Exploring a faulty LED driver (with schematic)

[@NworbLegin]

"They don't always dome when they fail" sung to the tube of "You can't always get what you want" by The Rolling Stones

[@bumpedhishead636]

Clive: I'm a retired engineer (Aero & SW, not EE), I have always been fascinated by circuit design, but I have no idea how you do it. I have a general idea of what each component does, but not WHY a circuit needs to look a certain way or have the components it does. I envy you guys that know how to make the pixies dance down the wire without making smoke...

[@NaoPb]

Thanks for going through all that trouble to show us the board.

[@umbraelegios4130]

"Oh wow its thick and hard." Thats what she said. Maxim 17: The longer everything goes according to plan, the bigger the impending disaster. Maxim 41: "Do you have a backup ?" means "I can't fix it".

[@gregorywildie37]

The problem with cock-ups is there is no way to know in advance how big they will be

[@hugegamer5988]

The difference between maxim IC and maxim always gets a rise out of me.

[@youdontknowme5969]

Oh my! 😲🤣

[@alexhajnal107]

@@gregorywildie37 True in so many contexts. ;^)

[@_SurferGeek_]

While in the Navy, we often had to work on potted assemblies... using a heat gun would make it a bit easier to remove the potting.

[@SoftTinyBunny]

WOW (lightbulb over head)! I cannot tell you how much I appreciate you discussing the feedback windings on the transformer. This absolutely helps to know there can be more than the primary and secondary coil windings; this was never discussed in electrical engineering classes (30 years ago). Having this knowledge opens the door in my thinking process of how much broader the circuit can actually be at the very beginning of exploring the circuit. Thanks. Loving you for years and sending a cup of coffee your way.

[@AlaskanInsights]

Brute strength and awkwardness always works. I feel that 22k resistor is there for both, keep open circuit runaway voltage in check and a lil load on the primary so the feed back still runs.

[@d614gakadoug9]

You don't want the auxiliary winding to run with no load. You want the IC to shut down. You do want feedback just until the filter cap on that winding discharges to the shut-down voltage. Voltage feedback in LED drivers is just for over-voltage protection in any case. They normally regulate current.

[@davidbudge8359]

Overzealous Force & Elections would be a great name for a series on destruction of terrible products.

[@superdau]

To remove compound like this (the rubbery soft one) what usually work well is to a drops of isopropyl alcohol and then just agitate the compound. The alcohol creeps in between the compound and components and board so it separates quite easily (you can do the same to remove hot glue, which I think you know already).

[@fromgermany271]

The „feedback winding“ is also to power the chip. These 3x 750k resistors give a small supply until the chip runs and supplies itself via that „fb“. The resistors just give 0.1mA to the 4.7y 50V and they only waste 22mW. The 0.1mA would not be enough to drive a 600V MOSFET.

[@AndreaZeta]

Great disassembly and reverse engineering! 🎉now that we know the schematic, next time, connecting an oscilloscope to the output will give us a clear indication if the problem is with the output capacitor. Then connecting an external capacitor in parallel to the output would be most of the time a quick solution! 😊 I used this approach successfully for an intercom making noise at 50 HZ

[@justpaulo]

In parallel with the 22kohm resistor I would use an appropriate Zener diode with a small limiting current series resistance to protect the secondary side in case the LEDs fail. Without it I think you are right in being suspicious of the 214 uF capacitor.

[@ford1546]

I have experienced that many LED work lamps 10w 20w 30w 40w 50w 100w are quickly destroyed. If you use it on the same fuse line as a current-demanding tool, the lamp may receive noise and current drops, which it does not like. there is no noise filter or other protection circuits in the power supply!

[@thepagan5432]

Interesting to see where the flashing probably came from. Thank you Clive, take care 👍

[@Alexelectricalengineering]

The wire link between the diodes and the main capacitor should probably be for a inductor as a filtering component and a little bit of PFC.

[@dcallan812]

Interesting dive, is the potting compound also a heat sink? Its not the worst thing we have seen across the bench. Interesting video 2x👍

[@bigclivedotcom]

It probably does help with thermal dissipation.

[@merlin5476]

I thought the potting compound was to either prevent any kind of reverse engineering or to contain any sparks / flames in specific environments.

[@PF-gi9vv]

​@@merlin5476 Its also used for moisture protection, I always find these in hydroponics LED drivers.

[@Mark1024MAK]

The primary purpose is protection against the environment, that is water. The hard type compounds also have the advantage of making reverse engineering far more difficult. But that’s not really a consideration in these SMPSUs, as the technology is well known now.

[@sihamhamda47]

​@@Mark1024MAKIt's also transfers heat to the metal sink, helping to balance all the internal temperature

[@twocvbloke]

Reminds me of back when those 10w floods for I think £3 were flying off the china shelves all them years ago, I took one of the drivers apart in the ones I got and the potting compound was so sand-filled that it all fell apart easily and came off mostly cleanly, still have them out in the shed, though not used them in some years, probably been rat chewed now... :S

[@gregorywildie37]

Got 150 metres of cat 6 cable in one metre lengths after some bloody mouse got into the box

[@curtishoffmann6956]

"Why did it fail?" I dunno, but from the thumbnail, I'd say there was an unfortunate encounter with a lawnmower.

