The zeners may have another function. If the supply voltage is briefly lowered below the LED+zener voltage, no current will flow in the leds and the supply line can be used as a data line to send data to the chip.
Only problem with that is voltage droop along the LED string which varies by location and current draw across the whole thing. You will also get detection range changes from LED and zener temperature. The pull-up resistors are likely what the chip uses to sense whatever signal it is looking for while the LEDs are blanked off.
@@teardowndan5364 I was thinking similar - if they dip the supply, the LEDs will all be guaranteed to be off, so power draw would be low and consistent, allowing data to be sent without voltage drop problems
Was thinking same first power up it'll be red and maybe that's enable programming mode recieving through other 2 while they off ..... Clever bit of tech😊😅
@@mikeselectricstuff it ads the "bypass" resistors that can be the address. electrically with 3 resistors you could have 8 addresses but the values might even ad addresses to them in the way it was done with the first wired resistor remote controls.
I'm not sure, but if you have a look at DCC used for control of model trains. It is a high frequency square wave AC. The voltage does vary depending on what scale you are modelling but the most common is specified at 13.8v, so after rectification, the full 12v DC is available for the motors. This gives a constant voltage available all the time and the data bits encoded into wave form. While this is a fairly complicated communication protocol. I think a much slimmed down version could be used for these. Also remember we have things like one 1-wire. Some really smart people have come up with some weird and wonderful things, that leave us mere mortals scratching our head on how they did it.
...Beaut, Thanks for teardown, indeed very interesting 2-wire methods used. Very pleasing that 'Twinkly' have warranty replacement circuit boards available. Well done Twinkly! Surprising that the resistors were the part that often failed first - would have expected SMT resistors to be hermetically sealed. Ya learn some thing new every day. Also rather surprising that a robust conformal coating had not been applied at the factory given that they are aimed at the 'premium' market and Twinkly obviously value their reputation. I feel for the manufacturer, imagine realizing that a batch of slightly non spec o-rings had snuck into your production line at some point.
@@LeifNelandDk Don't blame the space shuttle (well, the side solid fuel rocket booster) o-rings as they worked as expected. The problem was management. Previous to launch they were informed that it was colder than the vehicle had been tested, yet they went ahead with it anyways. If you use an item outside of designed and tested specifications while ignoring the experts who designed and build it whom are telling you not to use it, then don't blame the equipment if it fails. I'm not saying that the o-rings were great because there was a history of erosion on them after launching. O-rings made from another material probably would have been a better choice but it's easy to make such statements almost 40 years later when great advances in materials science have been made.
Oh, crazy! I only just recently learned about the pre-addressed LED pixels from some cheap curtain lights. The fact these can transmit data along the power lines is super fascinating.
Long ago when I was a boy, there were several devices that sent signals on the power lines. As long as you stayed below a certain power level, it was legal and worked quite well. This was in the analog days. You could have transmitters and receivers that used the power lines. Intercoms often worked that way. No need for antennas, just plug two of them in and they worked.
Another viewer mentioned already trains but this is also used at thermostats of Honeywell. These units have only two pins + and - (gnd) and send the data over these two lines by using a threshold voltage, the data is above the threshold voltage to make a digital signal. I think this method is developed to make easy to add newer hardware to an old house with existing (12v) wiring and don't need any modification. Very clever, very handy. There is detailed documentation available about the protocol they are using, very interesting if you want to know how it works. I bet if you use a scope on these lights to monitor the voltage lines, you can see the blips of data.
trains? don't talk to me about trains. a few months ago i saw James May recondition his 1971 toy train, it was from an old 'toy story' episode from 12 years ago. i thought, 'hmm a small complicated thing, with an electric motor', that is pretty cool, despite me not being interested in trains or model making ordinarily. then last week the guardian newspaper mentioned z scale trains. they are small enough to fit on 50p piece, the track gauge is .25 inches. one can put a loop of track with a mountain in the middle on a sheet of A4. i saw this and thought 'that is ruddy amazing'. i am getting good at soldering SMD, and so i am also looking at microscopes to buy (stereo or one with a screen?), so really small trains sound fun. it seems they may use coreless motors, rather than brushed DC like OO toy trains. obviously your trains, and my trains are completely different, but in lieu of any friends i ramble on here instead. cheers!
@@KarldorisLambley Did you mean T gauge, not Z gauge scale trains? T gauge are ridiculously small and use magnets so the trains stand a hope in hell of staying on the track.
@@nickryan3417 T are half Z, 3mm gauge rather than 6. but they are ludicrously expensive. at the moment i cant even find a kit on ebay. so ill just use the twice as large, but still minute z scale.
