A few points: 1. encapsulant over LED dice is soft, and it is fairly easy to damage the wirebonds on these multi-chip LEDs with any amount of pressure on the encap. The result is usually a short or intermittent short, which would explain the patterns you see. The lack of mechanically protective packaging is of more concern than ESD protective packaging, in my view. 2. Some people think that perfect matching is essential for proper current sharing, however, most high output LEDs are fairly resistive (see slope of VI curve) so 'self-ballasting'. This accounts for why strings of different effective lengths will still light up. 3. As others have suggested, these are probably reject units, but possibly just badly handled units. In my experience, many Chinese suppliers are capable of producing very high quality goods, but there are also plenty of vendors who will sell low quality stuff to anyone willing to pay. I am an LED illumination design engineer.
+Hendrik Hendrikson I just bought one on ebay (item 331620077265) and all 100 worked for $5. Not that they'll always sell perfect ones but it's a start.
I thank you sir ! After watching this I tested my just arrived led lights....... nine out of ten....... good ! I had checked them at full voltage and noticed that one did not seem quite right and it was the bad one. VERY helpful video as all of yours are, so again I thank you and I do appreciate your time !
+Jim Viau Really I don't think it makes much difference as to the individual seller from China. I got mine from Geek. Com and probably just got lucky! Wish I had a better answer for you!
The "real" issue is that the LEDs that light at very small current will be seriously overdriven when the voltage/amperage is set to achieve 100 watts. They will usually fail prematurely at open circuit and toss the whole string they are in. End result? A rooted light.
*****Clive, Good to see a fellow HVAC/R guy taking stuff apart. When you're showing this type of led array, please show it through a dark glass so even at higher power power we can see the individual LEDs. Thanks!
I can confirm your findings. Even with the more „expensive“ ones there seem to be 2 or 3 dim LEDs in the matrix of 100. But this refers to the 23..24Volt range in which they are quite faint. If I increase this to e.g. 26..27V my eye can’t assess the brightness any more as they are too bright. Using some sun glasses and welding equipment I managed to figure out that the brightness at that point is pretty even all across the matrix. Or in other words - the dim ones got up to brightness. So I would not worry so much about the voltage the individual LEDs need to “visually emit light” but rather focus on abnormalities in the range (28+Volts) they are typically operated. In case of mine there wasn’t a noticeable brightness difference. Additional remark: Just if one LED in the chain lights up - that doesn’t mean it is the only sink of energy. LEDs typically behave like 2V Zener Diodes (well 3V-ish in this case). But assuming that the reverse bias breakdown voltage is somewhat similar (say variance 1.9..2.2V) amongst LEDs the overall balance isn’t harmed significantly. To my experience with conventional LED’s - the breakdown voltage varies but even ones with the same current at the same voltage might differ in brightness. So unless you intend a very precise light distribution in very dim lights I wouldn’t worry too much about the inner balance of diodes.
they even out with voltage. his assessment is not correct, and it isn't the first time. i'm not even bothering to explain, they are all ok. try to "short"2 out of 10 and the others will instantly burn. or 9 out of 10. he's so wrong he's not even funny. he was refunded, because all sellers do that, afraid of bad rating.
@Matthayi Naalaaman Shutup dumbass. Or no 2 diodes are absolutely identical to the millivolt or microampere. Some will open before others, in the margin of error. I have a 12yo set of LED used for thousands of hours on my greenhouse that does exactly the same in low current... So yes, the fact that some series-connected diodes open before others is irrelevant. Now pay me for educating you.
Hi, nice test but try higher power, cooling and some light filter. I'm using one of those 100W in my underwater light and they stabilize themselves at about 27V. I'm running the LED 31V 3A and when I look on the LED using welding filter I see them all nicely glowing.
Thank you for the video. I myself wanted to experiment with some led lighting and had no idea this was an issue. Thanks for making me an informed consumer. Have a great day.
I learned most of my P4 (physics electronic circuits module) from this guy He's the reason I'm doing well in the electricity part of physics thank you very muchly
A better way to detect dead LEDs might be to turn it up to full voltage/current draw, then use a SLR camera or even your video camera if it has manual settings. If you set the aperture very high (High F number = Small Opening) and increase the shutter speed to very fast you'll be able to see each individual LED while they are running at full power. Then you could see if any are actually out and it's not just an imbalance of forward voltages of some of them where they're not quite on yet at the low voltages.
Welding glass is very helpful for this as already stated. He did however explain in the video, how he would not turn them up to full brightness without a heatsink attached. This is also a factor to consider for quick testing of LED chips.
from looking at this video.. it seems that there is both warm and cool LED's and the cool LED's light up first. and they are not placed evenly.. and also it seems when the voltage is low that it doesn;t spread the power across the LED's evenly.. but seems it might when they get the right voltage.
This may be a more recent thing. I puchased 2 100w warm white ones off a Chinese seller when they were around £4.50 each, and both were perfect. They have been installed in my car as auxiliary main beam headlights for 18 months at least & have not deteriorated at all.
I have had very good result with this type of LED. I tried what you did connecting to my bench supply and raising the voltage. I get the erratic lighting of some. This is at far below nominal current. I can say that when I properly heat sink them (I am using CPU heatsinks) and keep the die below 50-60C) and connect them to a current limited supply (about 3 amps) they are awesome. I measured 8-9000 lumens on mine. I bought a few cool white and warm white. I am going to use the for video lighting.
So I'm here, never thought Id be interested in anything of this, but this is the 10th video I see, of how people make cheap crappy electronics and god Im enjoying it a lot
Don't they even-out when you increase the voltage up to normal? Would putting a bunch of neutral filters over the camera lens and applying full power be a way to see this?
No. They dont even out. They will stay the same relative intensity. A properly matched LED produces a lot more light. I havent tried comparing light levels between a crappy LED and the same wattage through a good led. From what i have seen it seems a whole lot less.
@LazyH-Online This is quite an old post :p Since then I did do some testing. I had bought some 10W leds, some were very well matched and some were not. I could not get the poorly matched ones to take 10W, they max out at 7W and get very hot at 7W. There is quite a lot of difference in light output with a well matched LED. At much lower wattages it produced better light and MUCH less heat at 7W. The junk ones and good ones cost nearly the same, price doesnt matter. What was interesting is that the good ones came separately wrapped in plastic sleeves and had a serial number etched in the back... that was pretty cool since you know they passed quality control. The junk ones came in a large bag all loose. I was surprised at how much difference in light per watt. Given the lockdowns now is the perfect time to experiment :)
@LazyH-Online You're welcome! I find that the cheapest ad which is usually "sponsored" is junk. So anyone selling using the same picture is probably selling the same. I look for the next cheapest one with a different picture and roll the dice. China ebay sellers appear to all subscribe to drop ship items from several large sources. This is why they sell 60000 different items from hats and leds and cookware to vibrators. Another interesting source is aliexpress. You can look around at factories and you can often buy right from them. That way they arent hiding behind ebay anonymity. It makes me feel better knowing they make millions of something and have 5 star reviews. Also a good choice for lithium cells. I bought a bunch of lito... litoksomething something brand. I can look it up if you like. You could really use a bench powersupply to test and you can see how they work and learn at the same time. You can see the amperage increase as temperatures increase. You only need something that will do 30 or 40 volts and 2 or 3 amps to deal with most leds smaller than 100w. I had some luck with CREE leds, they are really nice but you need to spend for them. I am planning on converting some old crappy flashlights to use them. The very cheapest ones are fake, like everything from china. I am sort of impressed with large "straw hat" leds, 8mm i think. They put out about 0.5watt. I havent found a use yet. I got a whole bunch of small solar security lights from my friend that said they stopped working. They might be good for new leds. Lastly for fun and usefulness. My neighbour wanted leds for the trailer. I bought some of the cobs that are long and narrow. 150x10? They have the metal trim around the edges. I think bigclive looked at the recently. They are quite bright... a bit too much. Instead of running a resistor i picked up small motor controllers. They have a knob on them and pulse width modulate. These look neat. It gives full intensity control on the led. I will slide in a resistor to limit the maximum. The leds are about $1 or $2 usd and the motor controllers are about the same. If you have arduinos you could use a mosfet and control the light intensity. The current project... I want to use an arduino to control sequential LED taillights. An arduino nano is a tiny micro controller with several inputs and outputs. It can be programmed using a rather easy language. The chinese nano is about $3.
At first i thought some leds are dead too, but then i got a lense that focuses individual leds on a wall. i ran the matrix at full 100W power with heat sink on and all 100 worked fine. i concluded that each led had slightly different volt-ampere characteristics and in low power mode some turn on later then others. And led conducts some current, even if it is not emitting light so it explains why some in same series glow and others dont
The strings that light at a low voltage contain a blown led or two. Leds usually fail as a short circuit so you could wind up the volts and try and blow the dodgy strings into open circuit, the remaining strings might then behave sensibly.
A couple of things, after seeing this video. First question is, are these 100W LED's advertised as "dimmable?" Second question is about the power supply that you're running them with, is it a constant current source or a constant voltage source? This creates a huge difference in the final result. If the LED modules are not dimmable, then this is the behavior that one would expect out of a non-dimmable module. Non-dimmable arrays are not designed to run at current/voltages lower than their engineered thresholds and will generally operate quite erratically below those thresholds. Nowhere did I see this delineation called out and it's an important one. If these were dimmable LED's AND your input operating parameters (voltage and current) were consistent with the LED design parameters, then this behavior would be considered defective. Also, missing from your conversation, here, is the lumen output at the full operational voltage/current. That measurement is a critical one, as well, as it determines if the internal matrix is, indeed, operating appropriately, or accumulating failures.. Design centers require that each LED has a specific lumen output in order to achieve the final value (1000 lumens, 2000 lumens, etc). Should a single LED, in the matrix, fail, then the appropriate output value will not be achieved. As one, or more, LED's in the matrix fails the lumen output begins to fall, dramatically. LED diodes have a forward knee voltage/current parameter, as do standard diodes, beyond which they are gated on. Depending on manufacturing tolerances, there are a good number that may turn on early, and a good number that may turn on late and a middle range that all turn on about the same time. Basically one of those "bell curve" situations. Design centers can be as much as +/- 20% variation in less expensive products due to differences in substrate thickness, etc. That doesn't mean defect, it just means less accuracy. This is no different than the difference between 20% resistors and 1% resistors, where the 20% resistors are built to less rigorous standards of accuracy, not quality... but, even the 1% resistors have a temperature coefficient variations which can make the 1% tolerance meaningful only at room temperature, for instance, and outside of room temperature they can vary as much as 5-10% depending on whether the environment is hotter, or colder. A 1% resistor rated at 5PPM is MUCH more accurate than a 1% resistor rated at 50PPM, yet both are sold as 1% tolerance. And, in circuits where absolute accuracy isn't entirely critical it makes no sense to buy 1%, 5PPM resistors when a 20% resistor at 1/10 the cost will perform just as well. The same rules apply in non-dimmable LED modules where tolerances are not required to be tightly controlled and managed, as the only "correct" way to operate the LED module is at rated voltage/current. It is ONLY at that point that all LED's, in the matrix, are required to be both on, and emitting light at full rated power. Anything beyond that point (above, or below) doesn't not require that all LED's operate in unison, or equally. Finally, as individual LED's, in the matrix, operate at a sub-par level, the CRI and the chromaticity variations begin to degrade rapidly. Again, it's something that can be measured, that is both relevant, and scientific, rather than just looking at whether or not individual LED's light below specified operating thresholds. All of these factors, taken together, determine whether an LED is crap, or functional. All that I see, personally, in this video, is operation of an LED matrix below operating thresholds and calling it crap without actual measurements being made, and referenced. I'm not defending either you, or the eBay sellers in this comment... however, for me to make an accurate determination of product quality I would need to see a full spectrum of actual measurements made with the product operating within the correct thresholds, all presented together to make an accurate determination of product quality. Without that, it's all just conjecture and speculation...