[@joshwalker5605]

I’ve had good luck using lighter fluid to help soften & persuade the potting compound- definitely gotta have good ventilation tho. Thanks for all the videos Clive!

[@Subgunman]

It appears repairable by replacing all of the caps as you said however I would use a silicone conformal spray to reseal the board and ditch the potting compound. These sprays are doped with a phosphorescent compound that when you expose it to a standard hippie black light the board glows blue. In doing so the brighter the blue the thicker the coating. You can see all of the voids you miss and can re coat as many times needed. If you need to repair the board again it is easy y to remove the coating where needed and replace the faulty component and then re coat. Bootstrap capacitor, I recently repaired an Alinco 30 amp switching supply for ham radio, it was a bugger to figure out eve with a schematic. Turned out to be a cracked resistor that basically was a bootstrap resistor that provided excitation to the rest of the switching supply. Good as new! This was of the first generation supply that allowed one to switch input voltage from 110 to 220 VAC. Their new model is dedicated as a single voltage and no schematic is available for it so one can see where to start a repair. Not thrilled with it but it works for now as well, knock on wood.

[@Zerbey]

Just fixed a TV power board with a bootstrap capacitor failure. As I’ve learned from watching your channel, capacitors going bad is a very common failure.

[@ronwoodward716]

I had an issue with my bathroom light. It would delay about 5 seconds before coming on. I depotted the driver circuit like you show in this video and found a loose surface mount diode. Resoldered it and the driver began working properly. Still working fine for the last 2 weeks. Wish this video was up a few weeks ago when I was looking for a solution to my problem.

[@endymallorn]

Thank you again for the detailed explanation. I love how even though you and your brother have been doing videos for years, there’s still a spirit of pre-corporate KZbin unplannedness occasionally. A question tho: aside from dodgy electronic devices hiding how bad they are, or protecting a patent design, why would you feel a need for a device or component to be potted?

[@aaronmdjones]

There are 2 additional reasons, one of which he gave in this video. It provides a pretty effective means of weatherproofing something for outdoor use, and since this was potted into a metallic enclosure, it would also allow the enclosure to act as a heatsink for every component on the PCB.

[@NightsReign]

The other primary purpose (afaik) for potting tends to be to prevent against coronal discharge from high voltages.

[@ricoma6037]

Respect for your patience with the putty!!!! I wonder if a toothbrush and some alcohol might clean that up even easier? 👍

[@gregorythomas333]

Don't forget to order more of the good power supplies so you have some on hand :)

[@grantrennie]

Thanks for another great video Clive, have a great day everybody 👍

[@Solipsfilm]

Would have loved to see you fix it and pour some potting compound over it again 😄 Nice video though, as always!

[@hugegamer5988]

The problem with potting compound is my circuits never seem to sprout.

[@Solipsfilm]

@@hugegamer5988 Best to plant them in early February and keep them at a temperature of 19°C at least, water sporadically to keep the potting compound moist. Mine usually start sprouting around week 3 so be patient and good luck!

[@MrXBT2000]

I wouldn't be too sure about the input capacitor - I recently serviced a DAB radio where an internal 10V to 3.3V step down had hiccups. I eventually found the cause to be the input capacitor being bad - and it was not obviously bad bad. The input capacitor is after all also subject to the switching frequency, yet at a lower average current due to the higher voltage.

[@Mark1024MAK]

Yes, but it’s the discharge where it’s experiencing the high frequency rather than the charging.

[@matthewmarks6951]

It is, but I don't see what difference that would make as the resistance applies to current flowing in either direction.

[@Mark1024MAK]

@@matthewmarks6951 I wasn’t disagreeing!

[@teardowndan5364]

Burden resistors on output windings are there because transformer winding coupling isn't 100%. Energy from stray inductance on each winding has to go somewhere or the voltage will keep creeping up until something gets destroyed when you don't have direct feedback from the secondary.

[@d614gakadoug9]

Nope. That's what the diode, resistor, capacitor snubber on the input winding is for. The energy stored in the leakage inductance is dumped into the snubber cap when the switch turns off then discharged by the resistor. Though it isn't necessarily implemented you can tell to a good approximation what the secondary voltage is from the voltage on the auxiliary winding or even the primary. Some flyback converters rely entirely on such methods for quite good regulation.

[@teardowndan5364]

@@d614gakadoug9 Every winding has its own parasitic/stray/non-coupled inductance. The primary winding's snubber network only protects the main switch from stray inductance spikes, it doesn't do anything for secondary windings' stray inductance. Every output of a flyback converter must have enough load to take care of stray inductance energy. With opto-coupler feedback, you don't have that problem since the input is suppressed whenever output voltage exceeds threshold and the LED circuit is often enough load to deal with stray inductance on its own in single-rail supplies.

[@d614gakadoug9]

@@teardowndan5364 You don't need a load resistor on the output of a modern flyback converter. I've designed several and it has never been the slightest concern. In years long gone by flyback converters often did require some minimum load for regulation because the duty cycle could not be made low enough. There is a discharge path for the secondary inductance through the output rectifier into the filter capacitor. There is no discharge path for the primary leakage inductance once the switch has turned off, so a snubber is necessary. If you are careful and know what you are doing and using a discrete FET you may be able to rely on the repetitive avalanche energy rating of the body diode of the FET to discharge the leakage inductance. I've done that in housekeeping supplies for much higher power special purpose switchers. You do understand that the LED driver is a flyback converter and the thing being called a transformer isn't a transformer at all in the conventional sense but an inductor with separate charge and discharge windings, don't you? The concern with no load is that if you put ANY energy into the input winding it WILL go into the output circuit and the voltage will rise til something breaks, regardless of the "turns ratio" of the inductor, if there is no feedback.That has nothing to do with leakage inductance. It is the fundamental nature of the beast.