Reminds me of the eBus protocol used for heaters in Europe. It uses two wires for power and data. The voltage jumps between ~10V and ~20V, every device can send and receive. A device willing to send must wait for a "bus free" byte, which is sent at the end of every packet, and must not appear in any transmitted data. Then it sends its ID and simultaneously reads it back. Through clever overlapping of bits it can determine if another device also wants to send, and will wait if the other device has a higher priority. The addresses are one byte and determined by device type. Every device may pull up to 18mA from the bus for powering itself, which is enough to drive a display. Many devices even contain a full bridge rectifier so the polarity does not matter. I built an adapter recently to convert the bus signals to 5V TTL, now I can analyze and control what my heater does without a display.
There's an unpopulated resistor that would run between +24v and the _red_ drive signal. I think the placements (and resistances?) might be different on each board, as a way of baking in the address without needing to flash the chip or anything
Thanks for the video, Clive. It's -18° here which is probably too cold even for your liking! So I have refreshing beverages and UA-cam until things warm up.😊 I've been going back and re-watching your build/storytime/Q&A videos.
I messed up the polarity and blew the electrolytic capacitor ages ago on a gift of a philips striplight, and then I managed to crack the board. Always been too timid to try and fix it, but watching your videos has given me some gumption to try save some money. The details are scanty ((h2...100v...CL and is 6.3x7.7 ) so I started with a 10uF. Lights come on but are super dim, so I've ordered some more (47uF...etc) and will swap it out until I get the right one. Of course, I will inevitably be sending it to you when I cock it up. ;-)
The chip could alternate between lighting green and blue after every decoded packet if the protocol uses a delay between packets. That way one of the pins can always be used as an input. Maybe even between single bits? The voltage divider and zener on the left should be the supply, except if it has an internal capacitor. The DS18B20 for example can use one pin for data AND supply in parasitic mode. Would be interesting to have a look at the 24V rail with an oscilloscope to get the voltage levels!
I wonder if the addressing of the module isn't just set by R3, R4 and R5. Setting a resistor on R4 and R5, like on your board sets it to "2", but other combinations give other addresses. If I'm correct, this would mean you could have (3 bits =) 8 different addresses.
With the other light using 20k resistors, I wonder if the value also plays a part. Having the options as open, 8k2, or 20k gives quite a few additional addresses.
O-rings, the bane of my boiler, the damned diverter valve cartridge that fails annually is always cos of o-rings around the plunger failing, and usually not replaceable because they're pressed & crimped in, so the one that's in there now back in december I actually managed to take to bits (they were lazy with the crimps) and replaced the worn, stiff, cheap black things with the soft and supple red silicone kind, and put a load of silicone lube on the shaft (giggety) on assembly, and so far, no leaks, but time will tell, Baxi/Potterton boilers are shite at best anyway so I expect it to crap out at any moment regardless (alredy replaced the expansion tank too!)...
It's interesting that Twinkly (one of the first of these products at a consumer level) doesn't use addressable LEDs whereas the cheaper clones do. That makes the systems from Govee etc far more hackable than the Twinkly ones. In a (low budget) theatre production we used loads of the cheaper Govee LED stuff, but with our own drivers on the end to allow us control of them by sACN rather than the standard controllers.
Other 2-wire power and data schemes are eBUS used in heating systems ( en.wikipedia.org/wiki/EBUS_(serial_buses) ) and "1-wire" used for temperature probes etc ( en.wikipedia.org/wiki/1-Wire ). These are both bidirectional too. 1-wire uses baked-in unique addresses (like MAC addresses) and a clever mechanism to identify all devices on a bus without having to poll every address to do so.
I think the "bypass" resistors that can be the address. Electrically with 3 resistors you could have 8 addresses but the values might even ad addresses to them in the way it was done with the first wired resistor remote controls.
i never heard of these twinkly lights, so i had a look online. bloody hell, they cost 170 quid for 40, on 60 feet of wire. so each of those little PCBs cost about 4 quid each. seems a bit costly when one is buying 40 of them. mind you i get my e-waste from Poundland, so more than a tenner for lights seems lot to me.
24V LED strips usually use groups of six series LEDs and a dropper resistor. However, if you use a transistor configured as a constant current sink, you can use seven instead, increasing efficiency by 17%. (Obviously it's more complicated in this application and that would probably be impractical, not least because there may be considerable voltage drop over long cables.)
I use these LED strips in my kitchen and garden, powered from old laptop supplies which I have opened up and changed the potential divider so they output about 23V.
So the chip needs to quickly change the G and B pins between input and output modes? That would require a time synchronization method. Impressive, but seems like a lot of work just to avoid a third rail for communication. The factory defined addressing avoids the need of separate input and output pins.
