Sign In


Forgot your password?

Create Account


Teardown, autopsy and hack of dead SMD LED floodlight.

Teardown, autopsy and hack of dead SMD LED floodlight.

I was sent these two lights to take apart and see if I could find what had gone wrong. They are 10W LED floodlights with a very slim profile due to the use of a large array of half watt SMD LEDs on a PCB with their mains power supply.

If you like my videos you can support the channel with a dollar for coffee and cookies at


List your electronic components at

Tags: smd led

About the Author


  1. Date: February 4, 2016 at 10:13
    Author: leviathan937

    An easy way to remove components from Al. or copper based PCBs is use a fry pan on very low heat, or an electric sandwich maker. Then use a heat gun over the top to re-flow the board. If the PCB has thermal paste on the back, stick a sheet of Al. foil to it first so you don't get goop all over your kitchen utensils.

  2. Date: February 4, 2016 at 10:55
    Author: Graham R Dyer

    Um ! what to do with those  cases ? I know you could make some Led lights….Your way. Nice one Clive.

  3. Date: February 4, 2016 at 11:11
    Author: DjResR

    I had trouble with fixing one of the LED bulb with the same 5630 emitters (one chip version though), every time I replaced an LED, another one failed. I ended up gutting the switching power supply and replaced it with small 220nF capacitor dropper circuit with a bridge rectifier. Nice repair.

  4. Date: February 4, 2016 at 11:35
    Author: whatlions

    I've soldered/desoldered probably hundreds of opamps from various multilayer boards with no space around them because of other passives. Not wanting to use a heat gun for various reasons, my method is just to dump some flux onto the pins, cover them all with solder, use a soldering tool that has a flat angled tip to lift one side of the IC, leaving the solder on the iron, then doing the same for the other side. For re-seating the same IC, I just use some flat pliers to flatten and angle the pins correctly. If you put the IC into an alligator clip on a holder (by the IC body of course), you can run the soldering iron underneath the pins while gently brushing them from the top with a wire brush tool to remove all solder bridges and clean the pins up without bending them. Very quick process. I've done this many times on 48 pin SSOPs (0.5mm pitch) with 0 issues. The keys are careful inspection with a magnifier, using good flux, and being sensible about temperatures. Flooding the chip with solder means that you have to work quickly. If it's not working out, let the chip cool etc. This method also totally avoids lifted pads when executed correctly, since you should require very little torsion to lift one side of the chip once the solder is all melted.

    Also IR printouts, scope printouts, eBay printouts… it looks so wrong but feels so right haha.

  5. Date: February 4, 2016 at 11:57
    Author: electronbox

    I have had success soldering these high thermal mass boards, albeit a bit crude I admit, by heating them carefully with a heat gun (mine is temperature controlled), and sticking on the top of the board a thermocouple with kapton tape to monitor the temperature. You can take off the SMD's with a pair of tweezers, the power leads will need a high power soldering iron, but a pre-heat of the board may help a lot – J.

  6. Date: February 4, 2016 at 13:34
    Author: Geoffrey Cubus

    This video description in Chinese English:

    I had to take in addition to sending these lights, see if I can find anything wrong. Does anyone have a 10W LED light very thin plate due to the wide range of half a watt SMD LEDs and power consumption.

    If you like my video, you can support the dollar coffee and biscuits intersection of

  7. Date: February 4, 2016 at 14:20
    Author: Nadim Awan2

    Thank you for this, I don't have to throw away my 2 failed lights. Mine were advertised as 20W but were running at 10.2W

  8. Date: February 4, 2016 at 18:46
    Author: Autunite

    What do you think about these in the 12V version? Would they have a better reliability?

  9. Date: February 4, 2016 at 20:02
    Author: Gordon McLellan

    I like those enclosures, cast aluminum? Be a nice case for making my own floods with proper Philips LEDs and a smarter driver 🙂

  10. Date: February 4, 2016 at 20:07
    Author: MegaWayneD

    Those housings look like they'd be perfect for some 12 volt 5050 white LED car light panels. I've got some in the interior lights of my car and they're stupidly bright (it was cheaper to order the LEDs from Ebay than actually buy new incandescent bulbs!)

  11. Date: February 4, 2016 at 22:22

    Hey Big Clive, id love to see you take apart an electronic breathalizer. thanks for the videos!

  12. Date: February 4, 2016 at 22:46
    Author: John Wilde

    Get it right bigclive, it is Aluminium.
    Love your vids.