[@theminer49erz]

I have noticed that many incandescent alternatives even back to the CFL days, have much longer life spans on paper than you tend to see in use. Ballasts and LED drivers seem to fail before the actual light/Diode do. I have seen SO MANY street/parking lot lights that were replaced with LEDs have like 10-20% fail within a year. OR they flash like a strobe. Many street lights in my area are getting replaced with purple/black light lamps. I'm not sure if that is because they already stopped making fixtures for the units they installed or they are just cheaper/less likley to fail. Still, I have had incandescent lamps last much longer than any modern replacements. They use more power, but so does making a bunch of CFLs and/or LED lamps. Of course, buying cheap ones will make that more likely, but what isn't cheap anymore? I bet if the AC to DC conversation was done with a dedicated unit for a home or say the base of a streetlight etc, they would last longer. Having to build them with all of that inside the lamp and keep it cheap enough that people will buy them is prob adding to the issue. I'm actually retrofitting my house to have mostly 5V LED lighting that runs off a battery charged by 12v solar panels and topped off by the grid if nessasary. Using motion/photo sensors to make many come on and turn off automatically. Will also run 12v from the batteries to receptacles with USB to power/charge USB devices with it too. Was also thinking about building a custom hot water "pre heater" tank using some bigger PV Modules I have. I have a small new HWH tank without elements. Was going to hook it up before my gas HWH and use whatever power the sun can provide during the day to just heat that water with a DC water heater element. Im thinking about using one made for livestock water that can be hooked to batteries/PV modules in feilds without the need for an inverter. I have been collecting average usage data on the gas the hot water heater uses, so when I do this, I can see how much it helps. I'm going to super insulate the tank, so theoretically even if it can't make a tank of cold water hot, hot by the end of the day, it will certainly warm it up enough to require a lot less energy/fuel to bump it up to temp. Considering the modules and tank were free and I will only need to pay for the heating element, it should save a lot of money over time as well as lower my overall fossil fuel consumption. I'm also looking into using the insane number of apple trees and biomass from cover crops to distill ethanol so I can run a generators off of it if nessasary. We get a lot of brown/black outs here at the farm, so having battery back ups for basics and renewable fuels for more power hungry applications should be very helpful. I would like to start a local bio fuel "co-op" one day where everyone can throw down raw materials throughout the years and some money at first to build a LARGE stil and fuel storage systems. If anything happened, we would all have plenty of fuel. It's so unfortunate that EVs are almost ensuring we just stick with ancient battery tech now. We really need to come up with better ways of storing power. People like musk are not going to do it. Just for their BS "Gigaplant" plan or whatever it's called, alone they would need to extract somthing like 10x the amount of lithium we have mined in the past decade, every year. Not to mention the cobalt and other rare components. It's not viable imo at least not the way musk and Co want to do it. What else is new though right? Plug in hybrids would prob do a better job especially during the transition. We would need less battery per vehicle and even though they still use petrochemicals, they will use much less. 100% EVs will be getting charged via burnt coal most of the time, so why not? It would reduce demand for gas, lowering the cost and buy us time to innovate without having to dedicate ourselves to using 60+ year old battery tech just so we can make the transition sooner. What's the point of rushing that, if we end up where we could have been in the 70s?! OK I'm sorry I digressed like a mofo and moved onto full blown rant. Failing LEDs apparently triggers greenwashing/half assed pandering rage in me I guess. Great video as per usual! Peace and long life!🖖🏽

[@zacharytaylor8523]

Tends to be the curse of solid-state electronics. Everything truly does have the ability to last a long time but everything is also an assembly of tens if not hundreds of individual components where anyone can cause a problem if it fails early.

[@Broken_Yugo]

​@@zacharytaylor8523 Curse of solid state lighting is more that it can't handle a ton of heat, and it's cheaper to run less parts hotter so we get a lot of over stressed crap on the market at every price point that's only going to make the claimed lifespan with forced air cooling. Back when a basic 800 lumen 80 CRI LED bulb worth buying started around $5 they really did run until the diodes got noticeably dim 10,000+ hours later and you decided to swap it out for a nicer one, then in my experience swap half of them back when the nicer ones blow 50% in 6 months.

[@Slawek_EU07]

In my country, Poland, there are lamps (floodlights) with such power supplies. Most often, poor quality COB LEDs burn out. Some lower wattage power supplies do not have galvanic isolation and the headlight housings do not have a PE wire attached, using this can kill.

[@notahotshot]

I'm a little surprised you don't build your own power supply from discreet components.

[@bigclivedotcom]

Much more expensive than using a standard one.

[@notahotshot]

​@@bigclivedotcomyeah, I was just being a bit of a git.

[@curtw8827]

Easy to critique someone else's design, harder to start with a blank sheet of paper

[@petersage5157]

@@bigclivedotcom Sure, but you could engineer it to outlive the LEDs you're driving. Heck, at our age, you could build it to outlive us.