It all depends, as others, and I, have responded to in your comments - whether or not minimal operating parameters (BOTH current AND voltage) were met on the LED. If this is, for instance, an LED that requires constant drive current and you're powering it with a constant voltage source, the current parameter, in many instances will be insufficiently met when the supply is dialed in to the correct drive voltage. The same applies if the LED requires a constant voltage source and you're powering the LED with a constant current supply, the current condition may be met but voltage variations, at the adjusted current level, may not be within the operating thresholds defined by the device. The methods you use, and your explanation in the video, fail to take those highly critical factors, which govern LED performance, into account. Also, the LED modules you're using - if they're dimmable, what is the dimming range? Many LED modules will only dim from 10-15% to 100% and, below the the lower threshold value operate inefficiently and ineffectively. In fact, I just bought a pricey LED troffer that supports dimming for my garage/workshop. It dims from 10% to 100% and, below 10% some of the LED's in the matrix light, others don't, as the manufacturer suggests they will. Manufacturers, due to design tolerances, expected light output, cost and other considerations will, effectively, limit the dimming range to match design criteria. Below the accepted dimming range the LED's operate in, often, erratic ways. My workshop lights have 10 year, unconditional, warranties. They're not cheap, they're not junk, but they don't dim below 10% and 20% of them don't dim, consistently, below 15%. The company says that's acceptable and "within tolerances." To determine if an LED module is "good" one must match the power supply, used to power the module, to the module itself. Operating a constant current LED with a constant voltage source will yield inconsistent results as will the reverse. So, first, the supply has to be matched to the LED drive requirements. Just because you're operating a constant voltage LED on a constant current source supply at the recommended voltage does not mean that it will function per specification, or even normally because the voltage is set right. Current can, still, under those conditions, be compromised. The next step, once the supply/driver type (constant current or constant voltage) is matched to the LED requirements, then you have can only work the LED dimming through the range of MANUFACTURER SPECIFIED values. Throughout that range the manufacturer specifies lumen, or LUX, output values (usually in the form of a graph), CRI index values, etc. A device that is good will meet manufacturer specifications in BOTH light output and color stability. But those items MUST be measured ALONG WITH the appropriate "drive" parameter (current or voltage) with the other "constant" element (current or voltage), always set correctly. If one doesn't make the full gamut of measurements, under the appropriate drive conditions there can be no determination of product quality. You cannot ascertain, for instance, CRI, using your eye. Even high quality LED's can, and many times do, fail a CRI measurement across the span of their dimmable range. LUX/Lumen output is, also, something that's hard to tell w/ the naked eye, unless you're the equivalent of Ansel Adams who could, within +/- 10%, consistently, determine how bright a light source is. For the other 99.9999999999% of us, we need a light meter to measure light intensity. And, light intensity has to be measured under controlled conditions. If you measure at 1" from the source one time, and 2" the next, the law about the light dropping as a function of the square of the distance comes into play in a hurry... as 100% change has a considerable impact on measurement. My job is to test/characterize LED's, specifically high-end streetlights... I spend a good amount of time doing this sort of thing... many "cheap" LED's, such as the ones you display, are actually very good WHEN you operate them WITHIN acceptable tolerances and boundaries AND drive them with the appropriate type of source. If you don't operate within acceptable boundaries and drive them with the appropriate source they will operate erratically, as they should. That doesn't mean they're bad, or cheap, or junk, it just means they weren't operated appropriately, much like a car in which, rather than gasoline you put diesel fuel in the tank... doesn't mean the car is junk, just means it's not being operated correctly... you can't judge the quality of either, if you're not maintaining them by their individual standards...
I picked up a few 100w ones from Sure Electronics, so far so good. Started turning on, all 100 of them at around 24-25 volts. So bright I needed a welding glass to look at them by 29v, all 100 still lighting up. The 10w ones I got from the same seller works a treat too. sure-display on US eBay
those leds are not as bad as they seem, i bought lots of different ones too and most are indeed the same as the $2 100w cheapos. but they do work and put out a lot of light with the correct power supply. the reason the chips come on at different voltages is because the chips were not binned and were cheap as such or outside spec. i have used the cheap white, blue and red (orangy red) and have not had one blow yet but i have blown several drivers. led is a current device but whack a thermistor on your heatsink wired to the voltage setting on your driver and a capacitor across it for soft start. the spectrograms are very useful for hydro, only missing 660nm which should be supplemented by some non-phosphor chips.
They were probably damaged by static while being shipped. In the royal mail sorting office they use automated sorting machines and the leds have probably travelled about 20 miles just inside of the machines so they will obviously be hit by static at some point since some of the belts are 2 miles long.
+TheScrappingJeahaha You're assuming they're from responsible sellers :D Did you notice the dodgy LED's came in a gift box? :D I seriously doubt they would use anti-static wrapping with that.
You can get the best-binned individually-packaged LEDs and put them on a LARGE BOARD, in the same configuration, and this will happen the same. This is what happens when your parallel/series arrays get too large and the lighting pattern your LEDs display show exactly this. You can't avoid this without a control resistor on each series string. Even the best-binned 100w COBs do this below their appropriate drive voltage, ALL OF THEM. I design/consult on LED units. This is just basic physics kicking you in the butt.
Could you recommend a reliable LED chip source? I have an outside light that had failed due to water ingress and need to replace the chip, its a 30w warm white.
I think I found a really good seller. I bought a few more just to make sure. The one I got has all but two of the LEDs glowing evenly and two glowing dimmer. None were out. That to me is a VERY good for a $2 led!
+Sharp Paul could be the "night" batches, the company my uncle works for makes quality control tests on chips and other electronics to make sure their clients order is to exact specs. Some place will run a good batch with example a part that cost 10cents(the one you need) but when no one is supervising they replace that 10cent one with a 5 cent equivalent(so less quality. or in some case they will just use the same machine set up that in the "day" make yours and then they make their cheaper version at "night"
Probably not stolen, but they are rejects. For many years the same thing was done with multicore CPUs. They'd run the tests, and for example, if all 4 cores were good - that shipped as a 4 core. Three pass, one fails, disable the bad one and ship it as a 3 core. Two good, two bad, same, ship it as 2.
Fascinating video! After watching, I checked my 100 watt LED modules. Turns out all 100 LEDs lit fine. I got my modules for about $10 each on Amazon. Each one came in a static protective film inside of bubble wrap and more Packaging. I didn't have power source to bring the voltage up like you did so I turn the LEDs on and looked at the LEDs with my welding hood.
have you notice about the minimal voltage of diode to work? LED is a diode. more over, the series of diodes voltage drop shall not be the same on every diode! more more over, the parallel series of such diodes must consume a current that you have LIMITED. so how you supposed every single diode to light up similarly and evenly... you trying to "fire" the diode by applying a minimum energy, so some diodes will fire up first, dropping the voltage on them, not leaving the chance of other diodes to "fire up". try doing the same experiment by applying the correct voltage to all "set of diodes" as specified by manufacturer, then block the light by some sort of filter, then you should be able to see single led diodes by your camera, then and only then you shall speculate about their light intensity))) have a fun, you will be surprised.
back when all we had were the single red leds (1970's), I tested the voltage/current of a few batches. Similar results, top quality leds were almost all identical while the cheap ones especially in bulk were all over the map.
That's very informative, Clive - wow, I would have expected better quality on these LEDs - looks like the 20 LED versions are fine - the rest are rejects perhaps.... perhaps now, thanks to this video, more people know how to test, more will get thrown back!!!! LED lighting can be SO good but generally is so badly let down by rubbish manufacturing putting cost before quality - and that applies to many LED lamps (no, I don't sell LEDs before anyone asks - I just use a lot of them at home).
But it shouldn't be. If what Clive is saying about the circuit design is correct, all LEDs should light at an even brightness at the same time. An inconsistent light at lower voltages - "patchwork" if you will - suggests that some chips have a decreased resistance and/or power consumption comparative to the LEDs surrounding it. This can result in increased degradation of the LEDs, at a faster rate, due to an overcurrent situation occurring within the LEDs with the declined resistance, meaning the whole arrangement has to be replaced sooner than its decent, consistently lighting counterparts.
Thanks for sharing this info with us. I really didn't think about testing the matching of the leds because as far as I knew in practice, at full rated current they are going to be fine enough especially at their price. But now, I'll be wary of them. Cheers :)
i bought 20 days ago a 100w led from aliexpress for 2$ and im really happy with it. Difference in brightness is ~10% and only 1 led from 100 are lighting at 1.5V more than others so i think it is good deal.
Are they really damaged? It could be an intended feature. What do they look like when they're fully powered to spec, and you lay a filter over it? do they still have that pattern? these individual leds could be sitting on zener diodes with randomized breakdown voltages so that you can sensibly control the brightness of the entire unit by simply dialing down the voltage, and so you can cycle it on without getting a massive sudden current spike. you're running the unit at 0.5% of its intended workload and complaining that it doesn't work like you expect it to? sorry for being skeptical but this wasn't very scientific.
+sentientagent Highly unlikely that there would be a randomized Zener breakdowns. The LED junctions themselves could have a variance in doping if this type of function was required. You are correct of course about running the unit at such low level and expecting it to function. The demonstration is a good indicator that the matching may be poor but extrapolating that the units are faulty is a bit dodgy. The design and batch sampling at the factory should have ensured that the unit adheres to its specification, over the permitted drive levels. The cynic in me (and experience) however, tells me that they are very possibly the failures. Even that though, presupposes that the components are unusable which is not necessarily the case as the the testing may separate into more results than just good and bad.
The issue with the LED die not being matched lies in their IV curve characteristics. The reason why you need matched LEDs (that should all come on around the same time), is to ensure consistent and even current throughout the entire array. If you have a few die that have a lower forward voltage, they'll conduct proportionally higher amount of current over the rest of the die, leading to thermal stress in that area and eventually failure. Even though it is performing at "0.5%" of its expected workload, the die should still have similar (if not near identical) forward voltages, and should all turn on at the same time. What you're seeing in this video is a very big indicator of a low quality LED array.
@@joshdau they even out with voltage. his assessment is not correct, and it isn't the first time. i'm not even bothering to explain, they are all ok. try to "short"2 out of 10 and the others will instantly burn. or 9 out of 10. he's so wrong he's not even funny. he was refunded, because all sellers do that, afraid of bad rating.
Enjoyed your very well made and informative video. Thanks for the instruction and information. I subscribed immediately after watching, looking forward to watching your old and future posts.
The chips are not dead. As You raise the current going to the chips more of them light until the point when all of them light when the level of current necessary to drive all of them are reached. The diodes that require the less current remain lit the longest as You lower the current. You must have a driver, or power supply with enough overhead , or extra power to run all diodes effectively. I like Your work, don't give up, thanks for the knowledge , and video, respect.......Seven Thunders.......
+Ivan Quin The LEDs are in series so its all the same current. if some don't light as bright as neighbour its because the silicon is not well matched which is symptom of poor manufacturing quality. they need to put LEDs from same wafer or something to keep good match.
That could be true in theory , but looking at the results in Your video, shows all of the diodes lighting as more power was applied, and gradually , more going out as you decreased the power going to them. It would be nice if They behaved like You wanted them to and all go out at the same time, but they don't.
+Ivan Quin another problem I found wit these is that for the parallel series arrays, current is not even. Diodes start to fail along a series as they take a larger current than other lines. this drastically affects longevity at 75%+ output even with proper heat management.
