Near unity power factor LED driver - with odd schematic
Вставка
- Опубліковано 8 вер 2024
- A fairly deep dive into the odd but very interesting circuitry of a 0.99 power factor LED driver, that uses clever minimalist circuitry that is not documented on the chip manufacturer's datasheet.
I'm wondering if this is a pin compatible drop-in replacement for another brand that offers other schematic options.
The most obvious electrical weakness of the design is the unfortunate extra winding tap on the primary, that violates electrical clearance on the PCB. The extra screening track used between the MOSFET and sense circuitry is also routed directly under a sense resistor.
The chip is an MT7938 if you want to look at the datasheet. It's in English, but has some oddities to the wording. But it does convey most of the useful information.
The slight shimmer from the LED is low level ripple being picked up by the camera due to the intensity of the light output.
This video is probably not going to be popular. It will only appeal to a tiny percentage of UA-cam viewers, but it still needed to be made to cater for those who appreciate the data. That's the curse of technical channels. The more technical they are, the less views they get.
If you enjoy these videos you can help support the channel with a dollar for coffee, cookies and random gadgets for disassembly at:- www.bigclive.c...
This also keeps the channel independent of UA-cam's algorithm quirks, allowing it to be a bit more dangerous and naughty.
#ElectronicsCreators
See, Clive, this is why Alec uses the magic of buying two of things.
I like the "be a good neighbor" power factor, rather than the usual "every circuit for itself"
It is indeed, nice to see a circuit that behaves nicely.
It's a nice thought, but in high power land where I come from, I don't give a stuff about my neighbors; I just want to be able to get the kilowatts I need without needing a mains supply capable of over 9000 A rms.
I have enjoyed my 0.5% on time power supply sipping power for decades approach, why, because I have already paid for the pole, transformer and wires 3 times over.
Some of these designs never cease to amaze me. Many are breathtakingly bad, or even "delightfully" dangerous, and some are pretty brilliant. In my experience, China indeed is a country of extremes.
Yeah but it makes me wonder WHY they're coming up with different designs again and again and again. If one like this has such a great PF why don't they all start using it and be done with it? It's like they're trying to reinvent the wheel every time they need a driver.
@@Ni5ei Probably too expensive for many applications.
@@Ni5ei often it's to save cost. sometimes it's because someone thinks they can do better
@@Ni5ei Look up "Muntzing" on Wikipedia. I suppose in current times it is not about removing the last components, but making it work with what you have (such as a stockpile of transformers).
Simple ... Because the West does not want to pay for it since they want to make maximum profit here @@Ni5ei
I would be nice if you could show some scope shots of the circuit parts in operation.
It would be good for both the novice and pros.
Yeah. I’d like to see the voltage on that”extra” transformer pin in particular.
When Clive told us the actual circuit was unlike the recommended circuit and showed the reverse engineered circuit my first thought was that the designer wanted to use a slightly cheaper transformer with no sense winding, forcing some creative voltage divider trickery. However that unused transformer pin looks a lot like it could be a sense winding.
It doesn’t make sense…
I second that comment. I wanted to see the output waveform, not so much at the switching frequency, rather nearer the mains frequency. I could see in the camera shot, there was considerable flicker. That would be a game stopper for me to use this indoors. It would however be perfectly acceptable outdoors. They have traded quality of light (no flicker), for a good power factor. It has to be said that getting a good power factor using a small inexpensive circuit is surprisingly difficult. It would probably need two switching regulators one following the other. The first (just like this one, chopping the raw rectified mains, in step up config), would generate the `bulk supply` with ripple that does not reach zero and the true regulator would draw on that voltage, and would use a fast feedback loop to maintain true DC.
Damn...never seen a power factor that good before on a driver...really neat circuit :)
Using the stray inductance snubber as the auxiliary supply for the control chip and open circuit detection for the load side, clever. And while switching, the circuit floats itself above rectified line voltage to do high-side switching without needing a dedicated boost capacitor for the MOSFET either. Having lots of stuff attached to a switching node can be problematic for EMI but is likely fine here since it is all packed so tight together and inside a metal housing or fixture of some sort under normal circumstances.
This is still a LOW side switching unit. The mosfet source is still "dc" grounded through the winding...although lifted up by some potential diff. when "on".