[@timpontius]

Perhaps the resister across the output is to prevent a slow fade of light when power is removed?

[@Torbjorn.Lindgren]

Sounds likely, it's probably for that and to avoid "ghost glow" from induced currents when turned off.

[@Alex-mj7km]

@@Torbjorn.Lindgren I don't think switch-mode PSUs from "ghost glow". They usually blink, but I've no idea how resistor across the secondary would stop the blinking.

[@mysock351C]

@@Torbjorn.LindgrenIt’s probably needed as the LEDs eventually stop conducting below a certain voltage so the cap will just hover there, and I can say from experience that they will indeed ghost for quite some time due to the gradual leakage of current when they are at the threshold of conduction if there is no discharge resistor. Having a resistor there ensures that the filter cap can discharge past that point so the lights extinguish.

[@ralphj4012]

Many SMPSU designs do specify a minimum load. I have seen a few where if the minimum load is not present then the output pulses low / high (blinks, for want of a better word). Why this occurs is something to do with pulse stability in the feedback circuit (too technical for me).

[@mysock351C]

@@ralphj4012 In SMPS supplies whichever winding is in conduction determines what the output voltage is, and if there is no load with this type of supply (flyback), the induced voltage will just continue to climb until something starts conducting since its behaving basically just like an inductor. To stop this, they have burst or pulse mode which means the supply will just turn on briefly until the feedback satisfies and then it goes back into standby. Having designed a few of these for my projects, usually having something like 10-20mA of current from secondary things like indicator LEDs and such is enough to at least keep the IC on usually in FM mode or something similar since the burst mode can cause interference in audio equipment. Here, though its 22k so its only around 1-2mA. That might be just enough to allow the capacitor to stay at a reasonable voltage while the feedback is in the process of being satisfied. But, with LEDs you _do_ need a discharge resistor or they will glow very faintly for a long period of time once the power is removed.

[@kmoecub]

The first two minutes of this video are exactly why I keep a portable, precision fire nearby.

[@d614gakadoug9]

The capacitor after the bridge rectifier actually is subjected to substantial high-frequency current, though entirely as discharge current. For much of the AC cycle the bridge rectifier won't be conducting so the cap is the only thing that supplies the input requirements of the switcher. Fortunately, flyback converters are usually relatively kind to the input cap. Usually, but not always, a flyback is run in "discontinuous current mode." That really is a bit misleading because there is never a case with a flyback when there is continuous current (i.e. not dropping fully to zero) in either the input or output circuit. Both currents do go fully to zero, however the energy stored in the inductor may not. Assuming discontinuous current the current from the cap and in the inductor ("transformer") primary begins at zero and linearly ramps up, the abruptly goes to zero when the switch turns off. If the inductor didn't fully discharge every cycle the current would rise "instantly" to some level then ramp up from that ("ramp on pedestal" is a term sometimes used). Anyway, long story short (too late!) there is high frequency current flow from the input cap. The ripple current in the output cap can be absolutely brutal in a flyback converter - far worse than in other isolated topologies. With each cycle it always rises "instantly" from zero then ramps downward, The LED string across the filter cap acts as a shunt voltage regulator, though not quite as good as zener might be. It makes the output cap less useful than one might think because the cap can't charge up to a voltage higher than the instantaneous voltage of the LED string.

[@johnthefactfddict3281]

omitted NPT thermistor clive, surprised you didn't guess that inrush to the capacitor which is SPICY

[@Mark1024MAK]

One of the biggest problems with these types of SMPSU is the secondary (output side) rectification diode being placed near the secondary side electrolytic ‘smoothing’ capacitor. A diode that runs hot heating up an electrolytic capacitor can seriously reduce its lifespan. In a good quality SMPSU that uses a conventional Schottky diode, this would be mounted on an aluminium heatsink. [Edited to correct the spelling of Schottky].

[@NightsReign]

As a preamble, I'm not _trying_ to be a pendant, for all I know it was autocorrupt that turned *Schottky* into *Shockley.* Just thought I'd throw that out there.

[@Mark1024MAK]

@@NightsReign - Ahh, thank you for letting me know that I messed up the spelling of Schottky. The perils of using a touch screen (and no doubt getting the spelling wrong) with autocorrect “correcting it” and me not checking what “it” (me) had actually written…!

[@randycarter2001]

I call it the BF (brute force) method of disassembly. Very handy when it doesn't have to go back together and function. Sometimes a hammer comes in useful. Definition of hammer: It's a universal fix it tool, it will fix anything so it either it works or it doesn't work.

[@icexvi2573]

Today I have disassembled such driver for 220V to 12V 15W that recently started to go techno/rave operation mode from time to time. The failed part was also the capacitor at the low voltage side :)

[@nfx7414]

Output resistor might be to drain the cap if the leds fail, or to make sure cheap Chinese bulbs turn all the way off

[@youdontknowme5969]

OMG so my favorite hole-in-the-wall artery-clogging burger joint has an LED fixture right over their drive-thru pickup window that strobes erratically, it's been like that for over a freakin' year now, I mention it every time and they're always just like, "yeah yeah, we know..." just unplug it or take it down or whack it with a baseball bat if you can't fix it LOL! Luckily that flashing doesn't "affect" me like it might some people

[@bigclivedotcom]

Unfortunately the light is probably a random size and you can't just change a bulb. A new one has to be hacked in by an electrician, so it costs a lot to fix it.

