150W LED power supply teardown

What's inside a dimmable 150W Philips LED power supply and how does it work. Including a reverse engineered schematic of the power section, some tests, measurements and oscilloscope waveforms.
Please support my channel on Patreon:
/ diodegonewild
Instagram:
/ savage_danyk

[@drobotk]

"This thing is so complex that there is almost an unlimited amount of things to *explore* " I was sure you were going to say *explode* there LOL

[@Conservator.]

9:19 Wow, the schematic together with the device and a pointer to show the components! 👌

[@piconano]

He good.

[@Conservator.]

I learn far more from watching DGW videos than I would from taking one of these LED drivers apart. 😁 It seems like a (probably) well engineered driver quite different from most Chinese ones.

[@DrHouse-zs9eb]

Your camera setup became really good. These closeups are very good and readable.

[@airmann90]

Great explanation of an interesting design. I love how complex it is to just light something up lol.

[@iamdarkyoshi]

Your reverse engineering skills are top class.

[@amartyavlad]

you know its gonna be a good day when diode gw drops

[@carlosaugustorj]

arrow pointing the components was ❤❤❤ very good

[@d.k.9406]

Danke! Thanks ;-) The oscilloscope helped a lot for visualisation.

[@DiodeGoneWild]

Thanks for your support ;)

[@requiem4adreamc]

What is the model of ossiloscope? ​@@DiodeGoneWild

[@tiagoferreira086]

@@requiem4adreamc FNIRSI DPOX180H, i have one and it has annoying bugs mostly on trigger, and the voltage levels calibration are far from good... buy at your own disappointment risk... Of course it has also some good things like the screen but watch some reviews first at least, but don't believe in most of them saying that everything is great and awesome because they are paid or get the instrument for free to advertise sales.

[@jirij]

That complexity really smells of an engineer given no budget constraints. The synchronous converter, the discrete gate driver(s) adding extra BOM cost, the mosfet for shunting the current sense resistors, etc. ... If you already have a MCU for controlling the mosfet, why not use it to control the optocoupler (or voltage divided FB pin) directly? .. In fact - that's exactly what most sensible constant-current LED PSUs do. :) Specifically most Meanwell PSUs are a nice example of cost-optimized design (I opened a few of them), but without skimping on quality or filtering. Chips like AP4310 are super useful (and cheap) for both voltage and current limiting, all in one package. Anyway, greetings from Brno!

[@adnanmlivo5885]

Biggest problem with this led driver is thermals other than people. Every 8 out of 10 that come in for service are heavily discoloured boards around output transistors and every single capacitor cocked and leaking after 3 years of use. Other 2 are miss wired and dali + and - used for live and neutral. In that case replace a blown up trace and fuse and hope.

[@dosgos]

haha. Junk design.

[@stefflus08]

How much does the unit cost? I can't imagine this being something that touches my bench with our cost of living. PS. I rewind and looked up a Philips Xitanium. At €20 me opening and looking at it, if only to do a quick fix, would break-even with a new unit.

[@adnanmlivo5885]

@@stefflus08 Not sure about the price but here in Bosnia it makes sense to fix them for 15 BAM rather then to rewire and retrofit other makes or models but recently i experimented with Lisun and so far extremely happy with the results. Plus for 10 Bam more you can use flicker free driver and adjust the current down to increase lumen/watt. Depends what costumer wants in the end but most of them would rather pay 80 BAM and upgrade rather than buy 450 BAM light fixture.

[@tajtrlik1111]

Ďakujem ti za toto video, pekne si vysvetlil princíp fungovania tohto zariadenia.

[@mranvick3512]

A 48MHz Atmel Cortex-M0+ for a power supply... bloody hell... reminds me of Deus Ex Silicium (FR) video presenting a Dyson vacuum cleaner with two µC and some RAM chip...

[@tiagoferreira086]

That's what i though, why such a "powerful" micro in a "simple" led driver? I guess it handles everything including the inductor switching, and it may be cheaper and easier to implement that a bunch of different interconnected controllers and auxiliary chips.

[@romancharak3675]

Thank you for allowing closed captions - very helpful for me!