Just for info....back in 1990-1995, In high power "softstarter" 3ph. AC DRIVES. (100-700HP range with 6 thyristors) we implemented some isolated aux.power supplies powered from the 3 current sense transformers.
@@analoghardwaretops3976 Your definition is incorrect. Low-side switching switches the ground leg of a power circuit while high-side switching switches the V+ side. The MOSFET's drain is connected directly to V+ and its source controls the V+-facing side of the load, which makes it high-side.
It is the side of the DC rail the control device is connected to that determines whether it is high-side vs low-side, not the off-state DC potential of the FET's source.
Thank you for letting us know about the hoppi. I was concerned at first that the hoppi was no more.
It would have been fun to observe some waveforms of the circuit in operation.
I think Clive plans to be buried with him clutching it close to his chest
1. It sounds like a critical-conduction mode controller. This gives good power factor. However, it requires bigger output caps to keep the flicker down.
2. Clive, it would be great if you added the polarity dots to your schematic to show that the circuit is a flyback (i.e. transformer isolated buck-boost), and not another topology.
3. PFC operation requires two feedback loops: a fast loop at the switching frequency to control/limit the instantaneous current and a slow loop (slower than the utility mains frequency) to set the peak current and do the regulation of the secondary. Critical conduction mode is not constant frequency.
4. The IEC 61000-3-2:2018 power line harmonics standard doesn't have any requirements for lamps less than 5 watts. This being a 20 watt item says that there needs to be some PFC in order to be OK.
I do like seeing the schematics of circuit designs that have such good power factor scores.
I haven't looked at this driver's datasheet but I once designed a circuit using a somewhat similar one. The way these typically get close to unity power factor is that they run the MOSFET using a fixed on-time. That way the current through the transformer or inductor climbs up to an amount proportional to the voltage applied, and then is left to freewheel until it stops, then the cycle starts again. CS must be for current sensing; usually it quickly lowers the on-time when it's above a certain threshold and lets it creep up slowly if it's not. The feedback pin is probably there to sense when the transformer has finished freewheeling, triggering the next cycle. So the way it draws current is kind of like a high-frequency triangular wave with an amplitude proportional to the applied voltage, hence the great power factor. A neat side effect is that they usually work quite well with TRIAC dimmers, too. Only thing to look out for is that it does flicker if you don't add smoothing to the output.
I'm glad you told us where the Hopi meter was... I was literally going to ask that 😂
Instead I shall ask what the "FUN" button does on the meter you did use...
the "FUN" button peaked my attention too.
It means FUNction.
When the fun button is pressed, a random lawyer explodes somewhere in the world.
@@BigClive and you haven't eliminated all lawyers by building an auto-clicker? 🤔
@@notahotshotI'd personally finance whatever that auto clicker costs 😂
A very smart implementation... and a GREAT powerfactor!
Coming fom the 'old' analogue days, I am slowly being won over by switched-mode PSUs like this... No inrush current limit or MOV on the input will just kill this very early in its life!
Great video, thanks Clive!
In the early days I hated switchmode power supplies because they were to troublesome. Now they have been simplified and ruggedised to the point that I'd even consider building one into a project on the PCB.
@@BigCliveearly/ 1st. gen.controllers were just voltage mode with poor sequential and/or current limiting control....
Eventually up the learning curve brought about vast improvements in voltage mode..and now also a variety of current mode control techniques...
so they are easier to control and extremely reliable today.
What inrush current though? It's got one tiny cap on the primary side, that's it.
@@BigClive We are slipping Clive, we are slipping!😬
Give me a transformer the size of a small SUV and a heatsink the size of the Eiffel tower! I demand durability! And as many Deathbeam Caps as I can get!! 🤣
Very interesting idea, Deep explanation and analysis but i need to watch it two times more to fully understand. I remember that many years ago a colleague of mine, an electrotechnician, made an electronic circuit with some inverse laplace transform to minimize the garbage coming back to mains, even the connected circuit could be generating spikes , not only the power adapter.
Unity Powerfactor would be a great name for a Bond girl.
Who's the bad guy? Electro?
@@MostlyPennyCat Shocking!
@@TartyVesthandle
Will Mr bond, what do you think of my.....
......
_Electric Eels??_
*presses button*
As i didn't see this commented:
There is no need for a feed back to control the secondary's voltage. The load is a diode (or a string of LEDs) and those want to see a certain current, not a certain voltage.