Yes you are absolutely right those 100W are damaged LED's probably did not passed quality control. Guessing some of them are shorted else there is no way they will produce any light at 9V.
Gordon Lawrence They do actually fail like that as any other diode the junction melts if at high current and they get shorted but you may be right that is some sort of manufacturing process defect. This where rejected for sure at quality control and then sold as scrap at low price and then found their way on eBay.
I have seen plenty of rectifier diodes go short but due to the way an LED works it's more common for the bonding wires to go open circuit. An LED is not just a PN junction like a rectifier diode - it does not have huge bonding surfaces to the leads they are relatively speaking tiny.
Gordon Lawrence The way the LED's in this tests worked suggest that the defective LED dies are shorted. This 100W LED's are made out of a 10x10 x 1W matrix where 10 LED's are in series and 10 of this strings in parallel. One of them started producing light at around 9 or 10V that suggests that one of this 10 LED series strings had around 6 or 7 of them shorted so all the voltage drop will be on the working ones.
The point still stands that it is more likely the bonding process has shorted out the LED as i said originally IE there is a solder bridge across the pads.
there is certainly a reason why they only cost 10 dollars. buy some from CREE they will probably cost over 50 dollars but hey, you get what you pay for
@@constantbuzz like many chinese products they write the biggeat number from the best performing components and sell them all with the info even if they dont meet the mark.
They're okay for the price. I mounted one with a voltage converter in my Honda's dome light. I'm running it at 22V with no noticeable issues in consistency, and it is a fairly clean 5000-6000k in temp. It is true that they do not light evenly, you can see that when near the forward voltage. For $15, it looks like I have a 100W halogen in my cab. I have no cooling issues at 22V and I'm not using a dedicated heatsink for it.
It would be interesting to try looking at the LED light distribution at full power, maybe by diminishing the light with a welding glass window or something. Maybe these discrepancies at low voltages don't matter much, as long as you have sufficient heat sinking? It would also be really awesome to have an IR image of the back of the metal pad after turning on the chip for 1s, to check the heat distribution and difference in heat load chip-to-chip.
i got several of these, im using it since 2 years for my flashlight's (see my channel) at 22V they look screwed up, but with more than 24V it balances. i never had heat or current problems, not a single one broke. im running them on about 34V 3.33A(CC CV) fs2.directupload.net/images/150619/tgklzr69.jpg fs2.directupload.net/images/150619/exj7uws5.jpg fs2.directupload.net/images/150619/42b224l6.jpg There are 100W shitbricks for 80$ out there, but whos going to buy this? These one cost 3-4$ per led. you cant expect much, but you get much for the price.
You can dim an LED by either reducing the current through it or by turning it on and off very quickly and varying the on to off time. (Called pulse width modulation.)
Regular dimmers used for incandescent bulbs work on AC power and LEDs have to be driven with very well-regulated DC. Incandescent bulbs are basically just heaters and it doesn't really matter what quality of power you feed them. Your typical LED retrofit bulb contains a circuit that converts the 120V AC signal from your wall into DC to drive the LEDs. It is almost impossible to design a driver circuit that works well when fed with AC power that has been run through a dimmer first, simply because the AC->DC driver circuit no longer has a clean power supply to work from (Wallplug-dimmable LED bulbs do not work very well, especially at low dimmer settings). It's actually easier to make an LED bulb that screws works only in a non-dimmable socket with clean AC input, but that has its own remote control for doing the dimming on the DC side. What needs to happen is someone needs to develop an actual standard for LED lighting instead of trying to shoehorn LEDs into sockets intended for incandescent which completely different (and contradictory) needs.
Jarrod P what LEDs aren't exactly complex ICs. Undervolting them is not damaging and should either cause the lights to not show or to be dim. They should never be patchy as this video shows.
Jarrod P No. He is surprised that the diodes weren't matched. Even though they will (probably) all turn on at 100W, they will be collectively less bright (the power rating is what it is for a reason -- the light output from the diodes which consume too much power will not compensate for those which consume too little) and more importantly will have a much shorter lifetime. The lifetime of each column is decided by its weakest link and temperature differences can have dramatic effects on longevity. If the one "overeager" diode in a column burns out 3x quicker, the entire column dies. This is why matching is important even if you can't "see" the patchwork at full power.
Blacktronics How do you know they are just in series? They might have some resistance in parallel with the individual LEDs. I'd like to see what they look like at, say, somewhere between 30W and 100W. You'd need dark glass or something to be able to see if all the LEDs were working more equally in that situation.
the majority of the factories in Guangdong (they just assemble) here source their components from the same suppliers, so the quality is quite uniform, they just had the led expo in Guangzhou recently.
I bought a 100 watt 10,000 lumen one on eBay but it was almost $30.00 just for one. The emitter chips are twice the size as the $5 competitor 100 watt LED. The seller was showing just what you experienced if you use the common cheaper units. So I bought their $30 good one. All 100 LED elements turn on at the same time as your 20 watt examples. You have to get the better quality 100 watt units. Full power (3 amps at 33 volts). Not a single problem. Takes a good amount of heat sinking. So there are good 100 watt LEDs on eBay. You just have to pay for them.
It was ebay user "wholesale-led" and it is there Epistar series and are not cheap by any means. The 100 watt LED with the oversize die is now $40.00 USD. But definitely worth it. When I got it, every LED was identical even at fully dim using less than 1mA. It is so much brighter than those $5 crappy ones which did exactly as these did in the video and I chucked them. There is more than twice the silicon emitter die area in the Epistar brand. Now I did retest my 100W Epistar at 1mA after I have been using it to verify how well balanced the LED elements are. I do have 99 elements that are perfect, one that is slightly dimmer at the 1mA. At 4mA, they are all identical brightness though. Bear in mind the LED operates at 3 amps so the mismatch even after I ESD zapped it, it is still ideal. I marginally damaged one element from improper handling as all 100 were perfectly matched when I received it. You have to be very careful handling these and protecting from ESD. I wish they would have integrated a 36 volt zener diode into these to protect them when out of circuit. Some of the CREE brand multi LED units have the integrated zener. The ebay listing is 361294347502
Puts me in mind of trying to find high output incandescent headlamps for my bike, which only came with a hopeless single 35w incandescent, and for which 35w halogen only gave a mild upgrade and tended to burn out far too fast because of the smaller capsule and higher temperature... unfortunately it uses a Ba20d connector rather than a typical H-type, and although the connector is good for higher power hardly anyone seems to make bulbs higher than 35w. Managed to find a 45w incandescent and a 40w halogen, both of which worked fine for a while, but apart from that one seller that charged a fortune and has since disappeared, all the other offerings were obvious fakes claiming 60 to 100w output or the like (which were unlikely to be true and would burn out the alternator anyway), and LED ones with uncertain beam patterns and pie in the sky lumen claims... Seemingly aimed at 16 year olds who wanted as bright a lamp as possible for their 50cc scooter, more for bragging rights than utility as there's not really any point to high output on city streets (I was often having to ride on unlit country roads at the time, thankfully not so any more so I'm no longer bothered about finding replacements)... if it sounds too good to be true, well...
I'm jet lagged and woke up early. Watching this is so interesting! I have had a string of cob side lights for my car from eBay. Failed after a few weeks, but eventually found one type with silicone gel lenses that seem to last so far. Let's face it, these cob LEDs are new technology. Maybe in a few years the Chinese will have got their manufacturing quality up to speed.
Keep in mind that some of the COB style sidelights use resistors to pass lots of extra current (and generate lots of heat) to make them compatible with lamp failure detection systems. The ones that get very hot can burn out quickly and also cause damage to some lamp holders.
I picked up a bunch of 10w 12v versions of these, much smaller, but the whole thing is filled with LED's, interesting to see that the larger 20w ones have only two strips inside them. That's also some pretty severe patchwork on that, these 10w ones don't do that sort of thing. They came in the same box too. Even the 10w ones make quite a lot of heat, I planned on making a stroboscope powered by a 9v with one, however I was having trouble driving it for some reason. I have a couple small heatsinks with fans now though so hopefully it will work a bit better once I have some better transistors.
I bought mine from this guy, it was a pack of 10 for $9.55cad. Maybe quality control with larger ones is an issue though. www.ebay.ca/usr/happydeer88?_trksid=p2057872.m2749.l2754
Jesse Levesque The easiest trick is a very fast settling switchable constant current circuit. depending on the capacitance of the LED's (not something you normally have to consider) you should get settling time down to maybe a microsecond or so if you can reduce the overshoot etc.
Ah, you know that makes perfect sense, I hadn't even considered the LED's capacitance but since there are so many in one LED package it would definitely explain the issues I was having, I figured it was maybe a transistor issue but the LED's probably had enough capacitance to stay on in the brief moments when they were supposed to be off. I had only done this with a few single LED's before.
An LED only has a capacitance of about 10pF, its what that capacitive load does to the PSU that makes it a pain in the neck. Especially if your trying to strobe into the MHz band
Ah so really not all that much, hmm. I can't recall what frequencies I was getting into but it was enough to "freeze" the blades on a pretty powerful fan from a power supply. I'm trying to recall if I tried driving it from an external power supply or if it was just the one from my computer that was powering the arduino. Hmm. Either way I still think part of the issue was only using a basic transistor.
you buy XHP-70's from Cree vendors on quality copper heatsink stars, then heatsink those stars as good as you would have sinked the 100 watt emitter. You can end up with roughly the same lumen output at less than half the wattage. or you could buy a bunch of XM-L2's but the XHP-70 is 4 of those on the same die. If you push them at around 40 watts the light output won't be that noticeably different from these 100 watt emitters pushed at 100 watts.
My experience with theese are really good! They are cheap and super bright. But its important to have a power supply capable of delivering enough current, and heatisk!
Thanks! I just bought a bunch of cheap chinese 100W chips and of the 8 I bought, 5 were near-perfect, 2 were OK (better than the ones in your video, but not great) and 1 has a dead strip until it's too bright to look at.
Not a bad statistic, based on the fact that some are $3 and other are $10 . I wouldn't really care if I got 8 out of 10 at $3. Still even at $10 you are not guaranty that one of will not be defective. Thanks for sharing.
An excellent video as always, as an electronics engineer myself I have no issues with your testing of these, unlike a few uninformed critics. Sadly I still buy cheap electronic components on Ebay and often suspect from the price or the country of origin that what I might be getting is a "b" grade component. Even with my low expectations they still manage to frequently disappoint me. Deliberately mis-labelled transistors, diodes and various other components, it gets boring and you can't build anything you can trust using their components. It is criminal deception but Ebay do too well out of selling this stuff to take action.
+g0fvt High quality LED light manufacturer in China, using the chip, from the source to control the quality of the light source. Can email detailed chat: wiseled@aliyun.com
I bought regular 5mm LEDs from a very large far eastern country, and some of them had the negative leg longer instead of the positive. How can a machine do that ? ? ? I suspect the aim is to sabotage western electronics that use their components.
Some time ago I tested some 20 watt LED s and noticed that the emitters didn't come to life evenly as the current was slowly increased. But at a certain point they evened out and worked reliably up to 20 watts. I used a filter/attenuator like welding visor glass to see the individual LEDs as they got brighter. It would be interesting to know how the emitter matching affects long term reliability. Maybe the ones that light first are more stressed, or maybe there is a balancing effect possibly due to thermal coupling the evens them out at higher power levels; like paralleling mosfets.