Fascinated to know if these are using a microcontroller or some custom chip, I spent a little time searching for SOT-23-5 microcontrollers but they're really quite rare. I know it'd need some gear that's not necesarily in every hobbyist's workshop but I wonder if anyone's decapped one of the chips and stuck it under a microscope, might be interesting to see... Feeling the need to hunt down some 'Twinklys'
i got a set of addressable christmas lights that are literally a controller with bluetooth and wifi and a string of leds+wire+heat shrink that use 2 wires for power AND data, not 100% sure what type of chip they are using but they can show upto like 8 different colors at once before repeating the pattern
Twinkly festoon issues ... Sounds like a Python skit: Rumpletweezer ran the Dinky Tinky Shop in the foot of the magic oak tree by the wobbly dum-dum bush in the shade of the magic glade down in Dingly Dell. There he sold contraceptives... 😮
Another informative video from Big Clive - How to put the twinkle back in your Twinklys! 😁 Would the IC on board be programmed as they produce them? rather than holding stock of each individual number? Also leads to the question "how many of these are in each display/set?"
Am I the ONLY one who sniggered when Clive said "No. 2" ??? Great video once again, thanks Clive! Would it be possible to hook up a scope to the rail/components to see where the data flows (instead of the water? 🤣)
@@BigClive very interresting how they manage the protocol.. but a bit outside the scope of you typical teardown/fixing things! Nevertheless, a great video. Keep up the good work, thank you Clive! Oh! On the topic of OH-rings 🙂 - I see that suppliers normally only quote the inside diameter plus the cord thickness. Probably because the things are so squishy from the outside!
Certainly a very interesting system. I think the 15v zener is to strip off a large part of the 24v supply so that a larger proportion of the data, that is present, can be seen across the 3k3. I am not absolutely sure about that, but the 15v zener isn't to stabilise the 9v. I think the supply to the chip is via the two 8.2k resistors?
Interesting. I have a string of these outside the back door and six of the twenty are in various failure modes (when set to all warm white, some of the failing ones show green, two have no output, one is sometimes blue but corrects itself after a while). Time to get de/soldering….
There's a page called "Controlling Twinkly leds directly with a microcontroller" on AVR freaks where they've reverse engineered the protocol. "It appears that the waveform is stable for 6 microseconds and drops to half voltage for 2 or 4 microseconds depending on whether it is a HIGH or LOW bit. When I turned off all the lights, I only visualized the wave with 6us at full voltage and 2us at half voltage, so this makes me think that the 2us corresponds to a bit 0." Not sure how each module/LED gets its address.
Keith westley made two videos "Hacking Twinkly Part I" and "Hacking Twinkly Part II". The LEDs are QED3110 devices and it seems like each one has an address programmed.
I would think that the omission or inclusion of the four resisters, (three across the LEDs and the one across the high value zener) could be detected for numbers 0 to 15. so this one is number 2 and sure enough one resister is missing giving 1101 or 0010. How are some of the other numbered chips wired? It does seem odd that the power pin would be used that way.... but maybe that is not the (only) power pin but rather a LED replacement circuit. even with none of those resisters there may be enough power from the four i/o pins to power the chip.
either these pre-date pixel LED's with the driver on chip and a data line or Twinkly wanted something that would remain entirely proprietary. 16+ SKU's instead of one seems a little OTT to me..
I was wondering how the chips were programmed too. I couldn’t see any pogo pin spots at first making me think they were programmed before placement, then I noticed what look like seven through plated holes. If they connect to the chip they could be the programming port.
Resistors on the right are for calibrating white balance or matching perceived light intensity for all colors - current is fixed in chip and redirecting part of it will reduce intensity of this channel.
I am no expert on this and I will be perfectly honest and say that this is the first I have heard of these but the communication method sounds very much like a CanBus style protocol with a time synchronisation reset…maybe However, if it was something as simple as that then that would’ve been picked up already by people trying to reverse engineer it. Maybe some proprietary methods over the top of a well established high speed two wire method? Nice video as always
Isn’t CAN (2 wire) using both lines for positive voltage? But yes the signal is always in a range between 1.5 and 3.5 volts. Both wires sum up to the median 2.5V. I guess only 1 wire could be enough if the zeners put the voltage to the lower level to eliminate pulsing. (Probably not even visible if not if its a fast data rate)
@@domnik9062 Yes, CAN bus is two wire. The signal value is determined by comparing the relative voltages of the two wires, CAN H and CAN L, which almost entirely takes out the impact of the incredibly electrically noisy environment that CAN bus was designed for. The concept is that voltage spikes will affect both simultaneously and therefore it's just a case of comparing whatever happens to come through. Voltage drops can be more problematic due to them affecting the circuitry more directly than EM interference in the wires or voltage spikes in the power source. For something like this application I would expect a combination of data transmissions to one or more recipient nodes as well as an application broadcast transmission that triggers all nodes to apply the data they have just received. It's quite likely that there is also a combination of deltas as well as absolute values in the communications protocol, but that could be complicating things more when it comes to optimising bandwidth compared to control.