So the current in the primary and the transformer ratio are designed such that the secondary drives the required LED current (regardless of the LED voltage that will vary considerably over temperature).
And imho the circuit can be made dimmable by changing the voltage divider to the FB pin as that should influence the mean of the primary's current.
That spark gap looks set for electrostatic voltage levels.
That looked like a lot of work. Thank you, keep working.
Excellent! Smart design. Thanks as ever, Clive !
Well, I'm really looking forward to Rev:B of this board. Seems well designed for the first product!
I just like the fact that the AnTai has a "FUN" button. I think every product could be improved by adding a FUN button.
i figured out the pf on all my little usb stuff was so bad i consolidated most of them and made a pf correction circuit to sit between them. thankfully the big inductive appliance loads have built in pf correction.
Smart chip, smart circuit design and SMART CLIVE understanding it!
If every engineer just copied the data sheet, they would have to spend an infinite amount of time making a typical circuit that a fits every requirement. The typical circuits shown are normally just a simple example that will get it going. The job of the design engineer is to read the data sheet, their design requirements and build the final design around the chip. If your design is double insulated then you don't need mains isolation, if it goes above 3000m then you need greater creepage and clearance etc.
Looks like they already had a transformer without a feed back winding and did a trick to provide the chip supply, parts reuse in multiple products cuts down on a whole bucket of costs.
Dang giving totem pole pfc a run for it's money!
With the arrival of smart metering, it'll only be a short time before we're charged for apparent power as well as true power consumption. This sort of almost unity PF driver is going to become far more commonplace I suspect, especially as incandescent lighting is pretty much no longer in use.
@@BenRichards7 That's a relief, since with the move to LED, PF might have been a significant consideration.
Interesting design
- it's like a flyback based PFC circuit and LED driver in one unit.
I would be awesome if we could get some scope pics of the FET gate and output current in operation. That really is quite a clever little circuit. The MT7938 has a very poor datasheet though, I can't figure out what the frequency of the PWM is. They have a graphic showing roughly 7 slices per half cycle... that would be 700hz to 840hz depending. But no specs on the frequency.
Combining the snubber and the power supply for the chip is quite clever, but very sensitive to power. I don't think they could swing that at higher power levels. The extra energy has to be dissipated through that 62k resistor + consumption by the chip (also not specified for operating. Only specified at 1.2mA when the oscillator is turned off).
But I am impressed. The parts-count is impressively low for a power-factor-corrected current-regulated DC power supply. I don't think any of the capacitors actually have to be high voltage... 50V caps ought to do the job throughout.
Usually low cost LEDs are just run straight from an AC transformer and rectifier without any regulation. i.e. AC-driven LEDs (100-120hz after rectification). This works but winds up being very sensitive to any asymmetries in the AC input (every other post-rectified half-cycle can wind up being at a different energy level, resulting in visible flicker). LEDs can soak up extra voltage... it just turns into heat, and the power-factor is terrible. But the parts amount to one transformer and one fuse. You need to use thick-phosphor-coated LEDs to get a soft output.
-Matt
Clive, A great investigation work!
MT793x series ICs seem to use Critical Conductive mode to do the PFC, some of that company's products use QRM quasi-resonant mode instead.
The board you have uses N- channel mosfet while the load is on the Source side (kind of like BJT's emitter follower fashion.) I wonder if there's a drawback on such design.
BTW, I could not find the board you have from Ali. So I bought a 50W version which also do not have a large primary side storage cap, but just a 470nF filter cap. I found it is using a different chip LZC8650 which uses QRM flyback and the PF is measured as 99%. The IC's power circuit is traditional using an aux winding and the transformer winding is on N-channel Mosfet's (traditional design) Drain side.
Some regular SMPS MOSFET / IGBT gate drivers even from reputable manufacturers can actually be abused as a resonant active PFC driver as shown here in this particular power supply. It takes some analog wizardry (SPICE simulator may help a bit if you are lost) to make it work.
Super smart design. Love it.
This is how Bob Carver's "Magnetic Field Amplifier" power supplies worked. Basically, a big light dimmer driving a transformer.
That's a blast from the past. I worked at Carver back in the 90s. I loved working on those amps.
That took me down a good rabbit-hole! Also to Sunpower mag amps. They note that they operate fine in high neutron environments - I presume their location in Aldermaston is no coincidence!