[@CJWarlock]

Good video. I like the idea with the yellow arrow pointing to the components shown on the schematics. I appreciate all the work, seen and unseen in this video. Congrats. Cheers! :)

[@MeriaDuck]

6:09 So neat and tidy and compliant, quite the change from the typical dodgy power supplies ❤😂

[@objection_your_honor]

The cat squeezes the pillow, you squeeze the cat. The circle of life...

[@henriquelausch6999]

15:36 Quite a fancy gate discharging circuit! I'll use it in my designs from now on instead of a ol' capacitor

[@fabriziobrutti1205]

Nice! And you showed very well the waveforms of a CrCM PFC!

[@mikemccormac9368]

Thanks

[@DiodeGoneWild]

Giant thanks for your support!

[@mehmetturkcanerel761]

If you keep the neutral connection to the DALI line of the LED driver fixed and the phase connection touched and pulled once, you can turn the LEDs to the off or on position, or if you keep the phase connection solid for a short time, you can have a dimmer.

[@edvanes5178]

Knowledge is power and you have the power ! Great explanation !

[@andrewd762]

Thanks DiodeGoneWild, great video and much appreciated. I'm a little bit more knowledgeable than i was 15 minutes ago.

[@janno288]

New diodegonewild video! Today is going to be a great day!

[@sylvainbebert3704]

Very good idea to show arrows on pcb while explaining schematic ! Cool as usual !

[@aicisha]

I see an improvement in the quality of your videos, nice!

[@Trught]

SetLED and gndLED is for setting current for leds in range 0.2 to 0.7A and DALI is dimming protocol.

[@TomboRectify]

Nice! Maybe one day I'll be equally skilled at reverse-engineering PSUs

[@mikemccormac9368]

I agree with your cat (as always!). Why such complexity just to make light. I know; it gives your followers the pleasure of seeing the innerworkings of how things we take for granted, actually work😀.

[@user-ui6xt4fd1f]

11:08 i used to explore faulty driver for 22W philips industrial LED lamp, it has non-sync buck topology (w\o pfc) with UF54xx diode, and the fault reason was overheating of this diode lead to bloating of output electrolytic cap and bloody roasted to black pcb, so diode in 150W driver may cause quick house-burning event) its seems to be exactly conduction losses, coz mosfet body diode has even worse reverse-recovery time than diodes in continuous current mode. its also kinda funny that driver design has temperature-protecting circuitry but it was omitted)

[@eGnarF]

That atmel chip is a 48mhz arm m0+ mcu with usb. Seems a bit overkill for that application. But could probably be modded for fun features =)

[@ColBecket]

these LED power supplies are programable via NFC (you can se that blue antena near output) I dont see min value imho. Also that one you check is DALI power supply with additional output for LED control (like dimable switch or via smartphone) thats why there are some chips inside.

[@EJEuth]

Outstanding video, thank you!

[@d.t.4523]

Thank you, keep working.

[@xDR1TeK]

The L6388 is the worst gate driver imaginable. The internal bootstrap diode is very fragile. Substitute it with an external diode for safety because if it's overdriven internally, it goes short circuit when it fries under its load. The external bootstrap would help it stay alive. But if the internal bootstrap diode gets fried, nothing can save it, the whole thing fails. If anyone is reading this comment, take note, these things have to operate in switching mode, pwm. If you bypass the bootstrap by supplying the maximum voltage to drive the highside gate, note that the macimum voltage possible isnt high enough. The limit there is quite low. You'll just be disappointed. Go for a different gate driver if you dont want surprises. Its really duch a bad driver for high voltage applications.

[@b_70_saurabhvishwakarma84]

Sir please make a video on the RC EDM arc generator ( electrical discharge machineing)

[@davey2k12]

Get electroboom to do that it'll be so funny 🤣🤣😂

[@b_70_saurabhvishwakarma84]

@@davey2k12 🫡😂

[@SuperHaptics]

Great video again. Thank you!

[@diegoalfadelpinotasso2972]

Tiene una construccion similar a los antiguos lastres electronicos con microcontrolador de los tubos fluorescentes, tiene un circuito PFC identico, lo que veo que cambia es la salida que es para leds, si se puede explorar mas otro video seria interesante, cierto, hay mucha complejidad solo para solamente producir luz.

[@jp040759]

Great summary at the end.

[@1marcelfilms]

It would be nice to modify it so the power is 20% of the rated power for more efficiency

[@eDoc2020]

Just snip some of the current sensing resistors on the output.

