I have an LED light bulb in which there are 16 LEDs in series powered by a constant-current supply.

One of the LEDs burned out so I just replaced it with a blob of solder and it worked for a while, but then started flickering.

I measured the voltage drop on each LED and they all had a voltage of about 5.5V except one which had 5V. I bypassed this LED as well (putting a blob of wire parallel to it), but this did not help with the flickering at all.

The next thing I did was to add another capacitor across the LEDs. I kept the already present 2.2uF cap and added a 4.7uF parallel to it. This did not help either.

How do I go about solving this problem? Of course I could just buy a new light bulb, but I don't care about having a working lightbulb, I just want to know how to troubleshoot such issues in general.

Circuit schematic: 3

Videos of the LED flickering: video 1, video 2.

Datasheet of the driver (MT7828C): https://datasheetspdf.com/pdf-file/1260493/MaxicTechnology/MT7828/1 ‒ the very last paragraph indicates a possible the cause of this problem:

The VDD (Pin1) de-coupling capacitor (typically 1uF ~ 4.7uF ceramic capacitor) must close to the pin (Pin1) as much as possible. Better not exceed 5mm. This can greatly improve the system noise immunity.

On the actual PCB, the distance of this capacitor from the VDD pin is about 5mm.


The flickering is mostly gone after I re-soldered the wires connecting the PCB to the LEDs (which are mounted on a sheet of aluminum). During normal operation (i.e. no flickering), the total voltage across all LEDs stays near 69V. When the flickering happens, the voltage drops to around 61V. I shall now determine whether there is also a drop in current (in which case the fault is most likely in the power supply) or not (in which case the fault is in the LEDs).

  • 1
    \$\begingroup\$ If your LEDs are burning out, then de-coupling capacitor placement is not indicative of your problem. \$\endgroup\$
    – earl
    Feb 23 at 22:07
  • 1
    \$\begingroup\$ The 5-5.5V measured across a lit LED (should be 3 or 6V but not in between) makes me think that the controller is switching on and off and your meter is measuring a lower average voltage. Keeping in mind that the voltage is very high, is the board itself really warm? Can you measure the voltage drop across the parallel 4.7 and 10 ohm resistors safely? Be very careful given the voltages involved. \$\endgroup\$ Feb 23 at 22:20
  • 1
    \$\begingroup\$ @user253751 I realize it is a buck converter, but the controller should not be switching off. If the voltage across the capacitor is dropping well below the diode forward voltage, than the current is not constant and the device is out of regulation, which is why it is flickering. I'm curious if it is overheating, it isn't hitting the target current, or something else. \$\endgroup\$ Feb 23 at 23:12
  • 1
    \$\begingroup\$ @user1850479 The board is not particularly warm; the voltage drop on the CS resistors is about 45mV. \$\endgroup\$ Feb 27 at 22:07
  • 1
    \$\begingroup\$ @Nedd Vibration has no effect; soldering on the PCB seems fine. I suspect what's causing the flickering may be the blob of solder that I used to replace the burnt-out LED. \$\endgroup\$ Feb 27 at 22:11

1 Answer 1


Instead of adding to the excessive comments I'm listing this group of additional troubleshooting ideas as a potential answer.

In your update you say the flickering is mostly gone, so you're implying that the problem still exists but something has changed. Perhaps you just altered the original problem area somewhat. If you still suspect a current changing issue look again at the current sense resistors, they control the inductor current and LED current. As for other issues the tips below might help to finally track down the problem.

  1. A subtle solder joint crack is opening up along the LED path (or on other circuit points) only after a slight warming of the circuit.
  2. One or more LEDs have a weak contact to the aluminum heat-sink causing an LED to over-heat.
  3. Since two LEDs are already suspected of being defective perhaps yet another is still defective causing the flashing.

For the first two issues you might try using what used to be called freeze spray. This was originally a freon like spray that would super cool a single component or PCB area and if defective or marginally defective it would drastically change the operating characteristics. Cooling of a normally operating component may still have an affect but not necessary as significant. You could try cooling other components or solder joints too, even the controller chip. Of course if you want to use this idea try to find a freeze spray that is ozone friendly, no sense in riling up any environmental extremists, if you have them there.

For the last idea (another bad LED), a concept from one of the comments might be useful. If you were to temporarily place a three Si diode shunt across a single white LED you would effectively be taking it out of the circuit. (A 3 diode shunt would have a VF of about 2.1 V which should be significantly below one white LED's VF.) If you have small wire jumper clips they might let you temporarily connect the 3 diode string across each LED. By shorting each LED in this way you might eliminate the one that is causing the problem. Also instead of clipping directly to a specific LED's pins you might contact each adjacent LED pin, that way you could eliminate any potential problems with that LED's solder contacts.

Finally, for the above techniques I'll again stress that working on a live line powered device can be very dangerous so please use extreme caution.

  • \$\begingroup\$ Thanks for the pointers. I also considered the possibility of overheating but this is unlikely -- the videos were taken right after turning the bulb on and the flickering is apparent. I decided to first determine whether the fault is in the power supply or the LEDs by using a different power source for the LEDs, possibly also a different load for the supply. It's not going to be straightforward though, because the circuit behaves a bit unpredictably; e.g. one time I measured LED current of about 58mA, the next time it was barely above 30mA. Both measurements were long-running (30 mins). \$\endgroup\$ Mar 1 at 20:20
  • \$\begingroup\$ During these tests I was also measuring the total power consumption using a power outlet wattmeter: the bulb was consuming about 7.3 W in the first case and 3.4 W in the second case. \$\endgroup\$ Mar 1 at 20:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.