Nice LED driver !
You had me at unity power factor!
Thank you for making these video's, they are very interesting and educational for someone like me at least. Appart from the power-factor, I would also like to know the energy efficiency of the driver. It would be nice too if you could mention -or test- its compatibility with a dimmer and if so which type(s), even though for this one it seems quite obvious that it isn't.
Impressive design. I like it.
That’s an interesting bootstrap circuit for high-side Nmos driving. But I think it’s just that, there is no other low-impedance path for it to also act like a snubber, besides by dumping its energy into the power rails’ cap.
This doesn't look too bad. I'd buy one if I needed to drive an LED specifically from something like this.
MOST power factor power corrector devices have a working voltage of 80 to 400 volt rating (or 265Vac) very universal, nice !
I’m shocked, shocked I tell you, that a product purchased off AliExpress would be using a driver IC in some out-of-spec fashion!
The data sheet represents the feedback as the primary loop.
Your camera makes it clear that the LED does not have continuous light, but flickers very fast. The human eye really prefers constant light. Even high flicker rates that aren't obvious do have a detrimental effect, causing stress and headaches in most people. I modify any LED lighting I use so that it has pure DC light.
But do you keep the fluctuations and other deviations on the input down?
How do you create pure DC?
Battery or lots of filtering and capàcitance.
@@phillyphakename1255, 'approximate DC voltage', produced by inductor/capacitor filtering.
I’m curious. I guess the capacitor between primary and secondary is why you get a mains buzz from the case of many devices, but what’s its actual function? I’d have thought that a secondary isolated from the primary would have been safer.
It provides an easy return path to ground for coupled noise that could otherwise be radiated by attached cabling.
@@BigClive would not isolation plus a common mode choke not be a better solution?
@g8xft Yes - it can give rise to that tingling sensation. In small switching supplies, the capacitance is generally in the range 1000pF to 2200pF and so the current coupled is very low. You can often see the effect though if you measure the AC voltage between the load side of the supply and mains earth, since the resistance of the multiplier circuit in the meter doesn't load the capacitor. With UK mains, a reading of 80 to 100 Volts AC or so is very normal.
No matter the branding, it displays the frequence.
Why is there a spark gap? I am familiar with this in high voltage circuits like the Jacob's Ladder, where it acts as a crude switch: once the voltage is high enough a spark jumps the gap and what was an insulator becomes effectively a short. I've never seen it in a lighting circuit.
They are sometimes found in power supplies where there is a risk that an electrostatic discharge to the output connector could stress the electrical separation in the windings.
if this had improvements to your 2 concerns it would be great for anyone with solar power (partially or entirely solar) as most inverters don't "like" bad power factor (ranging from increased power consumption from batteries/panels to strange behavior to noisy AC output).
very clever bias voltage circuit, i tried this once and failed miserably!
Actually pretty clever.
reminds me of the "link switch" brand IC.
A very interesting chip indeed. The oddest thing about it is the voltage on the FB pin goes negative whilst the MOSFET is on. Also I doubt it does operate at resonance. There's no additional capacitance to resonant with, this just looks like a standard flyback converter, where the output is regulated by the peak voltage on the flyback transformer, rather than feedback through an optoisolator.
I pondered that too, but even with a separate FB winding it also goes negative.
By resonance I mean the natural time of building and collapsing the magnetic field. That will also be affected by the load.
Had a look at a similar module last week, the QH-20W variant, possibly the precursor. The rather expensive plant hibernation bulb it was inside was dead and the output caps domed/bulging. The sens resistor must have have gotten quite hot based on the blackening around. Strangely it appears to be a stack of two smd resistors on top of each other… Wasn‘t able to find which resistance value it should have been, the 7938 datasheet I found and translated didn’t provide any values, unfortunately.
The sense resistors often determine the output power in LED drivers. You may find you can tune it with some experimentation. Probably starting around 4.7 ohms or 10 ohms. Higher resistor value will usually result in lower LED power. The output caps would need replaced with new low ESR/impedance ones. If the dark resistors measure a few ohms then they may be OK.
Nice video as always Clive. Have you ever considered doing a video where you show how you reverse engineer a PCB?
Thank you for the video. Seems like a little wrong with the schematic. The diode D3 is also connected to the C3 as on the PCB you drew. The polarity of the D1 also looks strange... Thank you!
