LED bulb flickering

[edy]

I have a relatively new LED BULB that is flickering now when on, at about 20% brightness. And when I turn it off it remains on and flickering for a few more seconds until it shuts off (like it's draining a cap). Any ideas on where to start? Nothing seems burned. I've attached some photos.

[sleemanj]

One of the leds will be failing most likely, flicking off/on is a common failure mode.

[Kjelt]

First how are you applying 20% brightness, are you using a standard mains dimmer ?
If yes, read the specifications of your dimmer, usually they need 40W minimum for proper behaviour.
If so add a 40W lightbulb (not led) parallel to the led bulb and repeat the experiment, if it still flickers we can continue. 

[edy]

No dimmer... This is directly switched on/off through mains switch. I turn on mains, light flickers rapidly and is about 20% in brightness I guess due to it not turning on long enough during each flicker to fully illuminate. It is not just "dim"... I can perceive the flickering so it must be at a slow enough rate that my eyes can tell. It's unlikely to be doing it at 60 Hz the mains frequency.

When I unplug or turn switch off, cutting power out instantly, the bulb continues to flicker another few seconds and then dies out. When I plug it in, it starts up immediately and flickers.

I can test the LED's individually and also see what voltage is coming out of the power converter to the LED  board. What kind of voltage is normal to run these types of bulbs? I assume each LED has it's own resistor to control the current? I haven't looked at it carefully enough yet.

[abraxa]

I'm not an SMPS expert but to me it sounds as if the output cap is either not charged fully (creating a pulsed DC waveform on the output) or its capacity dropped and it can no longer sustain the load of the LEDs. I wouldn't be surprised if the output cap overheated and failed. If you have a suitable cap on hand (max temp doesn't matter for this test), I'd replace it and see what happens.

Please be careful, though. The fact that the cap that seems to be the output cap is rated at 250V suggests that the LEDs are connected in line and that the LED voltage may be unsafe. Either way, measuring the voltage at the LEDs won't help imo because the waveform won't be pure sine AC or DC and thus not providing much of a clue.

Edit: The turn-off behavior makes sense if the SMPS output draws too little power: the big input cap has plenty of charge left to keep the low-power output running for several seconds.

[Kjelt]

No dimmer... This is directly switched on/off through mains switch

Ah ok ignore my remarks in that case.

I can test the LED's individually and also see what voltage is coming out of the power converter to the LED  board. What kind of voltage is normal to run these types of bulbs? I assume each LED has it's own resistor to control the current? I haven't looked at it carefully enough yet.

Count the leds and multiply with 3,3V or 6,6V which is most common. hOwever there are white ledpackages  that have even more than two leds inside.
No resistor, it is current regulator.

[Armadillo]

Suspect the buck controller IC thermal protection or current limits, either the IC is faulty or the feedback network.
what is the IC number? cannot see clearly.

[SeanB]

I would check those 220k SMD resistors and the 2u2 50V capacitor, as those are the power supply for the chip, and the resistors going high value or the capacitor low will result in the chip trying a rapid series of startup cycles as it goes into and out of the undervoltage lock up as the voltage on the capacitor varies.

[orbanp]

I would touch up the solderings on the back, they do not look too good!
Though it could be the just the picture quality, but I would give it a go.

Good luck, Peter

[cdev]

How do you get the bulb dome off, is it just kept in with hot glue?

Also, i would not test anything on one of those bulbs while plugged into a circuit because they are often not isolated.. its unsafe.   Remove from power completely before testing continuity etc.

I have got a bunch of dead LED bulbs. I think it may have been simply that my ceiling fixtures hold the bulbs sideways not straight up or down. Could that cause bulb failure?

There is nothing on the packaging about anything like that.

The ones that are vertical have lasted, the sideways ones no.

[mariush]

A very often failure is the bonding wires that connect the led die to the terminals becoming loose... sort of like as the led heats up the wire bends or changes length and loses contact with the pad where it was soldered to or spot welded and so the led turns off and cools down and the bonding wire again moves/changes form and touches the pad again ...