[@danielegger6460]

Brilliant idea trying to discharge a cap with an uninsolutated screwdriver. Not convinced that would have lead to a lesser zap if it was charged than touching with bare skin...

[@TopEndSpoonie]

Thanks for part 2.

[@rpdom]

I would like to see it fixed if possible.

[@bigclivedotcom]

That is possible.

[@manolisgledsodakis873]

@@bigclivedotcom Yes, it would be most instructive to replace each capacitor in turn, to determine which failed. (Not 100% certain, of course, since all three may be contributing to the fault.) Something to try on a rainy day? Oh, wait. I'm in the Mediterranean where rainy days tend to occur mostly in Winter. OK, something for you to try on a dry day!

[@phonotical]

Sometimes with old transformers, if the load was disconnected, in this case if an led fails, the transformer would overload and burn itself up, so it could be to prevent fire/overheating and damage in the event of failure The other alternative is maybe it's a sort of impedance matching

[@acmefixer1]

I wonder if Isopropyl alcohol would help delaminate some of that potting compound. It might be interesting to try.

[@johnkey1682]

I've seen white spirit used to "swell" silicone in the past, it makes it quite crumbly.

[@PCM-3199]

It is noticeable though that the bench light on the right is slightly dimmer than on the left one..

[@MaNNeRz91]

Be good to see a breakdown of the common KT ebike controller. The one i have claims 2000w from a 48v battery. I had problems with one of the mosfets blowing on one of the phase wire circuts. I ended up replacing the FET with the same one. And it blew again. I damage one of the pins when repairing it a second time (with the orginal blown fet which was actually fine 🤯) so i ran a short piece of wire to bridge the drain to the FET next to it in the most bodged up way and its been perfectly fine since 😆😆😆 Gets extremely hot with regenerative braking to the point the phase wire started to melt. The same phase which ive had the mosfet issues. Im no electrical expert but im ok at repairing up to a certain type of pcb

[@amorphuc]

Nice. Thanks Big Clive.

[@_BangDroid_]

I'm 99% certain this is the same issue with my bench light, not sure I have the patience to de-pot the driver though. But it does seem difficult sourcing drivers from Ali right now.

[@Evergreen64]

I recently found out how bad I am at diagnosing failed electronics. I had a power supply that powered a small Roland mixer 90's vintage. It was supposed to output 9V @ 1A. One day (years ago) it failed. No output at all. So I decided to see what I could do. I figured it was a capacitor (Defpom: "It's always a capacitor."). So I replaced all the electrolytic caps. Still nothing. I measured the resistance across the output and it was shorted. So I looked at the output and found a shorted Zener diode. I replaced that and the symptom went from no output to on. Off. On. Off. On. etc. So at this point I'm stumped and now since I don't have a parts stock I've spent about 4 times what the power supply is actually worth. So I gave up.

[@FrontSideBus]

I really despise these things tbh. I've lost count at how many I have seen failed in "disco mode" or just barely glowing...

[@spokehedz]

I see everyone giving their takes on how to remove potting compound: compressed air! And not the stuff from the cans, but from an actual compressor and tank. And higher pressure in small bursts is better than long durations of air, and it wil blow it off in huge chunks, and it is sooooo satisfying. Also, freezing it solid, and giving them a solid WHACK on something hard can sometimes shatter the stuff into zillions of bits, also being able to be blown out via air. Also, if you do freeze it, and shatter it, the little bits might be really sticky when it warms back up to room temp. Dunno why. Weird Chemicals?

[@FireCrowsWorkshop]

If this potting compound is anything like the stuff used for sealing automotive control modules then brake cleaner should be able to help soften and remove it!

[@wiseoldfool]

@@Kevin-mp5of It's also not very nice (I hope that sounded like Clive!).

[@Sylvan_dB]

I want to like LED bulbs, but I'm replacing them far more often than expected, making the e-waste seem like a significant problem even if the cost were as cheap as the $0.25 incandescents I used before the government banned them. With dozens of bulbs I experience frequent LED bulb failures (usually one or two every 6-8 weeks) evidenced by flicker which is caused by the capacitor failure. This is only happening with "newer" LED bulbs. I was an early adopter of LED bulbs and my oldest bulbs are still going strong in spite of being on dusk to dawn every day for years (for a few years controlled by an X10 switch, in violation of the "not dimmable" specification). Those bulbs are "40 watt equivalent" and have a large aluminum finned section between the base and a hemispherical translucent dome. They were unknown brands at local retailers sold at $10 to $20 apiece over a year or two purchase window, and have spent their life in outdoor enclosed fixtures. I have a few spares removed from interior use in favor of bulbs with improved appearance, but they've been waiting for many years now. All the later bulbs, all more sleek and refined than the original chunky aluminum versions, have been failing in every type of fixture while used in normal household use. My first failures some years ago were the nice Cree branded bulbs with the rubberized glass bulb and the small finned heat sink section. I'm on my 4th LED bulb in that simple ceramic bare bulb fixture, which isn't any better than incandescent (if even that good). Yesterday I replaced my last two of the "Slim Style" (almost a 2 dimensional bulb, but the thin dimension is about 19mm) Philips bulbs that were originally advertised as safe for use in enclosed fixtures, but the ads changed (and they never were enclosed in my usage). I installed Philips Ultra-Definition bulbs at $7/ea. I wrote the date on the base as I've been doing for over 20 years. I'll see... This is the same pattern I experienced with CFLs. I had some original CFLs still working until after I was having LED bulb failures. Newer CFLs failed faster, just like newer LED bulbs are failing faster. At least LEDs don't smoke as often when they fail.