FOUR ! ! ! Zener ( MELFS ? ) , Metal - Ended - Leadless - Flange ( actually : Face , but , I prefer Flange ? errmm ! ... ) .. this has made my day , but lotz of MELF's on a cheapo PCB ? ... has 2 B good ( ? ) .... OMG .... DAVE™🛑
Would have been nice to see one of the actual boards in operation. Perhaps even diagnosing and replacing defective resistors. Good information but boring for actual demonstration. Thanks Clive
Twinkly has really let me down. As long as their products work, it's absolutely amazing, but the longevity is questionable. I have taken really well care of it, but the power supplies only last a season or three, my two strings had leds starting to fail at the end of the first season, and now by the thirds season they are no longer usable as strings, unless I pick an effect with "rainbow colors" since so many leds are the wrong color.
Clive, I wonder if the thing you have marked as the power input is actually the data input, and the chip gets its power from some or all of the LED drive lines. That would explain the lack of a filter cap on the "power" line.
Has anyone measured the data frequency on the power lines yet? I imagine it's up there a ways... I used to like to shop for deeply discounted RGB light sets after christmas, as they all used to have decent 4 pin RGB LEDs in them. But today they have mostly all went to these 2 pin RGB types, and arent nearly as useful as the 4 pin types as far as hacking goes.
Oh my goodness Clive - thanks so much for looking at them for me, what interesting boards!
Wait until Vince finds out you've been sending stoff to clive instead of him ;)
@@ruskythegreat hahaha! Vince knows how to fix things, but Clive knows how stuff works 👍🏼
recognised the name from the MMV Massive!@@ruskythegreat
@@ruskythegreat I was gonna say it.
The zeners may have another function.
If the supply voltage is briefly lowered below the LED+zener voltage, no current will flow in the leds and the supply line can be used as a data line to send data to the chip.
Only problem with that is voltage droop along the LED string which varies by location and current draw across the whole thing. You will also get detection range changes from LED and zener temperature.
The pull-up resistors are likely what the chip uses to sense whatever signal it is looking for while the LEDs are blanked off.
@@teardowndan5364 I was thinking similar - if they dip the supply, the LEDs will all be guaranteed to be off, so power draw would be low and consistent, allowing data to be sent without voltage drop problems
Was thinking same first power up it'll be red and maybe that's enable programming mode recieving through other 2 while they off ..... Clever bit of tech😊😅
@@mikeselectricstuff it ads the "bypass" resistors that can be the address. electrically with 3 resistors you could have 8 addresses but the values might even ad addresses to them in the way it was done with the first wired resistor remote controls.
I'm not sure, but if you have a look at DCC used for control of model trains. It is a high frequency square wave AC. The voltage does vary depending on what scale you are modelling but the most common is specified at 13.8v, so after rectification, the full 12v DC is available for the motors.
This gives a constant voltage available all the time and the data bits encoded into wave form. While this is a fairly complicated communication protocol. I think a much slimmed down version could be used for these.
Also remember we have things like one 1-wire. Some really smart people have come up with some weird and wonderful things, that leave us mere mortals scratching our head on how they did it.
Who you to call me a mere mortal? I'm at least a whole mortal...
...Beaut, Thanks for teardown, indeed very interesting 2-wire methods used.
Very pleasing that 'Twinkly' have warranty replacement circuit boards available. Well done Twinkly!
Surprising that the resistors were the part that often failed first - would have expected SMT resistors to be hermetically sealed. Ya learn some thing new every day.
Also rather surprising that a robust conformal coating had not been applied at the factory given that they are aimed at the 'premium' market and Twinkly obviously value their reputation.
I feel for the manufacturer, imagine realizing that a batch of slightly non spec o-rings had snuck into your production line at some point.
Finding out the o-rings were out of spec must be an expensive oversight. I hope they can go back to the manufacturer for the dud parts.
O-rings. The bane of everything from twinklies to space shuttles.
@@LeifNelandDk
Don't blame the space shuttle (well, the side solid fuel rocket booster) o-rings as they worked as expected. The problem was management. Previous to launch they were informed that it was colder than the vehicle had been tested, yet they went ahead with it anyways. If you use an item outside of designed and tested specifications while ignoring the experts who designed and build it whom are telling you not to use it, then don't blame the equipment if it fails.
