Help me figure out this LED lightbulb circuit

[MattHollands]

I have a Phillips 5.5W 240V LED bulb that has stopped working (ie. no light), and I want to fix it (I realise this is probably not an economical use of my time but it's fun).

I have attempted to reverse engineer the circuit and I think I've got it mostly, but something looks odd. Schematics and images attached. I have tested the chain of 6 LEDs and they turn on at around 30V. The 2.2uF capacitor was my first guess, but it seems fine and I am getting 320V DC across it, but I am ordering a replacement anyway so I can replace it. In the meantime, I want to understand the operation of the circuit better. I don't really want to scope the signal because I don't want to damage my scope, but I am able to use a multimeter as I set up the probes before turning it on.

I don't understand what the transformer is for... I would have guessed it was for isolation, but it seems like the LEDs are not isolated. Are they just using it for some kind of feedback? Also surely Z2 needs some resistor to pull it up?

Ignoring the transformer my guess is that R8 and R9 create some voltage feedback to regulate the current through Q1. Q3 is some kind of protection that turns off Q1 - perhaps R10 (unmarked) is a thermistor and turns on Q3 when it gets hot? I'm a bit lost. I would have thought that Z2 would need something pulling it high.

I am measuring about 400mV across Z1 and 0V across Z2.

Any suggestions for operation would be appreciated.

Matt

[jasonhanjk]

The simplest method to re-use my LED bulb.
Went to Taobao and ordered an LED driver, it's case up using ABS material so I won't get a rude shock. I just plug into my 230V mains and connect the output across my LEDs. Had to jumper across one of my LED to make it work since I bought a lower output voltage one. I also increase the feedback resistor to reduce the brightness to my need.

[sokoloff]

That soldering looks terrible to me for a production part. Did it come like that originally?

[oPossum]

Looks typical for a lead-free wave soldered TH/SMD board.

I don't know how that thing is supposed to work, but I would start troubleshooting by using diode test mode to check the bridge, transistors, and LEDs. If any of those are bad, it's not going to work.

[IanB]

That soldering looks terrible to me for a production part. Did it come like that originally?

Doesn't all lead-free soldering look like that?

[StillTrying]

Also surely Z2 needs some resistor to pull it up?

It looks that way, but Q3 would still conduct when the voltage from its emitter's T1 winding is negative going, but then Z2 is the wrong way around - used as a diode and not a zener. Q1, T1 and Q3 looks like an oscillator to me.

In fixing it my 1st 4 suspects would be Q1, the 2 2M resistors R2 and R3, and then the transformer T1.
In other words, I don't have much idea.

[tombi]

Does the transformer form some sort of resonant converter?

Like a joule thief where the transformer characteristics determines the frequency of operation.

The next question is why would it be built like this?

Surely the failure must be heat or transient related so I vote for R2 or R3 going HR. THat fits with 320V at C1 and nothing at Z1. Or Z1 is a short?

No idea - just a guess.

[MattHollands]

That soldering looks terrible to me for a production part. Did it come like that originally?

Well I've probed it quite a lot with my multimeter, but yes that's more or less what it looked like when I opened it.

Also surely Z2 needs some resistor to pull it up?

It looks that way, but Q3 would still conduct when the voltage from its emitter's T1 winding is negative going, but then Z2 is the wrong way around - used as a diode and not a zener. Q1, T1 and Q3 looks like an oscillator to me.

In fixing it my 1st 4 suspects would be Q1, the 2 2M resistors R2 and R3, and then the transformer T1.
In other words, I don't have much idea.

Z2 is definitely that way around

I've taken Q1 out of the circuit and it seems to work ok. R2 and R3 measure about 1.6M each which makes sense as they are in circuit so seems ok to me. What are the failure modes of a transformer?

Does the transformer form some sort of resonant converter?

Like a joule thief where the transformer characteristics determines the frequency of operation.

The next question is why would it be built like this?

Surely the failure must be heat or transient related so I vote for R2 or R3 going HR. THat fits with 320V at C1 and nothing at Z1. Or Z1 is a short?

No idea - just a guess.

Z1 definitely isn't a short - connected ~1.5v across Z1 and a 10R resistor and see about 700mV which makes sense for the gate of the transistor. R2 and R3 seem to measure ok... A bit lost

[alsetalokin4017]

The failure modes of a transformer are open circuit windings, interwinding shorts, and shorts to the core.

But if this were my project the first thing I'd try is to reflow all the solder joints, using a sharp tip iron and a touch of fluxcore leaded solder.
Second thing would be to replace Q1, even if it checks OK out of circuit.

I have rewound these little transformers to fix open windings. It's like microsurgery. Remember to count turns when you unwind it looking for the break!

[MattHollands]

The light is now working intermittently. It seems like reflowing various points in the circuit makes it work for a little while At first I thought it was something to do with R6 and R7, but it also seems to work if I heat up some of the transformer connections... After reflowing certain joints, the light works for a little while and then stops again!

For example, it wasn't working, and then I reflow the transformer joint where it connects to V-, and then it started to work for a bit, but then stops after a while...

I want to get this fixed because I want to know how the circuit works.

I suspect a crack in a trace somewhere that resolves as soon as any heat or pressure is applied. I'm slowly going to replace connections with jumper wires and see if that gets us anywhere

[TurboTom]

Did you check the choke L1? In your photo, it appears to be cracked close to the upper terminal. I had several LED lamps fail this way, even quality ones. The defect may also be intermittent.

[ocset]

Did you replace the electrolytic?...you mentioned you were ordering it...this is a highly likely cause of the problem.

[Gyro]

Did you replace the electrolytic?...you mentioned you were ordering it...this is a highly likely cause of the problem.

A perfectly reasonable suggestion, but if you're right, it would be the perfect comedic 'up yours!' to the rest of us. 

[Emergency getout] It is a hot running consumer bulb rather than an industrial luminaire.

[MattHollands]

Did you check the choke L1? In your photo, it appears to be cracked close to the upper terminal. I had several LED lamps fail this way, even quality ones. The defect may also be intermittent.

Good spot. I shorted out this inductor and the device is definitely more reliable. But sometimes when I turn on the light, it just doesn't work until I turn it off and on again. I have measured the voltage across the capacitor, and when it lights up it is 300V and 330V when the light doesn't work. Video demonstrates the problem (just uploading). Could this be to do with inrush current?

https://youtu.be/A99Tt___cB8

Did you replace the electrolytic?...you mentioned you were ordering it...this is a highly likely cause of the problem.

Part hasn't come yet :/ I agree that the cap is by default the most suspect part, but even when the light is turning on, there is still 330V DC across the cap so it seems to be holding up ok.

[StillTrying]

"I have measured the voltage across the capacitor, and when it lights up it is 300V and 330V when the light doesn't work."

That's proof that the cap is fine.