Sometimes just knocking the pcb with your finger or pressing on the led gently will make it turn on or off so you'll know which led fails.

Depending on the voltage drop of each led you may be able to just replace the led, or use a zener diode to replace a failing led or just take out the led and short the pads.

[cdev]

Note: I am not the OP but I have had a similar problem with one particular manufacturers bulbs (others have been fine) Unfortunately I bought a lot of them initially because I liked the natural color of the bulbs .
As I said I think they failed because the fixtures they were in held the bulbs sideways, not up and down.

Now I have a bunch of them that don't work. If I cant return them to the manufacturer I think I will just remove all the LEDs and test them individually and use them elsewhere. I guess I feel confident in other areas, but don't feel confident enough with my "skills" even if I was able to fix it, to plug a formerly malfunctioning LED bulb back into my AC power system without worrying yet.

I'd rather just use the LEDs in low voltage projects. I have enough now to last me several lifetimes.

[edy]

I'll try to attach some more images but having some trouble making them smaller than 1 Mb each.

[negativ3]

You have to be careful. Some manufacturers state the down light enclosure needs to be open type and I assume this is to prevent overheating. I had a batch of GE bulbs go bad and found it was the primary side of the transformer which had dry solder joints. Had to remove the potting to fault find but it was good to find the cause.

Sent from my SM-J710F using Tapatalk

[edy]

Ok I  figured out how to resize the photos on this phone and will post the rest here. It looks like the 9 LED are in series. There are lots of cracks lines on the board... Thermal stress? Also some of the solder job on the main board look shady but it was working fine before so I doubt it is that.

You can see the chip numbers clearly this time but they are upside down. One says HA5832E and 6A090SZ, and beside it one says ES1J. The one labelled MB10F seems connected directly to mains. Maybe a rectifier? Half or full wave? There are 3 caps on the board and a transformer.

Anyone have a schematic or can do a Dave-cad representation? I don't think the value of the caps are clear on the photos. I may have to look and take more photos. As far as the plastic dome, that just pried off with a thin knife. The glue was dry and crumbled away.

The bulb was installed in up position with socket at bottom and bulb standing straight up like a hot air balloon.

[edy]

Ok so my hunch was right, the MB10F is a rectifier diode, and the large cap is probably smoothing out ripple current. ES1J is a fast rectifier. The main IC on there is HA5832E is an efficient LED constant current controller power supply translated from Chinese as follows:

"HA5832E is a high-precision non-isolated step-down LED controller, anti-flicker, for 85V ~ 265V full voltage range of low power Non-isolated step-down LED lighting applications    HA5832E built-in high-precision sampling, compensation circuit, so that the circuit can achieve ± 5% or less constant current accuracy, and can achieve output Current to the inductor and the output voltage of the adaptive, so as to achieve excellent linear adjustment and load regulation    HA5832E internal integrated 500V power MOSFET, no secondary feedback circuit, and no compensation circuit, coupled with accurate and stable self-adaptation Should be the technology, making the system is very simple external structure, the number of peripheral devices in a small number of parameters under the conditions of loose to achieve high precision constant flow control System, which greatly saves the system cost and volume, and can ensure that in mass production LED lamp parameters of the consistency    HA5832E has a wealth of protection: output open short circuit protection, sampling resistor open short circuit protection, undervoltage protection, output overvoltage protection, too Temperature adaptive adjustment and so on."

I just noticed a spot on the board that says RS1 (next to the HA5832E chip and the blue resistor) seems unpopulated, as if an SMD component should be across there. Then again, there is RS2 and RS3 also written and I'm not sure what it is referring to. The other SMD resistors are listed just R1, R2, R3, so not sure what RS is supposed to be. Also one of the pads for the HA832E chip is not used (no solder). I don't think any of this has to do with the flickering bulb but just an observation on the design of the board.