I guess that they thought that it would be potted in something so the chance of that stray pin contacting the via would be minimal, except in reality, the pin's probably going to go bang... :P
The way that it is set up with current sensing makes me think that it is made to work with any transformer that you put into it, so maybe that is why that design flaw exists, because that transformer was not the one in mind.
LED drivers are fun to deal with them
Ya know Clive, you keep saying the Hopi is just off screen. I'm gonna need a picture of the Hopi turned on with a copy of today's news paper in the background to be sure it's really okay ;)
An LED driver done *Correctly*
Nice
Never hide the FULL BRIDGE RECTIFIER.
interesting circuit, but the issue is that you cant assume input is sine wave always, many of the invertors produce square wave and these cause flickering of many of these led lamps, and this one without any cap is probably going to be worse.
"IF" it works well for a fully rectified sinewave i/p it will work equally well on "non sinosudal " inputs...& on sq. wave i/p there shouldn't be any flicker... except if the deadtime between +&- is too large...i.e.>5-8 mSecs.
@@analoghardwaretops3976 no it doesn't, I can see it at my home, LED lamp flicker and some of them blink, sometimes I switch them off and use tubelight, they seem to function much better with square wave.
The kick-start, separate-feedback-winding-ok-keep-running design has always made me laugh (in an indecorous manner), making the inevitable hiccups difficult to diagnose and cure. Why not a simple am-I-present-and-correct monitor of the secondary, with a minimal load resistor (well, simple in my head, anyway).
That's quite a neat circuit. I love the good power factor it has. One question: is it dimmable?
Not dimmable.
Are there any tweaks we can do in the LED driving circuit when running on 50Hz mains that will make the LED not flicker when recording at 1/60th shutter speed?
Yep, those output capacitors and the much higher switching frequency greatly reduce output flicker, but no measurement is done in this exploration .
Actually there is. There's a circuit called a capacitance multiplier that provides additional smoothing to the capacitors that are already there. You give up a small amount of efficiency as compared to just using bigger capacitors but I reckon it'd work.
Driverless 12v LED modules and a decent size battery? There's no chopping and pulsing with a battery.
Use proper DC power supplies for electronics use, for example an ATX power supply. They should have negligible ripple.
@@ManWithBeard1990 Adding more capacitance only slightly lessens the intensity of the flicker but it still remains present if shooting at 1/60th of shutter speed.
You said capacitor is only over one of the ac lines, why different in schematic?
So i've noticed LED lightbulbs have been getting energy rating F a lot under new EU rules. Could this be why? a lot of LED bulbs (Cheap ones especially) seem to have the older drivers with bad power factor
The calculation for the energy rating seems to be calculated from the lumen output and the power usage, not the power factor. However I'm not sure if the power consumption is active or reactive.
I think it is called a chopper stabilised psu.
Just had such a strange experience.
A friend asked for help with a led strip she wanted to install. She hooked up the leds, they work but flicker a bit, and asked me where to put the power supply.
Then she gave me a packaged power supply, and showed that the leds worked when she flipped the switch.
I was bamboozled.
When we put a power supply in between, we saw that some parts of the strip were mostly dead, and some sill worked.
But this strip failed in a way that it mostly works on 230V
Do you want it to test?
Is that a FULL BRIDGE RECTIFIER in the top left corner?
Drink a shot for every utterance of "I Shall". Good luck.
What, a relatively well designed circuit!!!!????? (altho poorly designed board)...what the hell did they plan the Spark Gap to actually flash over...holy mackarel!!
Forgive me if I am misinformed, but I understood that power factor is not very important in a domestic (household) installation, since real (kW) is metered, not reactive power.
Your electric bill wont really be impacted by the power factor of your appliances.
So unless these LED drivers are being deployed by the hundred in a commercial / industrial setting, it really doesn't matter too much what the power factor of them is.
If I am incorrect, please explain.
Nice
This could potentially feed a lot of RF noise into the grid, doesn't it? There is no common mode filtering.
Could you cut the "extra" unconnected pin on the transformer?
It's not very accessible without removing the transformer and refitting it.
Would you be able to add a bigger capacitor to the output so there's no flicker at all? Would it effect the power factor?
The camera saw a slight shimmer. But it wasn't visible to the eye.
can you make a vid about tearing down a ballast for one of those decorative neon lamps unless you already have made one?