I think because they are wired in series is the main reason they light that way, I have a 50 watt LED light that does the same thing until voltage raises about 29.8 volts then all light evenly, also I'd like to point out that a driver and regulated power supply is also recommended for this application without driver attached the voltage does not always remain steady, current is regulated with the regulated power supply which is the mist damaging thing about LEDS. If you get a mismatch in both light output is affected, my 50 watt LED draws less then 2.0 amps but heat is def a factor and a good heat sink is required once you run close to the max amps and voltage.
I bought some 9W 120 VAC LED lamps from a Chinese company (Banggood) and they each lasted about 4 months or about 1000 hours. I took them apart and found several of the chips were bad. I replaced them with 100 mW 30 mA LEDs and got it working, but only for about 5 minutes (since the current in each string is about 180 mA). The LEDs seem to have been well matched but highly overrated. I plan to replace all 18 of them with good Osram LEDs which are only about $0.20 each. I can get Chinese chips for as little as $0.02 each but they are probably not 1/2 watt but maybe 200 mW. Thanks for your expose of the cheap modules.
I'm a bit confused by the voltages of these LED's. I bought a few 50 watt LED's a few years ago & the maximum voltage to run them must have been 24 volts as that's the supply I bought. Currently using one as a spotlight in my workshop. I did do this test with a bench supply & mine lit the same way these ones did, a few random chips light up first. I used a doubled up piece of welding curtain to look at the chips at full voltage & they do all light up in the end but obviously not matched as you said.
I have just bought 2 of the cheapest at the time (maybe $2.3) 100w Cool White COB LEDs from eBay (.com). I have ran them with 24v (from an ATX PSU) to test them. They light up _very_ dimly (total brightness like a classic GaAs indicator LED), and each of the LEDs inside the chip *light up extremely evenly* and consistently. It honestly surprised me. They came in a static dissipative thermosealed pouch, and the LED is physically the same as yours. I'll post the seller asap.
Hi Clive I think I might have an explanation for the seemingly random distribution of LED's lighting; your 20W are indeed wired in series hence there's a consistent current passing through both 'chains' the 100W one is wired as a MATRIX in effect all the LED's are wired in parallel, since in the event of a failure in a particular diode you don't lose the whole stripe, but when you've got a 10x10 square any manufacturing flaws are going to show up as an inconsistent pattern of lighting.
The main purpose for a Diode is so power only travels in one direction, like a check valve for electric, but in all reality, trace amounts of juice leaks backwards all diodes do, but the Zener is designed to do this for special reasons, not all LED's are created equal.
Maybe things have improved since this, but anyway I purchased online some 30 Watt LEDS similar to those shown, about 3 years ago. All of them are still working perfectly.
Ok, so the 100W model is badly balanced regarding the forward voltage. Did you power them up to specs and see if they all work at least? They may even out somewhat after a break in period. Another thing I noticed is that you look directly at the light source. I find it helpful to use a strong welding glass in order to be able to observe the led's behavior under working conditions.
I brought mine from eachbuyer. 100w ones ran fine, all shine at once. They come in plastic containers seperately.
9 років тому+1
There is a reasoning behind those trashy LEDs and it makes sense and it is pretty simple: 1: LEDs, _if_ produced and used correctly, can provide decades of zero maintenance, care-free operation. Imagine street lights and car headlights not needing repairs in one's lifetime. 2: Chinese LEDs device factories cannot survive decades of not selling their products. There is only so much market you could fill with one product and _it needs to break_ soon enough that the production line does not stop. That's that...
Hi, as a newbie to all this, you're videos are really good. The current thing that I'm really confused about is the difference between COB and other (discreet LED components such as SMDs etc) designs. i.e. I get your message about not wanting to overdrive the LEDs (say keeping them down to about 20mA) but I don't yet understand how you apply this to COB solutions. The reason is that I'm playing around with descreet component and COB solutions but am getting mixed messages (...or I've just missed the point) when considering which power supplies I need for each. For example, with discreet components (such as SMDs on LED strips) I'm hearing that the power supply should be rated higher than the (e.g.) LED string. i.e. With a 20% safety margin or more. ...but with COBs, you can use a PSU that's about the same (or less, so as not to over-drive?) as the the single or string of COBs your trying to drive. So, I currently have a 50W COB with a 50W PSU (DC: 20-40V and DC: 1.5A) and am scratching my head as to how the 1.5A current is shared between all the COB LEDs. The COB is spec'd as 10xCOB LEDs in series with 5 of these rows in parallel. Guess I'm trying to wrongly apply my understanding of discreet component LEDs circuits to a COB so, in my mis-guided approach, I seem to sharing 1.5A between the 5 parallel rows so I end up with 300mA going through each of the 10 series LEDs. Something not right here. ...so I guess that the 50 LEDs (somehow?) share the 1.5A resulting in 30mA per LED. ...but how? Have I got this all wrong? Guess I just don't understand the way COB LEDS work yet. Any guidance or references would really be appreciated. ...or can you do a figurative tear-down of a COB wafer itself? Thanks again, most enjoyable and informative.
I got one of those chinese 100W LED that light up in a spotty fashion, but when tested it indeed draws 100W of power and is as bright as I'd expect with no burnt out LEDs. Not saying there isn't a problem with yours, but some of these seems to work just fine even with a few mismatched LEDs.
I have a 500w one, when I got it all of the diodes worked, but now I have one or two strings that is not working, and two or three that choose by them selves if they will work, but if I put enough power in them almost all work
I've found that when projecting an LED's light onto a wall using a lens, even the chips that don't glow at all at low currents are glowing quite nicely. It seems to me like the chips tend to have a little parasitic resistance of a couple of kiloohms across them, rather than being shorted outright. I got the cheapest LEDs I have ever seen, like seriously bottom of the barrel factory reject items, twenty 10W ones for 6 euros, and at very low currents only three or four of the nine chips tend to emit any light. To my surprise though, their brightness is pretty consistent beyond 100mA, with only two of the modules producing significantly less light than the rest. Now they most definitely have a lower efficiency than the good ones, but let's be honest, who expects good quality from these things anyway. They definitely work for me!
+Dennis W The problem is, most of these LEDs came out of a LED flood light factory or something like this, they test them and the bad ones are returned, and however they'll come up on some ebay stores. You got it right, the problem is the resistance, some of the chips get mor, others less current, and when you power them nearly their specific values, some of the led chips get too much current and die sooner than the other ones on that LED, the other ones will even die faster when some of the chips are already dead. I've found out myself, that the good ones mostly start at 5-7 bucks
+Dennis W The problem is, most of these LEDs came out of a LED flood light factory or something like this, they test them and the bad ones are returned, and however they'll come up on some ebay stores. You got it right, the problem is the resistance, some of the chips get mor, others less current, and when you power them nearly their specific values, some of the led chips get too much current and die sooner than the other ones on that LED, the other ones will even die faster when some of the chips are already dead. I've found out myself, that the good ones mostly start at 5-7 bucks
TheScrappingJeahaha I think you mean the parasitic resistances, yes? I rest my case though, their effect diminishes when approaching the LED's rated current.
I know that this is an old video, but in my experience Sure electronics are the best supplier. They do sell a few different types of this style of LED, all packed in anti-static bags but I've found that if you want the good, matched ones you've got to get the ones sold as having 45mil chips. These should come with sure electronics branding on the bag, whereas the crap ones that I got from Sure are branded Wondom. Haven't tried the 30mil ones but they may be good ones too. The downside is that the 100W chips are about £12 each.
Try running them at full power and look at the results through a welding mask. Then you could tell if all the LED's are on or not and get a relative intensity of each. Maybe while they don't have matched minimum forward voltage, they perform better at higher voltage / current feeds.
I bought one for a flashlight and it came in a nice static bag with bubble wrap around it. All LEDs lit up almost perfectly. The flashlight gives of a ton of light. If wanted i can send an eBay link.
i would love to see your angle on some of these, new fangled, off the shelf, batteryless jump starters.. i bought some supposedly 500f caps on a board from ebay, i was rather underwhelmed. thanks
Alright! Thats good at least, too sad replacements will probably be of the same bad quality. The 20W seems to be very nice however =) Thanks for the reply!
Never thought I would be sitting on UA-cam watching videos about LED's but here I am XD
Anyways thank you for making this!
loll same
same :)
high five for not watching the bbc and the utter shite they put out!
Sergio Odintsov m
TrollFaceTheMan Well, it definitely beats frying one's brain with Happy Tree Friends for instance....
A few points: 1. encapsulant over LED dice is soft, and it is fairly easy to damage the wirebonds on these multi-chip LEDs with any amount of pressure on the encap. The result is usually a short or intermittent short, which would explain the patterns you see. The lack of mechanically protective packaging is of more concern than ESD protective packaging, in my view. 2. Some people think that perfect matching is essential for proper current sharing, however, most high output LEDs are fairly resistive (see slope of VI curve) so 'self-ballasting'. This accounts for why strings of different effective lengths will still light up. 3. As others have suggested, these are probably reject units, but possibly just badly handled units. In my experience, many Chinese suppliers are capable of producing very high quality goods, but there are also plenty of vendors who will sell low quality stuff to anyone willing to pay. I am an LED illumination design engineer.
Do you work on grow lights too?
I wish this guy was my Science and Technology teacher back in School days.. Well done on your great video and please make more :-)
Thank you for taking the time to post this video. I truly appreciate it Sincerity Tim D
Question remains is on where to buy the good ones on Ebay. I guess one can't.
+Hendrik Hendrikson It's hard finding good LEDs on ebay. Even sellers that provide better ones may provide crappy one the next time.
+Hendrik Hendrikson I just bought one on ebay (item 331620077265) and all 100 worked for $5. Not that they'll always sell perfect ones but it's a start.
Tommy Waldner Thanks a zillion!
+Hendrik Hendrikson good LEDS aren't on eBay try Amazon
SpeakerLover1000#EPIC Thanks mate. You're right of course :)
I thank you sir ! After watching this I tested my just arrived led lights....... nine out of ten....... good ! I had checked them at full voltage and noticed that one did not seem quite right and it was the bad one. VERY helpful video as all of yours are, so again I thank you and I do appreciate your time !
Ken can you tell us who that seller was. Id like to buy some myself but after viewing this I no longer know where to to buy?
+Jim Viau
Really I don't think it makes much difference as to the individual seller from China. I got mine from Geek. Com and probably just got lucky! Wish I had a better answer for you!
The "real" issue is that the LEDs that light at very small current will be seriously overdriven when the voltage/amperage is set to achieve 100 watts. They will usually fail prematurely at open circuit and toss the whole string they are in. End result? A rooted light.
*****Clive, Good to see a fellow HVAC/R guy taking stuff apart. When you're showing this type of led array, please show it through a dark glass so even at higher power power we can see the individual LEDs. Thanks!
I can confirm your findings. Even with the more „expensive“ ones there seem to be 2 or 3 dim LEDs in the matrix of 100. But this refers to the 23..24Volt range in which they are quite faint. If I increase this to e.g. 26..27V my eye can’t assess the brightness any more as they are too bright. Using some sun glasses and welding equipment I managed to figure out that the brightness at that point is pretty even all across the matrix. Or in other words - the dim ones got up to brightness.
So I would not worry so much about the voltage the individual LEDs need to “visually emit light” but rather focus on abnormalities in the range (28+Volts) they are typically operated. In case of mine there wasn’t a noticeable brightness difference.
Additional remark: Just if one LED in the chain lights up - that doesn’t mean it is the only sink of energy. LEDs typically behave like 2V Zener Diodes (well 3V-ish in this case). But assuming that the reverse bias breakdown voltage is somewhat similar (say variance 1.9..2.2V) amongst LEDs the overall balance isn’t harmed significantly. To my experience with conventional LED’s - the breakdown voltage varies but even ones with the same current at the same voltage might differ in brightness.