I'm not saying that the o-rings were great because there was a history of erosion on them after launching. O-rings made from another material probably would have been a better choice but it's easy to make such statements almost 40 years later when great advances in materials science have been made.
@@LeifNelandDk- and garden hose couplings
Oh, crazy! I only just recently learned about the pre-addressed LED pixels from some cheap curtain lights. The fact these can transmit data along the power lines is super fascinating.
Isn't it just similar to DMX? A technology that has been around for almost 40 years.
Is there a part number or spec sheet?
@@bojohannesen4352DMX doesn't mix power and data
Long ago when I was a boy, there were several devices that sent signals on the power lines. As long as you stayed below a certain power level, it was legal and worked quite well. This was in the analog days. You could have transmitters and receivers that used the power lines. Intercoms often worked that way. No need for antennas, just plug two of them in and they worked.
@@bojohannesen4352 DMX has a separate data line. I think CAN bus is the only widely known protocol that’s somewhat similar?
Another super interesting video, thank you Big Clive! These lights are not exactly cheap, hope this helps those out of warranty.
Another viewer mentioned already trains but this is also used at thermostats of Honeywell. These units have only two pins + and - (gnd) and send the data over these two lines by using a threshold voltage, the data is above the threshold voltage to make a digital signal. I think this method is developed to make easy to add newer hardware to an old house with existing (12v) wiring and don't need any modification. Very clever, very handy. There is detailed documentation available about the protocol they are using, very interesting if you want to know how it works. I bet if you use a scope on these lights to monitor the voltage lines, you can see the blips of data.
trains? don't talk to me about trains. a few months ago i saw James May recondition his 1971 toy train, it was from an old 'toy story' episode from 12 years ago. i thought, 'hmm a small complicated thing, with an electric motor', that is pretty cool, despite me not being interested in trains or model making ordinarily. then last week the guardian newspaper mentioned z scale trains. they are small enough to fit on 50p piece, the track gauge is .25 inches. one can put a loop of track with a mountain in the middle on a sheet of A4. i saw this and thought 'that is ruddy amazing'. i am getting good at soldering SMD, and so i am also looking at microscopes to buy (stereo or one with a screen?), so really small trains sound fun. it seems they may use coreless motors, rather than brushed DC like OO toy trains.
obviously your trains, and my trains are completely different, but in lieu of any friends i ramble on here instead. cheers!
@@KarldorisLambley Did you mean T gauge, not Z gauge scale trains? T gauge are ridiculously small and use magnets so the trains stand a hope in hell of staying on the track.
@@nickryan3417 T are half Z, 3mm gauge rather than 6. but they are ludicrously expensive. at the moment i cant even find a kit on ebay. so ill just use the twice as large, but still minute z scale.
Do you have a link by any chance to a page that describes the protocol? Would be interesting to make your own driver for a string.
Reminds me of the eBus protocol used for heaters in Europe. It uses two wires for power and data. The voltage jumps between ~10V and ~20V, every device can send and receive. A device willing to send must wait for a "bus free" byte, which is sent at the end of every packet, and must not appear in any transmitted data. Then it sends its ID and simultaneously reads it back. Through clever overlapping of bits it can determine if another device also wants to send, and will wait if the other device has a higher priority. The addresses are one byte and determined by device type.
Every device may pull up to 18mA from the bus for powering itself, which is enough to drive a display. Many devices even contain a full bridge rectifier so the polarity does not matter.
I built an adapter recently to convert the bus signals to 5V TTL, now I can analyze and control what my heater does without a display.
There's an unpopulated resistor that would run between +24v and the _red_ drive signal. I think the placements (and resistances?) might be different on each board, as a way of baking in the address without needing to flash the chip or anything
It's not using the resistors to set the address.
I've never seen Zeners used like that, very interesting
The bonus was hearing you say 'Twinkly Festoon' 😀
I just LOVE how you pronounce "schematic" :) Electronics engineer here, love your channel, keep up the good work!
I think it's a piss take.
In my day we British had "circuit diagrams".
Only foreigners spoke of "schematics".
Maybe a bit like the DS1820 so-called single-wire communications when using parasite power.
Reminds me of an automotive single wire scheme I looked at designing in the seventies
Power rail data transmission always intrigues me.
Thanks for the video, Clive. It's -18° here which is probably too cold even for your liking! So I have refreshing beverages and UA-cam until things warm up.😊 I've been going back and re-watching your build/storytime/Q&A videos.
I messed up the polarity and blew the electrolytic capacitor ages ago on a gift of a philips striplight, and then I managed to crack the board. Always been too timid to try and fix it, but watching your videos has given me some gumption to try save some money.