  13. Date: February 4, 2016 at 23:36
    Author: YourMUMY

    i have no idea what you say but its darn interesting .

  14. Date: February 5, 2016 at 04:27
    Author: Teardown Dan

    Are these lights actually rated for 240V? 20 chips at three LEDs each is 60 LEDs total vs the reference design's 72, that's an extra 36V the chip needs to drop in current-limit mode during about 110 degrees of each half-cycle. That's an extra 1.6W for those final stage current-limited drivers to get rid of on top of the original design's 4W. (Well, 0.8W and 2W each with two chips sharing load.)

  15. Date: February 5, 2016 at 06:08
    Author: Lierofox

    …Is it just me or is the bridge rectifier in the wiring diagram at 1:40 and the larger diagram at 1:50 connected so that it blows the ass out of itself for half the AC waveform?

  16. Date: February 5, 2016 at 16:29
    Author: Nerfyy

    I almost never know what you are saying because I know absolutely nothing about circuits and whatnot, but I love watching your videos, they are very interesting.

  17. Date: February 6, 2016 at 01:26

    Hi Clive. I commission Caterpillar generators for the offshore Oil & Gas industry. We have virtually all LED and electronic ballast fluorescent lamps on the new-build platforms.  I notice that with ONLY lighting loads running, the power factor is quite leading (capacitive) but the actual load is so small that the generator doesn't care. Once you start a few HVAC compressor motors or other inductive loads, the power factor of the entire facility is still lagging. Power factor of lighting load on a facility with a few megawatts of motor loads and a few kilowatts of lights is a non-issue.

  18. Date: February 6, 2016 at 03:17
    Author: Darryl 603

    Could you heat the aluminum on the back of the circuit board to prevent the heat sinking while soldering (or desoldering)? Thanks for the videos and looking forward to the next…

  19. Date: February 6, 2016 at 17:52

    Probably poor design.
    Just because a TO-220 transistor may be able to dissipate 100W, doesn't mean that it can under most circumstances!
    Transistors usually have a maximum rated power dissipation like that, which is measured with the transistor at 25C … try keeping the transistor die at 25C! It's VERY difficult to do!

  20. Date: February 7, 2016 at 01:40
    Author: V T

    I have two LED PIR floodlights on the front of the house. One is a 10 watt and the other a 20 watt. Different physically – not the same manufacturer.

    Just recently the 10w stopped working altogether and the 20w is just glowing a bit. It's appeared to have happened during the very low temperatures we just had recently. So I'm thinking that may have had something to do with it.

    Anyone got any thoughts?

  21. Date: February 7, 2016 at 12:08
    Author: AutomotivEivind

    For testing which chip was the bad one, could you have tried just lifting the one output leg of the chip of the board?

  22. Date: February 7, 2016 at 13:56

    Nice analytics. The 2835 can run up to 1 W THEORETICALLY. Now, if it comes down to: Low chip efficiency and or at 100% run and also that the heat of the outer leds is less so that their voltage might go lower….then this design is absolutely built to near beside the water. Apart from the fact that this 2835 led housing I would never run more than 0,15W per 2835 case. Now the aluminium PCB material might be not a clean one, that means only half or less heat conductivity. Then the housing….might be also a rubbish not RoHS standard. Even it is the ADC12……which complies to RoHS….then the heat conductivity is already 3 x less than pure aluminium. Now, if it is not RoHS… might be even less than that. Then the heat transfer must go smooth in the housing……..the casings are mostly not very smooth so that I would use hight thermal conductive silicone but in the case of using high voltage I would use a 5000-1000V protected isolation layer (standard is 3000V which actually also let the heat more easily passing through but also can be more easily damaged by hot soldering and then would not isolate 3000V – another EXTERNAL teflon layer beneath this aluminium board and glue it with heat conductive silicone. If grease would be used….it will melt and might shrink away or if not highest quality the oil will evaporate and lets only a part of the PCB good heat conducted. This design shown here is not good at all. Also front side….is not safe I guess since if the glas might shatter and someone touch the leds, it can cause get electric shock, moreover it MUST have at least 5 free mm if not covered to the next conductive element and that is the refelctor: What is if in case of a malproduction a small wire conducts one of the led pads with the refelctor or housing (that pads must be with a rigid cover or at least hard glued with rigid layer of resin or similar, but not soft). So: You get that what you pay for it. The leds run at 100%… so this design won´t work long and the lumne efficiency is little so you pay more electricity ( I guess up to 100% more than a flood lamp with 150 Lumens per watt, this one only might have 80). The lumens of the leds further correclate with their chip T. That is another reason why never let the led "core reactor" run over 0,2W for that 2835 Led-Casing.