[@omniyambot9876]

will you ever make a dc-dc converter designed by yourself, i mean a video

[@YoYoElectroFix]

Wow a bitmap 👌🏻😍

[@AKG58Z]

It can also be used with some other things.

[@tomashrabe9874]

Nice video like always. One question, could you ever create video about how you draw a schematic of some device? I dont know, how to manage components on paper. Thank you very much.

[@8Ugri8]

Again i learned something, thank you!

[@1pcfred]

That is one nice light driver.

[@simonilett998]

Dany, could you possibly do a video on how a self oscillating switch mode power supply works please? I have a couple of smps that have all the commonly seen input components, fuse, ntc thyristor X type capacitor, emi filter, bridge rectifier etc, but it doesn't have a switching chip or an optocoupler. It only has one main 400v/100Uf capacitor, 2 high voltage E13007 switching transistors, a small ferite ring oscillator transformer with (3×) 3 turns + (1×) 3/4 turn, a few resistors, a few diodes, 1× db3 diac, and a large output transformer with 3 wires primary, 4 wires secondary. The DC output is also very simple. It uses 1× double diode, 1× large electrolytic cap, 1× small electrolytic cap, 2x small ceramic caps, a few resistors and 1× ferrite core choke/inductor. I'm looking to repair them so I'm trying to better understand how they actually work?👌👍

[@eDoc2020]

That sounds like it might be an unregulated supply. It would be like the classic 2-transistor fluorescent driver circuit but with a transformer+rectifier instead of an inductor+lamp. Or I could be totally wrong. Since you're saying the output transformer has many wires I'm most likely wrong.

[@simonilett998]

​​@@eDoc2020Thanks for your reply. These are GME Electrophone PSS100 switch mode power supplies sold as 13.8V DC supplies for CB radio use. Rated at 13.8V-4A continuous, 12V-5A peak. On the box it states, fully suppressed for radio equipment, car stereos etc, but they don't state regulated output as such. I can find circuit diagrams for their other GME classic linear iron transformer power supplies ie. PSA123, PSA126, PSA1210 but nothing found for these PSS100 supplies. These PSS100's are made in Taiwan, and the circuit layout seems to share a lot of similarities to a half bridge design, but don't use a TL431, or optocoupler, or switching chip etc so I'm really not sure how they control the switching of the two MJE13007 transistors, or how they regulate output if they even do. The other thing I'm not sure of is what the DB3 Diac does, and if it is a significant part of the control circuit etc.

[@cambridgemart2075]

The marking on the power supply label looks like it was done on a worn out inkjet printer. Is this a genuine Phillips item?

[@godfreypoon5148]

Calculating efficiency by measuring rms input current can lead to a hugely pessimistic efficiency measurement due to the effect of the nonsinusoidal current measurement inflating the current measurement. rms voltage multiplied by rms current is not real power, it is apparent power.

[@RixtronixLAB]

Creative video, thank you :)

[@J.D-g8.1]

Great video as usual! I also like your accent. It sounds almost like singing. :)

[@hugoegon8148]

Is X6 the ISP interface for the Atmel? Perhaps one can access the chip, read the memory and reprogramm ... 😊

[@Freeman666Gordon]

Why 350V 68uF capacitors in series? What such significant difference made for? Is it not enough to have a one 450V 47uF?

[@DiodeGoneWild]

I guess a 33uF 450V capacitor would do, but the designer was in a series mood, or the 350V capacitors were cheaper, or maybe they had a much lower ESR or a higher temperature rating.

[@berndeckenfels]

The Dali control seems (according to the schematic on the package) just a momentan switch?

[@objection_your_honor]

Is it possible to put enough leds in series to match the mains voltage and not need any resistor?

[@DiodeGoneWild]

That would be very steep...

[@MichaelGiacomelli]

Lots of cheap LED bulbs put a lot of 9 or 15v diodes in series to get up close to mains voltage. Then a linear current regulator dropping a few volts to set the current to some safe level and keep it regulated as the bulb heats up and the diode voltage changes.

[@DiodeGoneWild]

Of course, but there still has to be the regulator. It can't be just the LEDs alone.

[@objection_your_honor]

@@DiodeGoneWild Why not come close to mains voltage, and then drop the rest using a resistor?