Previously in the thread someone mentioned 3.3V or 6.6V per LED, and there are 9 LED.  The LED board says "27mm8W9LED" which I assume to mean it consumes 8 watt. They appear to be in series which means if we chain 9 together the voltage drop is either 29.7V (if each is 3.3) or 59.4V (if they are 6.6). So does that mean we need 269 mA (if they are 3.3 V LED) or 134 mA (if 6.6V) current driving the circuit?

[Armadillo]

The RS1 position is vacant and is replaced by the bigger resistor shown in the picture. It is the Current Sense Resistor. Both of them are on the same trace. Pin 7 and Pin 8 is the current sense pin. The resistor is connected onto Pin 7 or Pin 8 and Ground. With the value of the resistor, is it 1.8 ohm. This resistor program constant current through the LEDs. In this case it is = 0.5 / 1.8 ohm = 280 mA.

Pin 2 is the Overvoltage Pin, It is tied to a resistor to Ground to program the over-voltage setting.

The arrow, please touch up the soldering work at the pin first.

Pin 4 is the VCC pin, measure the DC voltage at Pin 4 with respect to Pin 1 [Gnd Pin], it should be 17Vdc [be careful, dangerous voltage on the board, use isolation transformer if available].

Touch up the soldering first. May save you the trouble to measure Pin 4.

Previously in the thread someone mentioned 3.3V or 6.6V per LED, and there are 9 LED.  The LED board says "27mm8W9LED" which I assume to mean it consumes 8 watt. They appear to be in series which means if we chain 9 together the voltage drop is either 29.7V (if each is 3.3) or 59.4V (if they are 6.6). So does that mean we need 269 mA (if they are 3.3 V LED) or 134 mA (if 6.6V) current driving the circuit?

The controller ensure constant current through the series of LEDs by automatically adjust the voltage to achieve the programmed current [280mA].

[abraxa]

Previously in the thread someone mentioned 3.3V or 6.6V per LED, and there are 9 LED.  The LED board says "27mm8W9LED" which I assume to mean it consumes 8 watt. They appear to be in series which means if we chain 9 together the voltage drop is either 29.7V (if each is 3.3) or 59.4V (if they are 6.6). So does that mean we need 269 mA (if they are 3.3 V LED) or 134 mA (if 6.6V) current driving the circuit?

If you look closely you can see that each LED case contains two chips in series (they're slightly offset from each other), so this makes 18 diodes in total. Nice pictures, btw
You can see this the easiest on the LED case on the top right corner in your picture.

I noticed that one of the LED chips in the case on the left (above the PCB cut) has a black mark where it connects to the internal bonding wire. It's the only one with such a black spot. Maybe there has been arcing there, so this LED may have failed in the way cdev described. It is indeed a very common failure mode due to insufficient cooling and the resulting excessive thermal expansion.

[Kjelt]

Previously in the thread someone mentioned 3.3V or 6.6V per LED, and there are 9 LED.  The LED board says "27mm8W9LED" which I assume to mean it consumes 8 watt. They appear to be in series which means if we chain 9 together the voltage drop is either 29.7V (if each is 3.3) or 59.4V (if they are 6.6). So does that mean we need 269 mA (if they are 3.3 V LED) or 134 mA (if 6.6V) current driving the circuit?

Yeah that would be me. Indeed but to test I always use a nice power supply with voltage and current control. I set the current limiter to 40mA or something very safe like that and turn up the voltage till the current starts flowing and is controlled to the 40mA. Than you can see the voltage of the ledboard. Probably around 56-60V in this case.
Unfortunately the chinese bulb makers skimp out on the cooling, usually very thin aluminium carrier and nothing else or another thin aluminium heatsink so it survives for a year or so. I always recommend not using these bulbs at 100% due to heat stress but use them dimmed. I have some tear down topic of chinese bulbs here on the forum. Now the cheap A brands also use chinese electronics these days but they have better cooling, usually the enclosure is metallic and used as heat sink or the cooling vents are larger.

-> !!! Be carefull these are unsafe voltages that could give you a small shock, do not touch the leds or board during experimentation not even with a power supply, all voltages above 24V should be handled with care.