So unless you intend a very precise light distribution in very dim lights I wouldn’t worry too much about the inner balance of diodes.
they even out with voltage. his assessment is not correct, and it isn't the first time.
i'm not even bothering to explain, they are all ok. try to "short"2 out of 10 and the others will instantly burn. or 9 out of 10.
he's so wrong he's not even funny.
he was refunded, because all sellers do that, afraid of bad rating.
@Matthayi Naalaaman No, that's not how LEDs work. Or any diode.
@Matthayi Naalaaman Shutup dumbass.
Or no 2 diodes are absolutely identical to the millivolt or microampere. Some will open before others, in the margin of error. I have a 12yo set of LED used for thousands of hours on my greenhouse that does exactly the same in low current...
So yes, the fact that some series-connected diodes open before others is irrelevant.
Now pay me for educating you.
Hi, nice test but try higher power, cooling and some light filter. I'm using one of those 100W in my underwater light and they stabilize themselves at about 27V. I'm running the LED 31V 3A and when I look on the LED using welding filter I see them all nicely glowing.
6:35 The LED made a $, you got robbed and they are taunting you.
And now thousands of people know the product is shit. You think they're laughing still?
Doubt they found it real funny when they refunded BigC's cash.
Thank you for the video. I myself wanted to experiment with some led lighting and had no idea this was an issue. Thanks for making me an informed consumer. Have a great day.
I learned most of my P4 (physics electronic circuits module) from this guy
He's the reason I'm doing well in the electricity part of physics
thank you very muchly
If this guy was my teacher every one of my classes would actually be interesting.
A better way to detect dead LEDs might be to turn it up to full voltage/current draw, then use a SLR camera or even your video camera if it has manual settings. If you set the aperture very high (High F number = Small Opening) and increase the shutter speed to very fast you'll be able to see each individual LED while they are running at full power. Then you could see if any are actually out and it's not just an imbalance of forward voltages of some of them where they're not quite on yet at the low voltages.
***** or use a welding mask glass
Welding glass is very helpful for this as already stated. He did however explain in the video, how he would not turn them up to full brightness without a heatsink attached. This is also a factor to consider for quick testing of LED chips.
from looking at this video.. it seems that there is both warm and cool LED's and the cool LED's light up first. and they are not placed evenly.. and also it seems when the voltage is low that it doesn;t spread the power across the LED's evenly.. but seems it might when they get the right voltage.
This may be a more recent thing. I puchased 2 100w warm white ones off a Chinese seller when they were around £4.50 each, and both were perfect. They have been installed in my car as auxiliary main beam headlights for 18 months at least & have not deteriorated at all.
I have had very good result with this type of LED. I tried what you did connecting to my bench supply and raising the voltage. I get the erratic lighting of some. This is at far below nominal current. I can say that when I properly heat sink them (I am using CPU heatsinks) and keep the die below 50-60C) and connect them to a current limited supply (about 3 amps) they are awesome. I measured 8-9000 lumens on mine. I bought a few cool white and warm white. I am going to use the for video lighting.
I know nothing about electronics. Now I know about LEDs and I really like the guys accent
So I'm here, never thought Id be interested in anything of this, but this is the 10th video I see, of how people make cheap crappy electronics and god Im enjoying it a lot
Don't they even-out when you increase the voltage up to normal? Would putting a bunch of neutral filters over the camera lens and applying full power be a way to see this?
+metallitech YES
No. They dont even out. They will stay the same relative intensity. A properly matched LED produces a lot more light. I havent tried comparing light levels between a crappy LED and the same wattage through a good led. From what i have seen it seems a whole lot less.
@LazyH-Online This is quite an old post :p Since then I did do some testing. I had bought some 10W leds, some were very well matched and some were not. I could not get the poorly matched ones to take 10W, they max out at 7W and get very hot at 7W. There is quite a lot of difference in light output with a well matched LED. At much lower wattages it produced better light and MUCH less heat at 7W.
The junk ones and good ones cost nearly the same, price doesnt matter. What was interesting is that the good ones came separately wrapped in plastic sleeves and had a serial number etched in the back... that was pretty cool since you know they passed quality control. The junk ones came in a large bag all loose.
I was surprised at how much difference in light per watt.
Given the lockdowns now is the perfect time to experiment :)
@LazyH-Online You're welcome! I find that the cheapest ad which is usually "sponsored" is junk. So anyone selling using the same picture is probably selling the same. I look for the next cheapest one with a different picture and roll the dice. China ebay sellers appear to all subscribe to drop ship items from several large sources. This is why they sell 60000 different items from hats and leds and cookware to vibrators. Another interesting source is aliexpress. You can look around at factories and you can often buy right from them. That way they arent hiding behind ebay anonymity. It makes me feel better knowing they make millions of something and have 5 star reviews. Also a good choice for lithium cells. I bought a bunch of lito... litoksomething something brand. I can look it up if you like.
You could really use a bench powersupply to test and you can see how they work and learn at the same time. You can see the amperage increase as temperatures increase. You only need something that will do 30 or 40 volts and 2 or 3 amps to deal with most leds smaller than 100w.
I had some luck with CREE leds, they are really nice but you need to spend for them. I am planning on converting some old crappy flashlights to use them. The very cheapest ones are fake, like everything from china.
I am sort of impressed with large "straw hat" leds, 8mm i think. They put out about 0.5watt. I havent found a use yet. I got a whole bunch of small solar security lights from my friend that said they stopped working. They might be good for new leds.
Lastly for fun and usefulness. My neighbour wanted leds for the trailer. I bought some of the cobs that are long and narrow. 150x10? They have the metal trim around the edges. I think bigclive looked at the recently. They are quite bright... a bit too much. Instead of running a resistor i picked up small motor controllers. They have a knob on them and pulse width modulate. These look neat. It gives full intensity control on the led. I will slide in a resistor to limit the maximum. The leds are about $1 or $2 usd and the motor controllers are about the same.
If you have arduinos you could use a mosfet and control the light intensity. The current project... I want to use an arduino to control sequential LED taillights. An arduino nano is a tiny micro controller with several inputs and outputs. It can be programmed using a rather easy language. The chinese nano is about $3.
@LazyH-Online its always good to try out new things. Just dont burn the place down :p
At first i thought some leds are dead too, but then i got a lense that focuses individual leds on a wall. i ran the matrix at full 100W power with heat sink on and all 100 worked fine. i concluded that each led had slightly different volt-ampere characteristics and in low power mode some turn on later then others. And led conducts some current, even if it is not emitting light so it explains why some in same series glow and others dont
what do you recommend for 100 watt led? or just buy 10x10watt cree XM-l2?
Thanks for the heads-up, Clive. Keep 'em coming! :-)
Thank you I learn a lot from you
Can you provide an example link?
Hey
The strings that light at a low voltage contain a blown led or two. Leds usually fail as a short circuit so you could wind up the volts and try and blow the dodgy strings into open circuit, the remaining strings might then behave sensibly.
A couple of things, after seeing this video. First question is, are these 100W LED's advertised as "dimmable?" Second question is about the power supply that you're running them with, is it a constant current source or a constant voltage source? This creates a huge difference in the final result.
If the LED modules are not dimmable, then this is the behavior that one would expect out of a non-dimmable module. Non-dimmable arrays are not designed to run at current/voltages lower than their engineered thresholds and will generally operate quite erratically below those thresholds. Nowhere did I see this delineation called out and it's an important one. If these were dimmable LED's AND your input operating parameters (voltage and current) were consistent with the LED design parameters, then this behavior would be considered defective.
Also, missing from your conversation, here, is the lumen output at the full operational voltage/current. That measurement is a critical one, as well, as it determines if the internal matrix is, indeed, operating appropriately, or accumulating failures.. Design centers require that each LED has a specific lumen output in order to achieve the final value (1000 lumens, 2000 lumens, etc). Should a single LED, in the matrix, fail, then the appropriate output value will not be achieved. As one, or more, LED's in the matrix fails the lumen output begins to fall, dramatically.
LED diodes have a forward knee voltage/current parameter, as do standard diodes, beyond which they are gated on. Depending on manufacturing tolerances, there are a good number that may turn on early, and a good number that may turn on late and a middle range that all turn on about the same time. Basically one of those "bell curve" situations. Design centers can be as much as +/- 20% variation in less expensive products due to differences in substrate thickness, etc. That doesn't mean defect, it just means less accuracy. This is no different than the difference between 20% resistors and 1% resistors, where the 20% resistors are built to less rigorous standards of accuracy, not quality... but, even the 1% resistors have a temperature coefficient variations which can make the 1% tolerance meaningful only at room temperature, for instance, and outside of room temperature they can vary as much as 5-10% depending on whether the environment is hotter, or colder. A 1% resistor rated at 5PPM is MUCH more accurate than a 1% resistor rated at 50PPM, yet both are sold as 1% tolerance. And, in circuits where absolute accuracy isn't entirely critical it makes no sense to buy 1%, 5PPM resistors when a 20% resistor at 1/10 the cost will perform just as well. The same rules apply in non-dimmable LED modules where tolerances are not required to be tightly controlled and managed, as the only "correct" way to operate the LED module is at rated voltage/current. It is ONLY at that point that all LED's, in the matrix, are required to be both on, and emitting light at full rated power. Anything beyond that point (above, or below) doesn't not require that all LED's operate in unison, or equally.
Finally, as individual LED's, in the matrix, operate at a sub-par level, the CRI and the chromaticity variations begin to degrade rapidly. Again, it's something that can be measured, that is both relevant, and scientific, rather than just looking at whether or not individual LED's light below specified operating thresholds. All of these factors, taken together, determine whether an LED is crap, or functional. All that I see, personally, in this video, is operation of an LED matrix below operating thresholds and calling it crap without actual measurements being made, and referenced. I'm not defending either you, or the eBay sellers in this comment... however, for me to make an accurate determination of product quality I would need to see a full spectrum of actual measurements made with the product operating within the correct thresholds, all presented together to make an accurate determination of product quality. Without that, it's all just conjecture and speculation...
So Can you explain why the other chips were fine on the same power supply?
It all depends, as others, and I, have responded to in your comments - whether or not minimal operating parameters (BOTH current AND voltage) were met on the LED. If this is, for instance, an LED that requires constant drive current and you're powering it with a constant voltage source, the current parameter, in many instances will be insufficiently met when the supply is dialed in to the correct drive voltage. The same applies if the LED requires a constant voltage source and you're powering the LED with a constant current supply, the current condition may be met but voltage variations, at the adjusted current level, may not be within the operating thresholds defined by the device. The methods you use, and your explanation in the video, fail to take those highly critical factors, which govern LED performance, into account.
Also, the LED modules you're using - if they're dimmable, what is the dimming range? Many LED modules will only dim from 10-15% to 100% and, below the the lower threshold value operate inefficiently and ineffectively. In fact, I just bought a pricey LED troffer that supports dimming for my garage/workshop. It dims from 10% to 100% and, below 10% some of the LED's in the matrix light, others don't, as the manufacturer suggests they will. Manufacturers, due to design tolerances, expected light output, cost and other considerations will, effectively, limit the dimming range to match design criteria. Below the accepted dimming range the LED's operate in, often, erratic ways. My workshop lights have 10 year, unconditional, warranties. They're not cheap, they're not junk, but they don't dim below 10% and 20% of them don't dim, consistently, below 15%. The company says that's acceptable and "within tolerances."