[@andygozzo72]

maybe the boostrap or output cap intermittently shorts or goes low resistance when under voltage, i've had that on a couple of occasions, one with the audio boostrap cap in a roberts r505 radio, tested good on its own with a capacitor tester but when in the set, powered up, would go almost short after a few seconds

[@zedcarr6128]

13:55 I'd say that the 22k across the output serves two purposes. It provides a load on the output, so the power supply can start up, and it discharges the capacitor when the power is off. Congratulations on 1 Million Subscribers, Clive. I've been subbed to you for many years. 👍🏆

[@svenhoff2653]

I hate it when they use the potting compound. As far as my experiance goes, the Capacitor goes bad after some time. And replacing it with a decent one fixes the problem. After changing the cap my drivers work without problems (and did not fail again so far).

[@johnthefactfddict3281]

I love how it sounds like clive says "nanofart" when he says "nanofarad" too fast

[@channelsixtysix066]

The that was .... delightful, Clive.

[@sidetrack218]

This may be a stupid question but what is the reason for potting certain electrical circuit boards, seems it would keep components from cooling efficiently.

[@mississaugadave7270]

Potting of components and PCBs mainly serve as a sealant against moisture and as an electrical insulator. The practice has also been a method to prevent reverse-engineering through the use of potting compounds that become very tough, chemical-resistant blocks. Such indestructible concrete-like potting will also provide heat-dissipating properties.

[@sidetrack218]

@@mississaugadave7270 Appreciate your reply. The moisture and electrical insulation parts makes sense. I didn't think of the potting compound actually being a possible heat sink and it never crossed my mind it might also be a way to hide circuitry as a deterrent to those wanting to copy. Thanks

[@d614gakadoug9]

@@sidetrack218 Potting compounds can be "filled" with thermally conductive material, typically oxide of some metal. Zinc oxide is not bad but has fallen into disfavor as a somewhat hazardous waste. Aluminum oxide is good. Boron nitride is quite spectacular but with a spectacular pricetag.

[@ralphj4012]

Great explanation. I've often seen the feedback capacitor cause issues, especially with supplies operating for many years. I wonder whether a low value ceramic capacitor across the feedback and output electrolytics would reduce the likelihood of failure.

[@Mark1024MAK]

A ceramic capacitor (say 100nF) in parallel across both the bootstrap/feedback electrolytic capacitor and the secondary/output electrolytic capacitor would help, yes. But that adds to the cost. It’s only a small amount, but…

[@mk_scorpion_]

Congrats on 1 million 😎😎😎

[@SiaVids]

I once had a power supply that would start randomly, I eventually found that the problem was an intermittent bootstrap resistor.

[@mrtcntr]

very likely, would like to see what the resistance of the 750k has gone up to

[@Mark1024MAK]

Yep. Also failure to start may be due to an open circuit resistor (that has failed) or a dry solder joint.

[@Mark1024MAK]

So you fixed the flood light but not the original SMPSU!!! 😳

[@TechOne7671]

What do you think about the led light revolution Clive? Aye, they save energy but man they are just junk for reliability. I full a skip worth of them in a year easy. The led panels inside seem to be a more more reliable these days but the power supply always fails, whether it’s a cheap fitting or big brand one. The wholesalers can’t usually source the correct driver so the fitting is scrap, just great for the environment. Give me the sodium’s and fluorescents any day, new ballast or igniter and it’s good as new. I have 30 year old HPS fittings that are still running. Next time your in Edinburgh give me a shout and you can fill your boots with them. 😂😂😂Cheers.

[@rayoflight62]

There is a specific solvent for removing the resin, a liquid sold as "conformal coating stripper" on industrial supply catalogs. But it is a solvent in the class of carbon tetrachloride, isn't healthy and must be applied under a fume hood. The Chinese electrolytic capacitors can't stand more than 6 -12 months when operated at ultrasonic frequencies, typical of a switching power supply (~ 110 KHz). If you replace the capacitor with an original Nichicon brand, it is going to last for a lifetime.

[@lawrencegt2229]

Intriguing, I have a four LED bathroom light and two of the LEDs have started flickering on one side - maybe there is a starter for each side h one has failed?

[@dimitar4y]

here, we see a wild bear, in its natural habitat, mauling its next victim, tearing it apart mercilessly, to get at its guts- oh it's just clive.

[@petersage5157]

Any brand markings on those electrolytics? The colors of the sheathing all match appropriate types from major manufacturers (green Panasonic mains rectification, brownish Nippon-Chemicon low-ESR). Probably all generic knockoffs though?

[@bigclivedotcom]

Just vague two letter brands like BH and JP.