[EPTech]

Can be multiple things:

- Fault in the diode string. Look for a faulty (shorted or semi-shorted ) LED's, which result in higher current draw
- Secondary cap has become high ESR (or open)
- secondary rectifier diode fault (shorted)
- transformer isolation issue or core gap. Although not frequent in my experience they do fail. Only way to check is to desolder en zap it with a megger. It should be able to withstand 2500V. Core gap can be inspected visually with a small amount of mechanical stress.
- faulty switch IC, not likely in this case, more likely the peripheral passive components, especially those with high values
- faulty primary bulk capacitor, less often than secondary.
- Bridge rectifier. I would definitely check this. because this one is the first in line to fail if the cause is external.

happy hunting.

[Armadillo]

Can be multiple things:

- Fault in the diode string. Look for a faulty (shorted or semi-shorted ) LED's, which result in higher current draw
- Secondary cap has become high ESR (or open)
- secondary rectifier diode fault (shorted)
- transformer isolation issue or core gap. Although not frequent in my experience they do fail. Only way to check is to desolder en zap it with a megger. It should be able to withstand 2500V. Core gap can be inspected visually with a small amount of mechanical stress.
- faulty switch IC, not likely in this case, more likely the peripheral passive components, especially those with high values
- faulty primary bulk capacitor, less often than secondary.
- Bridge rectifier. I would definitely check this. because this one is the first in line to fail if the cause is external.

happy hunting.

Wouldn't you say is a comprehensive piece.   

[edy]

Thanks everyone! I'm using this more as a learning opportunity, obviously it is not worth the $3 to buy another bulb to go crazy trying to repair this thing. But I'm just curious as to:

1. The actual circuit design (I'm trying to reverse it)
2. Learn about how to trouble-shoot failure
3. Learn what and how to test the circuit safely, and what normal voltage readings I should be getting
4. If I can't fix it, see how I can utilize the components

If one of the LED's shows that it is faulty, can I simply short across it and let the remaining 8 LED's do the job? If the circuit limits current across the entire 9 series, and one of the LED's is remove, will it not keep the current constant but now across fewer LEDs, resulting in each LED taking a slightly higher current? Would I not want to short out the bad LED with a resistor equivalent?

I assume for the purpose of just testing the circuit behaviour I can short the bad LED and see if it resolves the flicker. The question is why the flicker in the first place? If the circuit applies a certain VOLTAGE across the LED string, expecting a certain resistance... If there is less resistance, for the same V the AMP goes up.. so the chip reduces the current to keep it at a set rate.

So for example, if I remove 5 of the 9 LED in the series, the same voltage will be applied to the string of 4 remaining LED. The resistance of this path will be way less, so to avoid burning out the LEDs the chip will reduce the current as it senses it. But the effect of having a higher voltage across the LED will not change the brightness at all... because they are current-driven. But will it cause any other damaging effect?

If I decide to drive an LED with 3.3V at 20mA, how does it compare to me using it with 6.6V at 10ma? In each case, the power is the same since P=VI. I am over the threshold voltage in either case to drive the LED, and limiting my current to keep power equal with either resistor or some other method to control the current dynamically. I might be missing something silly, sorry if this is obvious to everyone. I'm still learning! 

[Armadillo]

For thousand over posts, it cannot be you are struggling with a simple buck converter.
Maybe you got story to be telling us instead?
like a one way radio broadcasting.?

[edy]

For thousand over posts, it cannot be you are struggling with a simple buck converter.
Maybe you got story to be telling us instead?
like a one way radio broadcasting.?

   Yes, I know... I'll need to do my homework. Sadly I haven't had as much time as I like to devote and so my learning curve has stagnated.

[amyk]

Watch bigclivedotcom channel on YouTube. It's more random now but he has enough videos on LED bulbs to satisfy your curiosity more than you ever thought possible.

[spidola]

It looks like these HA5832E are not readily available.

[chhrisedwards]

There are many reasons due to which an LED bulb can flicker. As I think this problem may occurs due to problem in its circuit. The voltage supply may also cause the LED bulb flickering.