The details are scanty ((h2...100v...CL and is 6.3x7.7 ) so I started with a 10uF. Lights come on but are super dim, so I've ordered some more (47uF...etc) and will swap it out until I get the right one.
Of course, I will inevitably be sending it to you when I cock it up. ;-)
Very strange but interesting all the same thanks Clive 😊
I watched Kip the other day he said he was sending them to you. A interesting video.
Intresting thanks Clive and thanks kiphakes for sending this interesting item
I would love to see what these look like in action. Interesting break down in any case.
Nice light (I just looked them up) shame about the fault, but great they send spares out. 2x👍
The chip could alternate between lighting green and blue after every decoded packet if the protocol uses a delay between packets. That way one of the pins can always be used as an input. Maybe even between single bits?
The voltage divider and zener on the left should be the supply, except if it has an internal capacitor. The DS18B20 for example can use one pin for data AND supply in parasitic mode.
Would be interesting to have a look at the 24V rail with an oscilloscope to get the voltage levels!
Amazing twinkly lights 💡
I wonder if the addressing of the module isn't just set by R3, R4 and R5. Setting a resistor on R4 and R5, like on your board sets it to "2", but other combinations give other addresses. If I'm correct, this would mean you could have (3 bits =) 8 different addresses.
With the other light using 20k resistors, I wonder if the value also plays a part. Having the options as open, 8k2, or 20k gives quite a few additional addresses.
@@uzlonewolf Hmmm,, good point; didn't think of that.
Using the resistors for addresses was my thoughts too. Much cheaper and easier than using chips with different programming.
This was my thought too. No doubt Clive can look at some of the other boards to confirm or refute this, unless all the boards are number 2.
When I saw the different resistors I initially thought that they controlled the address. But it's not the case.
The 15v zener is stable enough no need for a cap that would be overkill for such a small circuit
Also a capacitor would mess up their signalling over the power line.
@@user-qf6yt3id3w yes, less sharp transitions. A clever design.
O-rings, the bane of my boiler, the damned diverter valve cartridge that fails annually is always cos of o-rings around the plunger failing, and usually not replaceable because they're pressed & crimped in, so the one that's in there now back in december I actually managed to take to bits (they were lazy with the crimps) and replaced the worn, stiff, cheap black things with the soft and supple red silicone kind, and put a load of silicone lube on the shaft (giggety) on assembly, and so far, no leaks, but time will tell, Baxi/Potterton boilers are shite at best anyway so I expect it to crap out at any moment regardless (alredy replaced the expansion tank too!)...
It's interesting that Twinkly (one of the first of these products at a consumer level) doesn't use addressable LEDs whereas the cheaper clones do. That makes the systems from Govee etc far more hackable than the Twinkly ones.
In a (low budget) theatre production we used loads of the cheaper Govee LED stuff, but with our own drivers on the end to allow us control of them by sACN rather than the standard controllers.
Other 2-wire power and data schemes are eBUS used in heating systems ( en.wikipedia.org/wiki/EBUS_(serial_buses) ) and "1-wire" used for temperature probes etc ( en.wikipedia.org/wiki/1-Wire ). These are both bidirectional too. 1-wire uses baked-in unique addresses (like MAC addresses) and a clever mechanism to identify all devices on a bus without having to poll every address to do so.
I think the "bypass" resistors that can be the address. Electrically with 3 resistors you could have 8 addresses but the values might even ad addresses to them in the way it was done with the first wired resistor remote controls.
i never heard of these twinkly lights, so i had a look online. bloody hell, they cost 170 quid for 40, on 60 feet of wire. so each of those little PCBs cost about 4 quid each. seems a bit costly when one is buying 40 of them. mind you i get my e-waste from Poundland, so more than a tenner for lights seems lot to me.
24V LED strips usually use groups of six series LEDs and a dropper resistor. However, if you use a transistor configured as a constant current sink, you can use seven instead, increasing efficiency by 17%. (Obviously it's more complicated in this application and that would probably be impractical, not least because there may be considerable voltage drop over long cables.)
I use these LED strips in my kitchen and garden, powered from old laptop supplies which I have opened up and changed the potential divider so they output about 23V.
So the chip needs to quickly change the G and B pins between input and output modes? That would require a time synchronization method. Impressive, but seems like a lot of work just to avoid a third rail for communication. The factory defined addressing avoids the need of separate input and output pins.
Fascinated to know if these are using a microcontroller or some custom chip, I spent a little time searching for SOT-23-5 microcontrollers but they're really quite rare.