Search my channel for CCFL.
Enlightning...
peace
So... The harmonics must be crazy. Just not that crazy because it's only 20 W, but still...
I can only guess that the design is the reason for the flicker from that LED? Did you try to measure the voltage on the output with a scope to check the ripple?
It does have slight ripple on the output. But not much and only visible on the camera.
The problem with chinise led bulb they over power the tiny led with bad thermal dicipation and they put 5 or more of them in parallel, after a chort period of time u will find a dead smd led or a blown capacitor.
I noticed on your order sheet there they offer a “waterproof driver” option. I wonder if that has a conformal coating or is epoxy potted to address your concerns over high humidity.
It's available in a small enclosure filled with a rubbery compound.
Good PF, horrible flickering. It’s like it’s one or the other, but you can’t have both. (You can though: there is DC output powersupplies with good PF. Its just the LED-drivers. And possible the non-flickery ones just don’t draw my attention)
The camera caught the slight ripple due to the intensity. It will even show it on a tungsten bulb.
Could the sense pin and voltage divider determine the input AC voltage in the first schematic because it's dealing with PF
Sounds like it would shut off if C4 & C5 were replaced with 250uf capacitors to get rid of that flicker :/
13:25 please carry on drawing Garfield 🙏
Hello, can you give a link to this on aliex? I wasn't able to find it by a brand search. Cheers!
www.aliexpress.com/item/32922801946.html
Now fiddle with the voltage sampling resistors to make it a dimmable controller.
{^_-}
The listing doesn't even seem to mention the great power factor…
It doesnt really matter.
Ahh Powerfactor.
You don't realize how big of an issue it is until you see what is needed to correct it on the powergrid side.
Good power factor .Now can you make it better ?
Is 80% load (400ma of 500) on an ac/dc wall warts okay 24/7 powering a computer fan?
I cracked the case from it always running quite warm and figured I could fix things by moving it inside the hepa box and into the airstream. Its running about 60c on the transformer.
Am I just creating a worse fire hazard by not getting a larger supply? I did put a 2 amp fuse with the fan but I dont think the problem will come from there.
*turned off autocorrect finally so I can actually type now. Still not an electrician tho.
I prefer to run long term use power supplies at half their current rating. So usually a 1A 12V power supply for a fan. I also use power supplies from a local component distributor to make sure it has safety compliance.
Chopping the sin wave off just sounds painful
Is it dimmable?
Yes very clever circuit, but what does the 1n cap do, connecting the transformer primary to secondary?
It provides a return path for coupled noise which would otherwise be radiated from the load side of the supply. Such capacitors (known as Class-Y) must be safety rated for use in that application. These are very common found in switching power supplies - everything from phone chargers upwards.
@@GodmanchesterGoblin Thanks. Just hope that these caps are really class-Y rated. If this PSU opened my parachute I don't think I would jump!
Both pri. & sec. windings have their own inter turn and interwinding capacitances...
The pri. & sec. windings get capacitively coupled through this..and if not bypassed it will cause the secondary output to swing w.r.t. the primary voltage that is rapidly switching between it's +&- .
Hence a pri.- sec. capacitor is connected between the two
" NON-SWITCHING" terminals of the pri. and main sec.( in case of multiple sec. wdgs. , the winding having highest current capacity is used).
The Y cap value chosen is usually 12-15 times more than the measured interwinding capacitance.
Just want to confirm, the LED will not flicker, right?
Very slight ripple detected by the camera.
Interesting
Next up an even prime schematic, which is also odd.
What does it do in the event of ultra-catastrophic failure (like throwing it into salt water)? I hope the answer involves lots of fire and poppage!
That's the standard failure mode of most switching supplies.
Is that a true power factor rating or just displacement power factor?
It's averaged over the sinewave, but still much better than the simpler primary circuitry.
Is there a known more dependable LED power supply? I'm glad the LEDs last 50,000 hours but the power supply lasts an hour and a half. I repair my stuff and would like to have an LED power supply that lasts a lot longer. Is there a type or class of supply that will hang on?
They could make them with solid polymer capacitors, but don't.
@@BigClive Not sure where my question went... If I replaced the capacitors with solid polymer ones before installing the LED fixture that would extend the life?
@@billbucktube Changing capacitors is usually a good upgrade. That and reducing the power to the LEDs a bit.