[@shivasisdash7343]

And the Cat is Right 👍

[@makoveliprod]

15:58 3 amps pnp transistor in sot23 package wow.

[@makoveliprod]

15:58 i tried to use low gate threshold p channel mosfet in this turn off circuit, but it works worse than pnp transistor. Why?

[@Alexelectricalengineering]

Amazing 😍👍❤️

[@worroSfOretsevraH]

Hi. 2 questions: Why they always use a 3.15A fuse on almost every power supply board? Why in every single transformer they use yellow (kapton?) tape, no matter if it's made in china or else? Thanks.

[@Ni5ei]

A typical Philips street light driver. Not even sure if these are available in retail stores.

[@LawpickingLocksmith]

Your cat is right: Humans are screwed.

[@mernokimuvek]

I will always prefer plasma based lighting over boring L*Ds.

[@andymouse]

Phillips, and its spelt correctly so not rubbish but the engineer asked to design a control for an LED light must have thought 'why me what have I done ?' so he took this piss with this design !....cheers.

[@liam3284]

Interesting it works on DC

[@piconano]

Half the input bridge rectifier carries the input current burden, and makes polarity irrelevant.

[@LMB222]

"Made in Poland" hasn't meant anything for ages. The real info is in the Certificate of Origin (COO) that the customs offices require. Of course you and me will never see it.

[@objection_your_honor]

The Chinese version, has one or two current limiting resistors, and the hell with the rest!

[@AKG58Z]

It is complex just to drive LED, but not for Philips.

[@zyghom]

I am always surprised when you say which connector goes to L and which to N - it is AC circuit and in Europe you can always connect opposite - the socket/plug don't always connect the same way - so you never know where is L and where is N. Unlike in other countries where socket/plug direct always the same connection.

[@DiodeGoneWild]

Czech sockets are polarized. And even in countries where sockets are not polarized, permanent wiring still keeps N and L. This thing isn't probably connected using a socket.

[@zyghom]

@@DiodeGoneWild a keyword: "Europlug" - for sure not polarized.

[@zyghom]

Practices differ mainly in the area of polarization. Typically, Europeans do not maintain consistent identification of line and neutral throughout their power system as is the practice in North America. Consistent with this practice, the Continental European plug can be rotated to either of two positions and plugged into the socket. Thus, the common electrical system in Europe is unpolarized (i.e., line and neutral are connected at random). In fact, most plug types used in Europe are not polarized.

[@maheen-]

We know they take the design seriously, when we see even an atmel microcontroller in it.

[@sanjikaneki6226]

at this level of complexity why not isolate from mains it would not cost probably that much more + safety

[@DiodeGoneWild]

It surely would be more costly and lossy to isolate it.

[@tiagoferreira086]

@@DiodeGoneWild I believe the losses would be the main reason due to certifications and energy efficiency class, they even went to the trouble to add a pfc circuit.

[@liam3284]

looks as complicated as a program start ballast.

[@jonlitchfield8888]

There is one thing you always fail to mention. Cats name. I want to donate regularly, but to do so will mean stopping longtime donations to fred hollows foundation, who restore sight to needless blindness in pacific islands, through free cataract surgery. Would that make me a bad person? Or a dirty donation? Edited to add: The cat that was sleeping on me suddenly threw up all over my legs 2 sec after i posted. Ill take that as a prompt to just keep things as they are, and get extra work hours. And a visit to the vet.

[@eDoc2020]

Her name is Veritaska (but with an accent or two).

[@jonlitchfield8888]

@@eDoc2020 thankyou, appreciate it 👍😀

[@arturtrzebinski2112]

Polandball mentioned! 🏳‍🌈 (no polish flag in emoticons tho :()

[@MieleW2573]

🇵🇱

[@mangalsutar7293]

❤❤

[@inseries5494]

your commenters get fascinated by stupid things: pointer to show components, a cat with 4 legs, etc

[@duzaliteraf7373]

1 second ago No views No comments No likes I'm first

[@user-vu4bi1rz3u]

❤❤❤❤❤❤❤❤❤

[@Weedic]

What is wrong with people spamming emojis

[@user-vu4bi1rz3u]

This is not a fake comment or spam. I love watching this man and I follow him from Egypt. I learn a lot from him.. Anyway, welcome.@@Weedic

[@isimisoko]

The videos are good but U talk too fast and subtitles disappear too fast