I know it'd need some gear that's not necesarily in every hobbyist's workshop but I wonder if anyone's decapped one of the chips and stuck it under a microscope, might be interesting to see...
Feeling the need to hunt down some 'Twinklys'
i got a set of addressable christmas lights that are literally a controller with bluetooth and wifi and a string of leds+wire+heat shrink that use 2 wires for power AND data, not 100% sure what type of chip they are using but they can show upto like 8 different colors at once before repeating the pattern
Twinkly festoon issues ... Sounds like a Python skit:
Rumpletweezer ran the Dinky Tinky Shop in the foot of the magic oak tree by the wobbly dum-dum bush in the shade of the magic glade down in Dingly Dell. There he sold contraceptives... 😮
Another informative video from Big Clive - How to put the twinkle back in your Twinklys! 😁
Would the IC on board be programmed as they produce them? rather than holding stock of each individual number? Also leads to the question "how many of these are in each display/set?"
Not sure how they're programmed or the maximum number available in a string.
Am I the ONLY one who sniggered when Clive said "No. 2" ???
Great video once again, thanks Clive! Would it be possible to hook up a scope to the rail/components to see where the data flows (instead of the water? 🤣)
Others have gone much deeper into the twinkly protocol.
@@BigClive very interresting how they manage the protocol.. but a bit outside the scope of you typical teardown/fixing things! Nevertheless, a great video. Keep up the good work, thank you Clive!
Oh! On the topic of OH-rings 🙂 - I see that suppliers normally only quote the inside diameter plus the cord thickness. Probably because the things are so squishy from the outside!
Twinkly Festoon
Certainly a very interesting system.
I think the 15v zener is to strip off a large part of the 24v supply so that a larger proportion of the data, that is present, can be seen across the 3k3. I am not absolutely sure about that, but the 15v zener isn't to stabilise the 9v. I think the supply to the chip is via the two 8.2k resistors?
Interesting. I have a string of these outside the back door and six of the twenty are in various failure modes (when set to all warm white, some of the failing ones show green, two have no output, one is sometimes blue but corrects itself after a while). Time to get de/soldering….
If they're still under warranty they may send new o-rings and modules.
Interesting stuff. 😊
There's a page called "Controlling Twinkly leds directly with a microcontroller" on AVR freaks where they've reverse engineered the protocol.
"It appears that the waveform is stable for 6 microseconds and drops to half voltage for 2 or 4 microseconds depending on whether it is a HIGH or LOW bit. When I turned off all the lights, I only visualized the wave with 6us at full voltage and 2us at half voltage, so this makes me think that the 2us corresponds to a bit 0."
Not sure how each module/LED gets its address.
Keith westley made two videos "Hacking Twinkly Part I" and "Hacking Twinkly Part II". The LEDs are QED3110 devices and it seems like each one has an address programmed.
Love a good festoon!
I would like to have seen one of these working Clive after the fixes.
I don't have a set or controller.
Very interesting. Never considered SMT resistors would be a point of failure. Thought they were completely sealed and invulnerable to moisture.
It's probably sneaky electrolytic corrosion creating a microscopic gap at a thinner area.
Where they're not always flat against the board if any water gets under them it'll be hard hard to get it out
Ooohhh, twinkly, Said every squirrel ever.
I would think that the omission or inclusion of the four resisters, (three across the LEDs and the one across the high value zener) could be detected for numbers 0 to 15. so this one is number 2 and sure enough one resister is missing giving 1101 or 0010. How are some of the other numbered chips wired? It does seem odd that the power pin would be used that way.... but maybe that is not the (only) power pin but rather a LED replacement circuit. even with none of those resisters there may be enough power from the four i/o pins to power the chip.
either these pre-date pixel LED's with the driver on chip and a data line or Twinkly wanted something that would remain entirely proprietary. 16+ SKU's instead of one seems a little OTT to me..
I've seen fairground cabachons which "know" their index, wonder if it's the same idea?
Thanks :)
Addressable LEDs are basically rgb 5050 led with IC.
I looked at buying a set of these, but I'm too tight to shell out for them!
Yeah, they're quite pricey.
strange, the IC is supplied with 9V.
The input voltage can perhaps drop to 12V and the IC continues to receive a supply of 3V
I was wondering how the chips were programmed too. I couldn’t see any pogo pin spots at first making me think they were programmed before placement, then I noticed what look like seven through plated holes. If they connect to the chip they could be the programming port.
Not really enough pins on the chip. Unless it was done on the output/data pins.
Not gonna lie I thought this said Twinkie Festoon at first and I was hoping for snack cakes.
Resistors on the right are for calibrating white balance or matching perceived light intensity for all colors - current is fixed in chip and redirecting part of it will reduce intensity of this channel.