To determine if an LED module is "good" one must match the power supply, used to power the module, to the module itself. Operating a constant current LED with a constant voltage source will yield inconsistent results as will the reverse. So, first, the supply has to be matched to the LED drive requirements. Just because you're operating a constant voltage LED on a constant current source supply at the recommended voltage does not mean that it will function per specification, or even normally because the voltage is set right. Current can, still, under those conditions, be compromised. The next step, once the supply/driver type (constant current or constant voltage) is matched to the LED requirements, then you have can only work the LED dimming through the range of MANUFACTURER SPECIFIED values. Throughout that range the manufacturer specifies lumen, or LUX, output values (usually in the form of a graph), CRI index values, etc. A device that is good will meet manufacturer specifications in BOTH light output and color stability. But those items MUST be measured ALONG WITH the appropriate "drive" parameter (current or voltage) with the other "constant" element (current or voltage), always set correctly. If one doesn't make the full gamut of measurements, under the appropriate drive conditions there can be no determination of product quality. You cannot ascertain, for instance, CRI, using your eye. Even high quality LED's can, and many times do, fail a CRI measurement across the span of their dimmable range. LUX/Lumen output is, also, something that's hard to tell w/ the naked eye, unless you're the equivalent of Ansel Adams who could, within +/- 10%, consistently, determine how bright a light source is. For the other 99.9999999999% of us, we need a light meter to measure light intensity. And, light intensity has to be measured under controlled conditions. If you measure at 1" from the source one time, and 2" the next, the law about the light dropping as a function of the square of the distance comes into play in a hurry... as 100% change has a considerable impact on measurement.
My job is to test/characterize LED's, specifically high-end streetlights... I spend a good amount of time doing this sort of thing... many "cheap" LED's, such as the ones you display, are actually very good WHEN you operate them WITHIN acceptable tolerances and boundaries AND drive them with the appropriate type of source. If you don't operate within acceptable boundaries and drive them with the appropriate source they will operate erratically, as they should. That doesn't mean they're bad, or cheap, or junk, it just means they weren't operated appropriately, much like a car in which, rather than gasoline you put diesel fuel in the tank... doesn't mean the car is junk, just means it's not being operated correctly... you can't judge the quality of either, if you're not maintaining them by their individual standards...
AKA TLD--Didn't answer. Mismatched LEDs cause that.
I picked up a few 100w ones from Sure Electronics, so far so good. Started turning on, all 100 of them at around 24-25 volts. So bright I needed a welding glass to look at them by 29v, all 100 still lighting up. The 10w ones I got from the same seller works a treat too. sure-display on US eBay
those leds are not as bad as they seem, i bought lots of different ones too and most are indeed the same as the $2 100w cheapos. but they do work and put out a lot of light with the correct power supply. the reason the chips come on at different voltages is because the chips were not binned and were cheap as such or outside spec. i have used the cheap white, blue and red (orangy red) and have not had one blow yet but i have blown several drivers. led is a current device but whack a thermistor on your heatsink wired to the voltage setting on your driver and a capacitor across it for soft start. the spectrograms are very useful for hydro, only missing 660nm which should be supplemented by some non-phosphor chips.
I agree, as I have several and at the proper working voltage they work fine and for the price it is a hell of a deal. Quit crying of 3 buck.
But they won't last. The brighter ones are pulling more current and will fail much sooner.
@@Steve_Just_Steve First step: what do you know about series connection? Can the current be different for serially connected LEDs?
They were probably damaged by static while being shipped. In the royal mail sorting office they use automated sorting machines and the leds have probably travelled about 20 miles just inside of the machines so they will obviously be hit by static at some point since some of the belts are 2 miles long.
+weirdstuffanimation7 the LEDs are shipped in propper antistatic bags ofcourse ;)
+TheScrappingJeahaha You're assuming they're from responsible sellers :D Did you notice the dodgy LED's came in a gift box? :D I seriously doubt they would use anti-static wrapping with that.
You can get the best-binned individually-packaged LEDs and put them on a LARGE BOARD, in the same configuration, and this will happen the same. This is what happens when your parallel/series arrays get too large and the lighting pattern your LEDs display show exactly this. You can't avoid this without a control resistor on each series string. Even the best-binned 100w COBs do this below their appropriate drive voltage, ALL OF THEM.
I design/consult on LED units. This is just basic physics kicking you in the butt.
Could you recommend a reliable LED chip source? I have an outside light that had failed due to water ingress and need to replace the chip, its a 30w warm white.
Nichia, Cree, Epistar, Edison, Osram.
....Phillips.....lumi-leds...
I can't trust Philips further than I can throw them, not after getting sold 25w grow lights which were advertised as 40w.
+techkitsune it's common practice to advertise the led light as the equivalent amount of light it will replace
I think I found a really good seller. I bought a few more just to make sure. The one I got has all but two of the LEDs glowing evenly and two glowing dimmer. None were out. That to me is a VERY good for a $2 led!
Seems maybe you got the factory rejects stolen by an employee
Pete Brown lol I really doubt they'd have to steal those :P
+sinephase then again, I really doubt they give them away for free in China.
+Sharp Paul could be the "night" batches, the company my uncle works for makes quality control tests on chips and other electronics to make sure their clients order is to exact specs.
Some place will run a good batch with example a part that cost 10cents(the one you need) but when no one is supervising they replace that 10cent one with a 5 cent equivalent(so less quality. or in some case they will just use the same machine set up that in the "day" make yours and then they make their cheaper version at "night"
Probably not stolen, but they are rejects. For many years the same thing was done with multicore CPUs. They'd run the tests, and for example, if all 4 cores were good - that shipped as a 4 core. Three pass, one fails, disable the bad one and ship it as a 3 core. Two good, two bad, same, ship it as 2.
@@dr666demento Goes back even further. The 486 SX was a regular 486, except that the FPU failed tests so they disabled it.
Fascinating video! After watching, I checked my 100 watt LED modules. Turns out all 100 LEDs lit fine. I got my modules for about $10 each on Amazon. Each one came in a static protective film inside of bubble wrap and more Packaging. I didn't have power source to bring the voltage up like you did so I turn the LEDs on and looked at the LEDs with my welding hood.
have you ever found a good source to fined the 100watt LED's
331620077265
Hmmm? Link to store?
Ebay, obvs
Pigeon
Not really obvious tbh, but thanks anyway man.
Mick C.
I've had some luck actually getting the dispute system on Aliexpress to work and getting free LEDs as a result. But I wouldn't count on it.
Channel idea that's probably already been done: test all no-name ebay products and give reviews.
have you notice about the minimal voltage of diode to work? LED is a diode.
more over, the series of diodes voltage drop shall not be the same on every diode!
more more over, the parallel series of such diodes must consume a current that you have LIMITED. so how you supposed every single diode to light up similarly and evenly... you trying to "fire" the diode by applying a minimum energy, so some diodes will fire up first, dropping the voltage on them, not leaving the chance of other diodes to "fire up".
try doing the same experiment by applying the correct voltage to all "set of diodes" as specified by manufacturer, then block the light by some sort of filter, then you should be able to see single led diodes by your camera, then and only then you shall speculate about their light intensity)))
have a fun, you will be surprised.
back when all we had were the single red leds (1970's), I tested the voltage/current of a few batches. Similar results, top quality leds were almost all identical while the cheap ones especially in bulk were all over the map.
Okay, their quality aren't very good but the best part is they're so cheap!
Plus he got a full refund, sounds like free leds to me
Oh hi
Mr.mosfet
That's very informative, Clive - wow, I would have expected better quality on these LEDs - looks like the 20 LED versions are fine - the rest are rejects perhaps.... perhaps now, thanks to this video, more people know how to test, more will get thrown back!!!! LED lighting can be SO good but generally is so badly let down by rubbish manufacturing putting cost before quality - and that applies to many LED lamps (no, I don't sell LEDs before anyone asks - I just use a lot of them at home).
I have one of those 100w LEDs, all the LEDs in it light up on full power, it's only patchwork at lower voltages.
But it shouldn't be. If what Clive is saying about the circuit design is correct, all LEDs should light at an even brightness at the same time. An inconsistent light at lower voltages - "patchwork" if you will - suggests that some chips have a decreased resistance and/or power consumption comparative to the LEDs surrounding it.
This can result in increased degradation of the LEDs, at a faster rate, due to an overcurrent situation occurring within the LEDs with the declined resistance, meaning the whole arrangement has to be replaced sooner than its decent, consistently lighting counterparts.
Then the one you have is JUNK
CoolDudeClem -- not known for having any clue what he's talking about.
Thanks for sharing this info with us. I really didn't think about testing the matching of the leds because as far as I knew in practice, at full rated current they are going to be fine enough especially at their price. But now, I'll be wary of them. Cheers :)
Hi Clive xD
bibib
i bought 20 days ago a 100w led from aliexpress for 2$ and im really happy with it. Difference in brightness is ~10% and only 1 led from 100 are lighting at 1.5V more than others so i think it is good deal.
LEDs are not sensitive to static electricity.
+Gianmario Scotti (Mario) The Gallium, Nitride ones are. Good suppliers ship them in metallised bags.
+bigclivedotcom OK.
Funny how this reply appears among my G+ notifications as if I had shared the video on G+.
+bigclivedotcom this seems to be the case for LEDs using sapphire as substrate, which now seems to be the majority of them. So, I stand corrected.
thanks, i have bought several identical wattage spot lights from ebay china, and they vary greatly. Now i understand why, cheers
Are they really damaged? It could be an intended feature. What do they look like when they're fully powered to spec, and you lay a filter over it? do they still have that pattern?
these individual leds could be sitting on zener diodes with randomized breakdown voltages so that you can sensibly control the brightness of the entire unit by simply dialing down the voltage, and so you can cycle it on without getting a massive sudden current spike.
you're running the unit at 0.5% of its intended workload and complaining that it doesn't work like you expect it to? sorry for being skeptical but this wasn't very scientific.
+sentientagent
Highly unlikely that there would be a randomized Zener breakdowns. The LED junctions themselves could have a variance in doping if this type of function was required.
You are correct of course about running the unit at such low level and expecting it to function. The demonstration is a good indicator that the matching may be poor but extrapolating that the units are faulty is a bit dodgy.
The design and batch sampling at the factory should have ensured that the unit adheres to its specification, over the permitted drive levels.
The cynic in me (and experience) however, tells me that they are very possibly the failures.
Even that though, presupposes that the components are unusable which is not necessarily the case as the the testing may separate into more results than just good and bad.
If you know about zener then a series run of LEDs shouldn't be so hard to comprehend resulting in your OP. ...ugh
The issue with the LED die not being matched lies in their IV curve characteristics. The reason why you need matched LEDs (that should all come on around the same time), is to ensure consistent and even current throughout the entire array. If you have a few die that have a lower forward voltage, they'll conduct proportionally higher amount of current over the rest of the die, leading to thermal stress in that area and eventually failure. Even though it is performing at "0.5%" of its expected workload, the die should still have similar (if not near identical) forward voltages, and should all turn on at the same time. What you're seeing in this video is a very big indicator of a low quality LED array.
@@joshdau
they even out with voltage. his assessment is not correct, and it isn't the first time.
i'm not even bothering to explain, they are all ok. try to "short"2 out of 10 and the others will instantly burn. or 9 out of 10.
he's so wrong he's not even funny.
he was refunded, because all sellers do that, afraid of bad rating.
@@atruebrit6452 "He's wrong and I'm right but I'm not going to bother explaining" Fuck off clown, might as well change your name to "A True Twit"
Enjoyed your very well made and informative video.
Thanks for the instruction and information. I subscribed immediately after watching, looking forward to watching your old and future posts.
The chips are not dead. As You raise the current going to the chips more of them light until the point when all of them light when the level of current necessary to drive all of them are reached. The diodes that require the less current remain lit the longest as You lower the current. You must have a driver, or power supply with enough overhead , or extra power to run all diodes effectively. I like Your work, don't give up, thanks for the knowledge , and video, respect.......Seven Thunders.......