  23. Date: February 7, 2016 at 17:44
    Author: Ed ash

    I find it funny, that they complicate things TO MUCH…

  24. Date: February 8, 2016 at 07:23
    Author: Ian Oliver

    Milliohmmeter? (I've seen some cool designs for those recently, including one from Scullcom here on YouTube.) Or just cut the one pin that's shorted, on only one chip, and re-test. That way you only have one pin to resolder if you picked the wrong one.

  25. Date: February 9, 2016 at 02:12
    Author: Ken Smith

    Aluminum castings tend to come out a different size than the mold. If they took the dimensions from the mold the screws will not quite line up.

  26. Date: February 18, 2016 at 07:23
    Author: Rich Booth

    +BigCliveDotCom Do you take requests?

    I've been watching you teardown and diagram LED circuits for a while, and still trying to grasp running COB LEDs without an LED driver, just a high wattage resistor.

    If one has a 10W COB LED(3×3) with the following ratings: Forward Voltage (VF): DC 9-12V and Forward current (IF): 1050MA. I know that running the LED on 12V with a 3.9Ohm resistor works, but I don't know how to calculate that it will. I know the answer resides in V=IR. How would you explain this? I did a quick Google search, and it appears in not the only one confused.

  27. Date: March 9, 2016 at 19:08
    Author: Sean Ross

    Are you trying to help the Chinese engineers? lol Keep in mind that even though a chip is labeled something such as "CYT3000A" it very well might be a counterfeit inferior chip in these cheap devices. Thanks for the video though, one might conclude that since they both failed in exactly the same way that they might just be engineered to fail!

  28. Date: March 15, 2016 at 16:31
    Author: Nscale

    Is the flickering of the LED normal? Should not compensate the capacitor it?

  29. Date: April 4, 2016 at 07:20
    Author: Jack Smith

    +bigclivedotcom would you be able to put a 10w chip on there? Obviously insulating the previous tracks?

  30. Date: June 14, 2016 at 20:48

    +bigclivedotcom I just bought a "100 Watt" version of this design, and made a 1200 Frames Per Second video that (sort of) shows how the CYT3000A LED driver switches in and out the several banks of LEDs. My half-arsed analysis of what can be seen is in the description under the video.
    It can be seen at the following link:

  31. Date: August 29, 2016 at 01:11
    Author: MIKES0029

    These are selling in he US (120v, 10w) for $1.99 free shipping on eBay, couldn't help to buy a few to play with

  32. Date: October 24, 2016 at 18:52
    Author: Jeremy Lister

    Chips used instead of cap dropper for better power factor.
    I'm looking for a decent PIR LED floodlight that will last 10 years.
    Most on the market are IP44 which isn't high enough. Even IP65 ones appear to have poor design on the PIR with many opportunities for water entry.
    Anyone make an IP66 PIR floodlight?

  33. Date: November 16, 2016 at 21:35
    Author: Syn Kronos

    Too much thermal compound can be as bad as not having any.

  34. Date: November 17, 2016 at 10:58
    Author: bjtaudio

    The idea of the smart chip flood light and running the leds directly from the 240v mains, is all great but, the smart chips have been failing and blowing up the led chips…for higher reliability the cap dropper and smoothing cap is a better idea. excellent video, good work.

  35. Date: February 13, 2017 at 10:09

    Hi great video just the right balance between theory & practical
    Did you remove the surface mount fuse to fit the 680nf capacitor ?

  36. Date: February 20, 2017 at 02:02
    Author: Raymond J

    i acutually seen the black spots on them leds before you said it . i learned that from you brother hehehe

  37. Date: May 23, 2017 at 19:09
    Author: Bradxyz

    Hi. I have one of these, mine is a bit bigger, 100 watts. 3 of them have failed with the same method. After about 6 months of use, the light stopped glowing brightly and appears to only glow the multiple LEDs just a very little bit. I have tested the fume resister with an ohm meter, and it is passing current. Any idea? Perhaps I could send it to you.

  38. Date: October 19, 2017 at 21:09
    Author: TheWacoKid1963

    Nearly bought one of these lamps from B&Q, There own Blooma brand, Till I saw the reviews on them. The majority of them were failing within a year. there not cheap either. Going to stick with the old halogen lamps.

Leave a Comment