I am no expert on this and I will be perfectly honest and say that this is the first I have heard of these but the communication method sounds very much like a CanBus style protocol with a time synchronisation reset…maybe
However, if it was something as simple as that then that would’ve been picked up already by people trying to reverse engineer it. Maybe some proprietary methods over the top of a well established high speed two wire method?
Nice video as always
Isn’t CAN (2 wire) using both lines for positive voltage? But yes the signal is always in a range between 1.5 and 3.5 volts. Both wires sum up to the median 2.5V. I guess only 1 wire could be enough if the zeners put the voltage to the lower level to eliminate pulsing. (Probably not even visible if not if its a fast data rate)
@@domnik9062 Yes, CAN bus is two wire. The signal value is determined by comparing the relative voltages of the two wires, CAN H and CAN L, which almost entirely takes out the impact of the incredibly electrically noisy environment that CAN bus was designed for. The concept is that voltage spikes will affect both simultaneously and therefore it's just a case of comparing whatever happens to come through. Voltage drops can be more problematic due to them affecting the circuitry more directly than EM interference in the wires or voltage spikes in the power source.
For something like this application I would expect a combination of data transmissions to one or more recipient nodes as well as an application broadcast transmission that triggers all nodes to apply the data they have just received. It's quite likely that there is also a combination of deltas as well as absolute values in the communications protocol, but that could be complicating things more when it comes to optimising bandwidth compared to control.
When will he show us how to build his grow ray?
Had you considered that the 8k2 pull-ups might be setting the address? Do all the boards just have those 2 fitted?
I thought that at first but they're all the same
FOUR ! ! ! Zener ( MELFS ? ) , Metal - Ended - Leadless - Flange ( actually : Face , but , I prefer Flange ? errmm ! ... ) .. this has made my day , but lotz of MELF's on a cheapo PCB ? ... has 2 B good ( ? ) .... OMG .... DAVE™🛑
Would have been nice to see one of the actual boards in operation. Perhaps even diagnosing and replacing defective resistors. Good information but boring for actual demonstration.
Thanks Clive
They require a very specific data driver and protocol. They will do nothing at all without that data.
So if they'd sprayed a thin layer of protective lacquer over the finished boards, all of them would probably still work. That's a shame.
interesting chip. but i can't get over that brand name. maybe their main competitor is just called Bulb.
I have one but uses IR instead of BLE
Twinkly has really let me down. As long as their products work, it's absolutely amazing, but the longevity is questionable. I have taken really well care of it, but the power supplies only last a season or three, my two strings had leds starting to fail at the end of the first season, and now by the thirds season they are no longer usable as strings, unless I pick an effect with "rainbow colors" since so many leds are the wrong color.
They do water resilient versions. Are you using the standard versions outdoor, and are there visible signs of corrosion?
A set of 20 Twinkly lights are a quite expensive at £130.
What bulb would you buy to light your rooms?
Most of the bulbs in my house are custom modified.
Clive, I wonder if the thing you have marked as the power input is actually the data input, and the chip gets its power from some or all of the LED drive lines. That would explain the lack of a filter cap on the "power" line.
Has anyone measured the data frequency on the power lines yet? I imagine it's up there a ways...
I used to like to shop for deeply discounted RGB light sets after christmas, as they all used to have decent 4 pin RGB LEDs in them.
But today they have mostly all went to these 2 pin RGB types, and arent nearly as useful as the 4 pin types as far as hacking goes.
the pcb looks like a raver on a trip lol...im ok man i sware
MELFs galore! Woohoo!
Do you have a video of these working
Kip Hakes has one.
That's just.... odd!
For some reason, I read "Twinky festoon"...
Would the 8201s stop ghost glow when those strings are off ? They have a place for the 3rd one but it is not populated.
Why would there be ghost glow? If "off" is called for, the controller won't be pulsing the power to the LEDs?
Clever these Chinese.
@@obd6HsN perhaps the driver outputs have some leakage. blue and GaN green are very sensitive and can glow with only 100nA
Strange they'd add a ton of zeners but there isn't a chip that has multiple regulated outputs(c) 🤷
Might the resistors on the R (unpopulated), G, and B lines be used to set the module's address?
At one point I thought they were, but different addresses had the same resistors.
iook up ws2811 pixel ic chip.
That's 3 wire.
can you put them on 240v for us to see how bright they can go? like Andy Photon used to do
Dead Circuits Society.
You can plug a new one in, and all it took was a bigger o ring? Can you say, we did this on purpose😂
You didn't show one running!
I can't show one running. Even giving it 24V it stays off because it needs data.
@@BigClive what kind of data