+Ivan Quin The LEDs are in series so its all the same current. if some don't light as bright as neighbour its because the silicon is not well matched which is symptom of poor manufacturing quality. they need to put LEDs from same wafer or something to keep good match.
That could be true in theory , but looking at the results in Your video, shows all of the diodes lighting as more power was applied, and gradually , more going out as you decreased the power going to them. It would be nice if They behaved like You wanted them to and all go out at the same time, but they don't.
Never mind all that, You're doing a good experiment, and building good lights, keep up the good work, Thanks and respect.......Seven Thunders.......
+Ivan Quin another problem I found wit these is that for the parallel series arrays, current is not even. Diodes start to fail along a series as they take a larger current than other lines. this drastically affects longevity at 75%+ output even with proper heat management.
Thanks
Wow, I had no idea! No wonder the lights that come with theses big LEDs are so expensive, like the Zylight F8, which has a 200w version now!
I would bet that an employee is pocketing the bad chips and selling them.
Yes you are absolutely right those 100W are damaged LED's probably did not passed quality control.
Guessing some of them are shorted else there is no way they will produce any light at 9V.
***** LED's usually dont fail like that so it's more likely a screw up with the bonding process.
Gordon Lawrence They do actually fail like that as any other diode the junction melts if at high current and they get shorted but you may be right that is some sort of manufacturing process defect. This where rejected for sure at quality control and then sold as scrap at low price and then found their way on eBay.
I have seen plenty of rectifier diodes go short but due to the way an LED works it's more common for the bonding wires to go open circuit. An LED is not just a PN junction like a rectifier diode - it does not have huge bonding surfaces to the leads they are relatively speaking tiny.
Gordon Lawrence The way the LED's in this tests worked suggest that the defective LED dies are shorted.
This 100W LED's are made out of a 10x10 x 1W matrix where 10 LED's are in series and 10 of this strings in parallel. One of them started producing light at around 9 or 10V that suggests that one of this 10 LED series strings had around 6 or 7 of them shorted so all the voltage drop will be on the working ones.
The point still stands that it is more likely the bonding process has shorted out the LED as i said originally IE there is a solder bridge across the pads.
there is certainly a reason why they only cost 10 dollars. buy some from CREE they will probably cost over 50 dollars but hey, you get what you pay for
Interesting video.
Having said that, I am surprised how people get rude at eachother over a video showing dodgy 100W Ledś
+Ed19601 because they bought some and now realize that's why they don't match :P so instead of using the if its cheap its cheap rule they get rude.
Many Chinese on ebay sell used or rejected stuff so looks like you got some junk.
they may be factory seconds- instead of recycling them, sold with misinformation for cheap.
@@constantbuzz like many chinese products they write the biggeat number from the best performing components and sell them all with the info even if they dont meet the mark.
They're okay for the price. I mounted one with a voltage converter in my Honda's dome light. I'm running it at 22V with no noticeable issues in consistency, and it is a fairly clean 5000-6000k in temp. It is true that they do not light evenly, you can see that when near the forward voltage. For $15, it looks like I have a 100W halogen in my cab. I have no cooling issues at 22V and I'm not using a dedicated heatsink for it.
I would not be surprised if whats being sold are QC rejects.
+VIDEOSUPERHIGHWAY Quality Control? HA!!!
Like selling sausages made from "selected meat", they select all the grisly meat and put it in sausages.
It would be interesting to try looking at the LED light distribution at full power, maybe by diminishing the light with a welding glass window or something. Maybe these discrepancies at low voltages don't matter much, as long as you have sufficient heat sinking? It would also be really awesome to have an IR image of the back of the metal pad after turning on the chip for 1s, to check the heat distribution and difference in heat load chip-to-chip.
i got several of these, im using it since 2 years for my flashlight's (see my channel)
at 22V they look screwed up, but with more than 24V it balances. i never had heat or current problems, not a single one broke. im running them on about 34V 3.33A(CC CV)
fs2.directupload.net/images/150619/tgklzr69.jpg
fs2.directupload.net/images/150619/exj7uws5.jpg
fs2.directupload.net/images/150619/42b224l6.jpg
There are 100W shitbricks for 80$ out there, but whos going to buy this? These one cost 3-4$ per led. you cant expect much, but you get much for the price.
Didn’t expect that. Well briefed. Cheers.
i thought LEDs aren't dimmable? i know nothing...
You can dim an LED by either reducing the current through it or by turning it on and off very quickly and varying the on to off time. (Called pulse width modulation.)
bigclivedotcom how come most LED bulbs say 'not dimmable'?
It's their power supply circuitry that is not suitable for dimming on a standard wall-plate dimmer.
i see. tnx
Regular dimmers used for incandescent bulbs work on AC power and LEDs have to be driven with very well-regulated DC. Incandescent bulbs are basically just heaters and it doesn't really matter what quality of power you feed them.
Your typical LED retrofit bulb contains a circuit that converts the 120V AC signal from your wall into DC to drive the LEDs. It is almost impossible to design a driver circuit that works well when fed with AC power that has been run through a dimmer first, simply because the AC->DC driver circuit no longer has a clean power supply to work from (Wallplug-dimmable LED bulbs do not work very well, especially at low dimmer settings). It's actually easier to make an LED bulb that screws works only in a non-dimmable socket with clean AC input, but that has its own remote control for doing the dimming on the DC side.
What needs to happen is someone needs to develop an actual standard for LED lighting instead of trying to shoehorn LEDs into sockets intended for incandescent which completely different (and contradictory) needs.
Yet another very comprehensive review.
Thank you sir. Keep up the good work!
Your running 100 watt leds at a few watts and your surprised it doesnt work?
Jarrod P what
LEDs aren't exactly complex ICs. Undervolting them is not damaging and should either cause the lights to not show or to be dim. They should never be patchy as this video shows.
Jarrod P No. He is surprised that the diodes weren't matched. Even though they will (probably) all turn on at 100W, they will be collectively less bright (the power rating is what it is for a reason -- the light output from the diodes which consume too much power will not compensate for those which consume too little) and more importantly will have a much shorter lifetime. The lifetime of each column is decided by its weakest link and temperature differences can have dramatic effects on longevity. If the one "overeager" diode in a column burns out 3x quicker, the entire column dies. This is why matching is important even if you can't "see" the patchwork at full power.
Jarrod P get out. grow a functioning brain.
Jarrod P ever thought about learning how LED's work in series?
Blacktronics How do you know they are just in series? They might have some resistance in parallel with the individual LEDs.
I'd like to see what they look like at, say, somewhere between 30W and 100W. You'd need dark glass or something to be able to see if all the LEDs were working more equally in that situation.
the majority of the factories in Guangdong (they just assemble) here source their components from the same suppliers, so the quality is quite uniform, they just had the led expo in Guangzhou recently.
I bought a 100 watt 10,000 lumen one on eBay but it was almost $30.00 just for one. The emitter chips are twice the size as the $5 competitor 100 watt LED. The seller was showing just what you experienced if you use the common cheaper units. So I bought their $30 good one. All 100 LED elements turn on at the same time as your 20 watt examples. You have to get the better quality 100 watt units. Full power (3 amps at 33 volts). Not a single problem. Takes a good amount of heat sinking. So there are good 100 watt LEDs on eBay. You just have to pay for them.
Gavin Curtis what from what supplier? i am thinking of buying some
It was ebay user "wholesale-led" and it is there Epistar series and are not cheap by any means. The 100 watt LED with the oversize die is now $40.00 USD.
But definitely worth it. When I got it, every LED was identical even at fully dim using less than 1mA. It is so much brighter than those $5 crappy ones which did exactly as these did in the video and I chucked them.
There is more than twice the silicon emitter die area in the Epistar brand. Now I did retest my 100W Epistar at 1mA after I have been using it to verify how well balanced the LED elements are. I do have 99 elements that are perfect, one that is slightly dimmer at the 1mA. At 4mA, they are all identical brightness though. Bear in mind the LED operates at 3 amps so the mismatch even after I ESD zapped it, it is still ideal. I marginally damaged one element from improper handling as all 100 were perfectly matched when I received it. You have to be very careful handling these and protecting from ESD. I wish they would have integrated a 36 volt zener diode into these to protect them when out of circuit. Some of the CREE brand multi LED units have the integrated zener.
The ebay listing is 361294347502
Puts me in mind of trying to find high output incandescent headlamps for my bike, which only came with a hopeless single 35w incandescent, and for which 35w halogen only gave a mild upgrade and tended to burn out far too fast because of the smaller capsule and higher temperature... unfortunately it uses a Ba20d connector rather than a typical H-type, and although the connector is good for higher power hardly anyone seems to make bulbs higher than 35w. Managed to find a 45w incandescent and a 40w halogen, both of which worked fine for a while, but apart from that one seller that charged a fortune and has since disappeared, all the other offerings were obvious fakes claiming 60 to 100w output or the like (which were unlikely to be true and would burn out the alternator anyway), and LED ones with uncertain beam patterns and pie in the sky lumen claims... Seemingly aimed at 16 year olds who wanted as bright a lamp as possible for their 50cc scooter, more for bragging rights than utility as there's not really any point to high output on city streets (I was often having to ride on unlit country roads at the time, thankfully not so any more so I'm no longer bothered about finding replacements)... if it sounds too good to be true, well...
I'm jet lagged and woke up early. Watching this is so interesting! I have had a string of cob side lights for my car from eBay. Failed after a few weeks, but eventually found one type with silicone gel lenses that seem to last so far. Let's face it, these cob LEDs are new technology. Maybe in a few years the Chinese will have got their manufacturing quality up to speed.
Keep in mind that some of the COB style sidelights use resistors to pass lots of extra current (and generate lots of heat) to make them compatible with lamp failure detection systems. The ones that get very hot can burn out quickly and also cause damage to some lamp holders.
I make lots of lights from 3w chips they are a lot more reliable and will last thousands and thousands of hours, keep up the good work
I picked up a bunch of 10w 12v versions of these, much smaller, but the whole thing is filled with LED's, interesting to see that the larger 20w ones have only two strips inside them. That's also some pretty severe patchwork on that, these 10w ones don't do that sort of thing. They came in the same box too.
Even the 10w ones make quite a lot of heat, I planned on making a stroboscope powered by a 9v with one, however I was having trouble driving it for some reason. I have a couple small heatsinks with fans now though so hopefully it will work a bit better once I have some better transistors.
I bought mine from this guy, it was a pack of 10 for $9.55cad. Maybe quality control with larger ones is an issue though. www.ebay.ca/usr/happydeer88?_trksid=p2057872.m2749.l2754
Jesse Levesque The easiest trick is a very fast settling switchable constant current circuit. depending on the capacitance of the LED's (not something you normally have to consider) you should get settling time down to maybe a microsecond or so if you can reduce the overshoot etc.
Ah, you know that makes perfect sense, I hadn't even considered the LED's capacitance but since there are so many in one LED package it would definitely explain the issues I was having, I figured it was maybe a transistor issue but the LED's probably had enough capacitance to stay on in the brief moments when they were supposed to be off. I had only done this with a few single LED's before.
An LED only has a capacitance of about 10pF, its what that capacitive load does to the PSU that makes it a pain in the neck. Especially if your trying to strobe into the MHz band
Ah so really not all that much, hmm. I can't recall what frequencies I was getting into but it was enough to "freeze" the blades on a pretty powerful fan from a power supply.
I'm trying to recall if I tried driving it from an external power supply or if it was just the one from my computer that was powering the arduino. Hmm.