[@paulstaf]

I am surprised you didn't fix it. :D

[@bigclivedotcom]

It would be better to do the fix on a driver that was completely faulty instead of just intermittently. That way the fix could be seen. I could repopulate the PCB with new capacitors and keep it as a spare though.

[@Mark5mith]

Ive been changing quite a few failed led drivers, always seems to be a failed capacitor and theyre all domed with heat damage to the plastic case.

[@wtmayhew]

Aluminum electrolytic capacitors are hygroscopic and absorb humidity from the atmosphere. That is why electrolytic caps often have slotted spacers to keep them slightly away from the printed circuit board. The spacer also keeps the caps away from heat which may be in the board that would tend to dry out the cap. I’m not sure about a completely potted capacitor because it should neither absorb humidity nor outgas. Heat is a high on my suspect list. Vibration is also a likely suspect, but the shop light should not have gotten much mechanical shock. The third suspect is that it is simply an average Chinese capacitor of dubious quality. Way down on the list is shorts in the transformer, but hey you never know…

[@d614gakadoug9]

The "slots"in the sealing plug are there so solvents used for board cleaning can escape."Snap in" caps can be a problem because they seat tight against a PCB. I've resorted to adding drain holes in the PCB under such capacitors in designs where I knew the boards were going to be cleaned. Chlorinated solvents can penetrate ordinary seals and "poison" an aluminum capacitor. There are "solvent proof" seals that have pretty good resistance and permit the use of chlorinated solvents with care. Some spray-can board cleaners are chlorinated. Heat generated internally in an aluminum electrolytic capacitor due to ripple current acting on the ESR is far harder on the cap than external heat that raises the cap to the same temperature. I had one sample board come back from the user with a bunch of components, including the main input filter cap, broken off. They'd run the board on a shaker table. No one had bothered to tell me survival on a shaker table was a design requirement.

[@cambridgemart2075]

I noticed the IP67 rating on the case in the previous video; I was dubious, but seeing the potting makes me think it could well be true.

[@chatrkat]

As I watch you reverse engineer that short lived power supply, I think about the fluorescent fixtures I am sitting under, with magnetic ballasts over 50 years old still functioning normally.

[@bigclivedotcom]

Yeah. Same for the new led replacements for street lights. The whole expensive head needs replaced every few years where it used to be just a bulb replacement.

[@carlospenalver8721]

If you pointed at a solid titanium cube 2 meters x 2 meters x 10 meters stainless steel plated with a wire coming out opposite ends and tell BigClive there’s a germaninium diode inside he’ll still show up with tiny wire snips and a precision flat head screwdriver to get at it .

[@MalinDixon]

On LED power supplies where the input capacitor has degraded, I have added an external rectifier and input capacitor to get the power supply working, without having to open it up. I don't know if it would also be possible to put a larger capacitor in parallel with the output, in case the output capacitor isn't as good as it used to be

[@bigclivedotcom]

Adding an external output capacitor can extend the life of older power supplies.

[@christophertstone]

Would love to see you replace one cap at a time, see exactly what the failure was.

[@bigclivedotcom]

Unfortunately it was very intermittent and any cao change would probably solve it for a while even if it wasn't the primary culprit.

[@alexandern8hgeg5e9]

2:04 If you clamp it into a vise, you can position some demolition screw driver / chisel and then punch it with a hammer. Then It's also easy to use a hacksaw.

[@TheDefpom]

Yeah it's always a crapacitor....

[@richardbriansmith8562]

Awesome Video Big Clive

[@Roy_Tellason]

That potting compound is nothing at all like the stuff I encountered when trying to see what could be recycled from a Commodore 64 power supply. That stuff was as hard as a rock! I tried a couple of those, using different methods to get the stuff out of there, and was not very successful. I'm thinking that it must have been some kind of epoxy because when it got to the point of trying to remove it from around epoxy-cased components like diodes and a regulator chip, the cases of those parts ame out as well. I gave up on those after that...

[@bigclivedotcom]

It usually takes heat and lots of effort to remove the black epoxy.

[@McTroyd]

Perhaps a follow-up video opportunity? "Hack" the unit by doubling up the output capacitor to prevent this failure mode, in the same way you often "hack" LED lights by snipping a current sense resistor off the PCB. Even using cheap capacitors would probably work, because the ESR would only be half the value when there's two, right? 👍

[@bigclivedotcom]

Doubling the output capacitor capacity can cause the unit to fail to start, as the bootstrap circuit fails to maintain the initial burst of power long enough to charge the output capacitor and that is reflected on the bootstrap capacitor.

[@McTroyd]

@@bigclivedotcom Interesting. Did not know that, but seems logical in hindsight. Maybe if the capacitance was kept the same, just using two capacitors...? 🤷‍♂

[@kira07]

if yo uwant to remove even to toughest epoxy, there is a checmical you can use to do so, you just dunk the unit into it and leave for few hours and potting turns into mush that you can easily take off. I wont say the name of it as its not safe, as the vapour it releases is very bad, so this must be done outside or with proper ventilation. Als if yo uwant to dispose of it I suggest to let it evaporate naturally. Also this is only sold to busienesses in UK , as 'common folk' cannot buy it as its very close to being illegal due its health hazards

[@darrinswanson]

I would like a nice, long, boring, and repetitive soldering project video, Clive.

[@tonysheerness2427]

Is it done for waterproofing or to stop people repairing them? How does it disperse heat? It is not as if it is going to be used under water.