Either way I still think part of the issue was only using a basic transistor.
And the obvious question - Where does one find quality LED's?
It's very hard to find them. The best are reserved for the bigger equipment manufacturers.
Those ones were quite alright, and this is not how you test it
you buy XHP-70's from Cree vendors on quality copper heatsink stars, then heatsink those stars as good as you would have sinked the 100 watt emitter. You can end up with roughly the same lumen output at less than half the wattage. or you could buy a bunch of XM-L2's but the XHP-70 is 4 of those on the same die. If you push them at around 40 watts the light output won't be that noticeably different from these 100 watt emitters pushed at 100 watts.
Jack Sprat I learned my lesson,when buying on eBay,I stay away from items shipped from China
stars? sinked? same die? emitters pushed? Are you trying to tell us something?
My experience with theese are really good! They are cheap and super bright. But its important to have a power supply capable of delivering enough current, and heatisk!
Thanks!
I just bought a bunch of cheap chinese 100W chips and of the 8 I bought, 5 were near-perfect, 2 were OK (better than the ones in your video, but not great) and 1 has a dead strip until it's too bright to look at.
Not a bad statistic, based on the fact that some are $3 and other are $10 . I wouldn't really care if I got 8 out of 10 at $3. Still even at $10 you are not guaranty that one of will not be defective. Thanks for sharing.
...Right.....Good ones are expensive....100 Watt CREE units start around $60-70+....
Nice job man! thank you for taking the effort to make a video.
An excellent video as always, as an electronics engineer myself I have no issues with your testing of these, unlike a few uninformed critics.
Sadly I still buy cheap electronic components on Ebay and often suspect from the price or the country of origin that what I might be getting is a "b" grade component. Even with my low expectations they still manage to frequently disappoint me. Deliberately mis-labelled transistors, diodes and various other components, it gets boring and you can't build anything you can trust using their components. It is criminal deception but Ebay do too well out of selling this stuff to take action.
+g0fvt High quality LED light manufacturer in China, using the chip, from the source to control the quality of the light source. Can email detailed chat: wiseled@aliyun.com
I bought regular 5mm LEDs from a very large far eastern country, and some of them had the negative leg longer instead of the positive. How can a machine do that ? ? ? I suspect the aim is to sabotage western electronics that use their components.
Good one mate - had similar result with crappy LED's from Ebay - Chinese manufacture of course.
Some time ago I tested some 20 watt LED s and noticed that the emitters didn't come to life evenly as the current was slowly increased. But at a certain point they evened out and worked reliably up to 20 watts. I used a filter/attenuator like welding visor glass to see the individual LEDs as they got brighter.
It would be interesting to know how the emitter matching affects long term reliability. Maybe the ones that light first are more stressed, or maybe there is a balancing effect possibly due to thermal coupling the evens them out at higher power levels; like paralleling mosfets.
I think because they are wired in series is the main reason they light that way, I have a 50 watt LED light that does the same thing until voltage raises about 29.8 volts then all light evenly, also I'd like to point out that a driver and regulated power supply is also recommended for this application without driver attached the voltage does not always remain steady, current is regulated with the regulated power supply which is the mist damaging thing about LEDS. If you get a mismatch in both light output is affected, my 50 watt LED draws less then 2.0 amps but heat is def a factor and a good heat sink is required once you run close to the max amps and voltage.
I bought some 9W 120 VAC LED lamps from a Chinese company (Banggood) and they each lasted about 4 months or about 1000 hours. I took them apart and found several of the chips were bad. I replaced them with 100 mW 30 mA LEDs and got it working, but only for about 5 minutes (since the current in each string is about 180 mA). The LEDs seem to have been well matched but highly overrated. I plan to replace all 18 of them with good Osram LEDs which are only about $0.20 each. I can get Chinese chips for as little as $0.02 each but they are probably not 1/2 watt but maybe 200 mW. Thanks for your expose of the cheap modules.
It's a very professional testing.
I'm a bit confused by the voltages of these LED's. I bought a few 50 watt LED's a few years ago & the maximum voltage to run them must have been 24 volts as that's the supply I bought. Currently using one as a spotlight in my workshop. I did do this test with a bench supply & mine lit the same way these ones did, a few random chips light up first. I used a doubled up piece of welding curtain to look at the chips at full voltage & they do all light up in the end but obviously not matched as you said.
I have just bought 2 of the cheapest at the time (maybe $2.3) 100w Cool White COB LEDs from eBay (.com). I have ran them with 24v (from an ATX PSU) to test them. They light up _very_ dimly (total brightness like a classic GaAs indicator LED), and each of the LEDs inside the chip *light up extremely evenly* and consistently. It honestly surprised me. They came in a static dissipative thermosealed pouch, and the LED is physically the same as yours. I'll post the seller asap.
enlightening video, I always know I am taking chances when buying product from China on EBay.
Hi Clive I think I might have an explanation for the seemingly random distribution of LED's lighting;
your 20W are indeed wired in series hence there's a consistent current passing through both 'chains'
the 100W one is wired as a MATRIX in effect all the LED's are wired in parallel, since in the event of a failure in a particular diode you don't lose the whole stripe, but when you've got a 10x10 square any manufacturing flaws are going to show up as an inconsistent pattern of lighting.
The main purpose for a Diode is so power only travels in one direction, like a check valve for electric, but in all reality, trace amounts of juice leaks backwards all diodes do, but the Zener is designed to do this for special reasons, not all LED's are created equal.
Maybe things have improved since this,
but anyway I purchased online some 30 Watt LEDS similar to those shown, about 3 years ago.
All of them are still working perfectly.
The Cree CX-1507 LEDs are decent for light fixtures. A good 550 lumen, 93 CRI chip can be had for about $4, which in small quantities is decent.
I ordered the cheapest 100W LEDs I could find on AliExpress and they have all PERFECT matched chips - jackpot! ;-)
Ok, so the 100W model is badly balanced regarding the forward voltage. Did you power them up to specs and see if they all work at least? They may even out somewhat after a break in period. Another thing I noticed is that you look directly at the light source. I find it helpful to use a strong welding glass in order to be able to observe the led's behavior under working conditions.
I brought mine from eachbuyer. 100w ones ran fine, all shine at once. They come in plastic containers seperately.
There is a reasoning behind those trashy LEDs and it makes sense and it is pretty simple:
1: LEDs, _if_ produced and used correctly, can provide decades of zero maintenance, care-free operation. Imagine street lights and car headlights not needing repairs in one's lifetime.
2: Chinese LEDs device factories cannot survive decades of not selling their products. There is only so much market you could fill with one product and _it needs to break_ soon enough that the production line does not stop.
That's that...
Hi, as a newbie to all this, you're videos are really good.
The current thing that I'm really confused about is the difference between COB and other (discreet LED components such as SMDs etc) designs. i.e. I get your message about not wanting to overdrive the LEDs (say keeping them down to about 20mA) but I don't yet understand how you apply this to COB solutions. The reason is that I'm playing around with descreet component and COB solutions but am getting mixed messages (...or I've just missed the point) when considering which power supplies I need for each. For example, with discreet components (such as SMDs on LED strips) I'm hearing that the power supply should be rated higher than the (e.g.) LED string. i.e. With a 20% safety margin or more. ...but with COBs, you can use a PSU that's about the same (or less, so as not to over-drive?) as the the single or string of COBs your trying to drive. So, I currently have a 50W COB with a 50W PSU (DC: 20-40V and DC: 1.5A) and am scratching my head as to how the 1.5A current is shared between all the COB LEDs. The COB is spec'd as 10xCOB LEDs in series with 5 of these rows in parallel. Guess I'm trying to wrongly apply my understanding of discreet component LEDs circuits to a COB so, in my mis-guided approach, I seem to sharing 1.5A between the 5 parallel rows so I end up with 300mA going through each of the 10 series LEDs. Something not right here. ...so I guess that the 50 LEDs (somehow?) share the 1.5A resulting in 30mA per LED. ...but how?
Have I got this all wrong? Guess I just don't understand the way COB LEDS work yet. Any guidance or references would really be appreciated. ...or can you do a figurative tear-down of a COB wafer itself?
Thanks again, most enjoyable and informative.
watching this at 4am.. time to re-evaluate my life.
I got one of those chinese 100W LED that light up in a spotty fashion, but when tested it indeed draws 100W of power and is as bright as I'd expect with no burnt out LEDs. Not saying there isn't a problem with yours, but some of these seems to work just fine even with a few mismatched LEDs.
I have a 500w one, when I got it all of the diodes worked, but now I have one or two strings that is not working, and two or three that choose by them selves if they will work, but if I put enough power in them almost all work
I've found that when projecting an LED's light onto a wall using a lens, even the chips that don't glow at all at low currents are glowing quite nicely. It seems to me like the chips tend to have a little parasitic resistance of a couple of kiloohms across them, rather than being shorted outright.
I got the cheapest LEDs I have ever seen, like seriously bottom of the barrel factory reject items, twenty 10W ones for 6 euros, and at very low currents only three or four of the nine chips tend to emit any light. To my surprise though, their brightness is pretty consistent beyond 100mA, with only two of the modules producing significantly less light than the rest. Now they most definitely have a lower efficiency than the good ones, but let's be honest, who expects good quality from these things anyway. They definitely work for me!
+Dennis W The problem is, most of these LEDs came out of a LED flood light factory or something like this, they test them and the bad ones are returned, and however they'll come up on some ebay stores. You got it right, the problem is the resistance, some of the chips get mor, others less current, and when you power them nearly their specific values, some of the led chips get too much current and die sooner than the other ones on that LED, the other ones will even die faster when some of the chips are already dead. I've found out myself, that the good ones mostly start at 5-7 bucks
+Dennis W The problem is, most of these LEDs came out of a LED flood light factory or something like this, they test them and the bad ones are returned, and however they'll come up on some ebay stores. You got it right, the problem is the resistance, some of the chips get mor, others less current, and when you power them nearly their specific values, some of the led chips get too much current and die sooner than the other ones on that LED, the other ones will even die faster when some of the chips are already dead. I've found out myself, that the good ones mostly start at 5-7 bucks
+TheScrappingJeahaha The problem is the resistance? That's such a stupid statement, I don't even know where to begin trolling you.
Jimmy Gregorio
then maybe you just fuck off?
TheScrappingJeahaha I think you mean the parasitic resistances, yes?
I rest my case though, their effect diminishes when approaching the LED's rated current.
I know that this is an old video, but in my experience Sure electronics are the best supplier. They do sell a few different types of this style of LED, all packed in anti-static bags but I've found that if you want the good, matched ones you've got to get the ones sold as having 45mil chips. These should come with sure electronics branding on the bag, whereas the crap ones that I got from Sure are branded Wondom. Haven't tried the 30mil ones but they may be good ones too. The downside is that the 100W chips are about £12 each.
Try running them at full power and look at the results through a welding mask. Then you could tell if all the LED's are on or not and get a relative intensity of each.
Maybe while they don't have matched minimum forward voltage, they perform better at higher voltage / current feeds.
I bought one for a flashlight and it came in a nice static bag with bubble wrap around it. All LEDs lit up almost perfectly. The flashlight gives of a ton of light. If wanted i can send an eBay link.
i would love to see your angle on some of these, new fangled, off the shelf, batteryless jump starters.. i bought some supposedly 500f caps on a board from ebay, i was rather underwhelmed. thanks
Do you get any replacements for these? Great video and thanks for sharing! =)
Chris Fredriksson I got a refund from both suppliers. I've not even tried ordering replacements yet.
Alright! Thats good at least, too sad replacements will probably be of the same bad quality. The 20W seems to be very nice however =) Thanks for the reply!