[@bigclivedotcom]

It's for waterproofing. The material can also help transfer heat to the case.

[@jerrydurand4127]

Is the resister to make it go out quicker? I bought a cheap (300 rubles) ceiling light for a storage space and it takes quite a while to go out. Otherwise works fine.

[@bigclivedotcom]

It probably is to make it fade out faster.

[@d614gakadoug9]

Slow turn-off is primarily due to the input capacitor. The output cap simply can't charge to any voltage higher than the LED string forward voltage at any instant and the LEDs go off completely as soon as the cap voltage drops below the low-current forward voltage of the LED string.

[@Gav_Rez]

"Violence" You make me laugh always, Keep up the great vid's

[@shaunewing8973]

love that clay type potting lol .

[@jasonkuehl639]

I could watch Big Clive chip away at potting compound all day! But I still think he needs to revisit lighting with cheese since cheese technology is keeping pace with lighting technology. 😆

[@artillerest43rdva7]

hello Big Clive so sorry I know very little about circuits and electronics, but I enjoy your videos teaching about things that are around us in our daily lives. ok what exactly is a “moss fit “ { spelling way off } that integrated circuit. I am in the US do you happen to know a good starter book to learn about electronics? I have been always fascinated about electronics and how they are made and work. thanks so much, thank you for all your great videos, have a great day.

[@gordonirving24]

Metal oxide field effect transistor.

[@bigclivedotcom]

A MOSFET is a type of transistor that can switch high current, but is very easy to drive.

[@gordonirving24]

@@bigclivedotcom Hi Clive, change of topic. I've tried for ages to find out about the American street light that is acorn shaped. It appears in many films and seems to be a standard luminare shape. Wonder if your US followers could advise. Kind regards from Falkirk.

[@fredfred2363]

Can't beat a linear power supply for very long life... I can't figure out why we changed to switched mode, other than to get rid of the metal transformer.

[@Alex-mj7km]

Semiconductors became cheaper than copper/aluminium and iron.

[@johngangemi1361]

Linear regulators have poor energy efficiency.

[@PF-gi9vv]

I made a linear power supply for my 50W LED once and it was a whopping 65% efficient! 😐

[@Alex-mj7km]

@@Kevin-mp5of power-grid transformers (and in general high quality transformers) are really efficient, often 95+ % efficient. Problem is that they are quite heavy and big and expensive. SMPS can get reasonably high efficiency, and transformer itself is super cheap and small and lightweight, and weight matters a lot when stuff is made in China and transported via airplane. But all this comes at cost of being less reliable due to increased complexity.

[@d614gakadoug9]

@@Kevin-mp5of Low frequency transformers are generally significantly more efficient than high frequency transformers. Linear supplies are inefficient because they regulate by converting electrical energy into heat.

[@Lmay_3809]

Violence is always the way in electronics!

[@robinc.5077]

That wire link is where they would put an inrush ntc

[@Mike_5]

There must be quite a large reverse EMF current with that setup no wonder it went into Boogey Boogey and put it's dancing shoes on

[@d614gakadoug9]

What large "reverse EMF current?" Where? How?

[@Mike_5]

@@d614gakadoug9 On the supply side Phase to Neutral

[@PsiQ]

Not long ago i was shopping for usb power supplies and because of a certain channels info and because they also are the cheapest i ended up ordering from IKEA :-) I noticed they also have new LEDs, that even with the tightened efficiency ratings are in the good B and C group again. Also claim longer life/better tech on the specs, so i ordered a few along and was not disappointed.. Did you happen to analyze those already?

[@bigclivedotcom]

I've not seen the new LED lights yet.

[@MattThaDJ]

I wonder if anyone of you guys has thought of putting their multimeter or any electronic diagnostic tools in a Faraday cage in case of solar flare/EMP? 🤔🧐🔍

[@elgorrion52]

What about discarded vape tubes for potting electronics 💡

[@TDOBrandano]

“Maxim 6: If violence wasn't your last resort, you failed to resort to enough of it. -The Seventy Maxims of Maximally Effective Mercenaries” ― Howard Tayler

[@DigitalIP]

Question for ya Clive. Have you ever seen or heard of a dual voltage COB LED (3/3.7v on one side and 5v on the other) that pulls 4amps 40+watts on the 3v side when connected to 5v USB? I have an ATX PSU Adapter Board that has 6 5v USB Ports along with all the other terminals for 3.3v and 12v etc, now ive had this light since 2020 but barely used it, but i just hooked the COB up to the ATX PSU and the Board to my killawatt meter cuz i knew the light drew at least 2amps or more which most power banks obviously dont like, but it was Pulling 4amps over 40watts!! How the hell could that happen just by hooking up a 5v USB input to the 3v side? If you're curious about the light its listed on Aliexpress as 120x10mm Battery USB Powered COB LED Strip Light 3V 3.7V 5V LED Bulb 5W for DIY (Gotta love their titles), on the 5v side its actually a very nice light for the record.

[@DigitalIP]

Nevermind, ended up contacting the seller in case they knew. Turns out the 3v side doesnt have any current regulation so it'll basically go as high as possible. Which makes sense. I had forgotten that the 1 side was for 3v so when i first ran into that when i bought it back then i hadnt considered that.