SMPS failure

[johnmx]

I tried to fix an SMPS power supply without success. The schematic of the power supply is practically identical to the one attached. One difference is, for example, that C11 does not exist.
The switch mode regulator (FD9020D) was burned out with an almost short circuit (~6 Ohm) between the power pins (1 and 8 ).
I checked all the other components and they were good. I particularly checked all diodes and resistors. I measured the capacitor C10 with an LCR meter and it had 9.8uF with 4.5 Ohm series resistance. A somewhat high value, but I don't have any of these capacitors to replace it. I even tested the secondary circuit and it seemed to be working fine.
I bought 10 units of the FD9020D on AliExpress, replaced it, turned on the power supply, and the 12V appeared at the output (correct value), but after about 10 seconds it burned out again.
The failure was the same, low resistance between pins 1 and 8.
I replaced the PWM controller again, turned on the power supply, and it burned out again after 10 seconds. But this time the 3 current sensing resistors (R4 and R9 from the attached circuit) also burned out, probably because the chip burned out and kept the output transistor on.

What do you think it could be?

[wraper]

С1 would be a first suspect to me. Also R2/R3, C3, D5.

[xavier60]

If C5 goes high ESR, flyback doesn't get clamped properly on the primary side. Also check D5.

[johnmx]

I desoldered the main capacitors to measure them on the LCR meter and obtained the following results:

	Original value |	Meas. cap. |	Meas. Rs
C1	22uF/50V	19.6uF	1.13Ω
C5	470uF/16V	450uF	0.07Ω
C4	470uF/16V	435uF	0.07Ω
C10	10uF/400V	10.0uF	3.45Ω
C3	2.2nF/1200V	2.3nF	274Ω

The only capacitor that stands out is C3 with a very high equivalent series resistance.
R2, R3 and D5 are fine.

Note, L3 does not exist in the real board. The circuit is slightly different. All component designators mentioned here are based on the schematic, not on the actual board.

[wasedadoc]

Does your LCR meter give a similar apparently high ESR reading for another capacitor in the 2.2nF ballpark?  Does not need to be a high voltage one for that check.

[johnmx]

Does your LCR meter give a similar apparently high ESR reading for another capacitor in the 2.2nF ballpark?  Does not need to be a high voltage one for that check.

Yes, it gives similar results with new ceramic capacitors. I was measuring at 1kHz. At 100kHz, the Rs of C3 dropped to 9.8 Ohms. The same happened with a new SMD ceramic capacitor of similar value.

[wraper]

IC you got might be a counterfeit (different IC relabeled) too.

[johnmx]

IC you got might be a counterfeit (different IC relabeled) too.

I doubt it. The IC is a low-cost Chinese brand, and the markings appear identical to the original (observed through a microscope).
Additionally, the original one had the same issue, a short circuit between the power pins.

[wasedadoc]

Does your LCR meter give a similar apparently high ESR reading for another capacitor in the 2.2nF ballpark?  Does not need to be a high voltage one for that check.

Yes, it gives similar results with new ceramic capacitors. I was measuring at 1kHz. At 100kHz, the Rs of C3 dropped to 9.8 Ohms. The same happened with a new SMD ceramic capacitor of similar value.

That suggests your C3 is not faulty.

[johnmx]

I'm running out of ideas on what could be wrong with this power supply... 

[Coordonnée_chromatique]

I'm running out of ideas on what could be wrong with this power supply... 

Delayed failure caused by an ESD?

[boB]

I know you said you checked the components but my first thought was something wrong with the input bridge rectifier.

You might try checking the voltage across what would be pins 1 to 8 without the controller IC in circuit to see what it is.

[johnmx]

I know you said you checked the components but my first thought was something wrong with the input bridge rectifier.

You might try checking the voltage across what would be pins 1 to 8 without the controller IC in circuit to see what it is.

I tested all diodes of the bridge rectifier. I don't recall the exact value of the rectified voltage, but it was around 300 and something volts, which seemed correct.
I couldn't check the voltage around pins 1 and 8 because the chips burned before I had a chance to measure it.
My initial suspicion was that the chip burned due to excessively high voltage in that pin. However, I didn't verify this with an oscilloscope. The IC itself should limit this voltage, at least there is no external component to limit it.

[boB]

OK then.  If those two 1M Ohm resistors are OK, then that part is most likely fine

[johnmx]

OK then.  If those two 1M Ohm resistors are OK, then that part is most likely fine

Yes, there are. I even removed one of them to measure them correctly.

[magic]

Maybe try replacing C3 for peace of mind.
Inspect or redo solder joints, maybe a connection is intermittent somewhere.

Is it possible that the current sense resistors were already too low in value before you blew them?
These may be hard to measure accurately with an ordinary DMM.

[fmashockie]

What is the resistance between pins 1 and 8 on the board with FD9020 removed?  I'd be suspecting D7 or D5 if you measure low resistance without IC soldered in.

But that is based on the schematic you provided.  Which is just an application schematic from the FD9020 datasheet.  How sure are you that this schematic matches the SMPS you are working on. You said the only difference is C11?  I find that hard to believe...

EDIT: My bad I see you said it is 'pratically' the same.  Could you possibly trace the schematic out a bit?  Maybe notate some more of the differences? That might help people to better help you

[Audiorepair]

Its possible the transformer core is cracked, or shorted turns.

Either could drastically change the primary inductance into more of a short.

[wraper]

Its possible the transformer core is cracked, or shorted turns.

Either could drastically change the primary inductance into more of a short.

It's s flyback with gaped core. Some crack won't make difference large enough for it to fail.

[Audiorepair]

Personally, I would never rule out anything bought from Ali Express wasn't fake or reject.

[eblc1388]

I measured the capacitor C10 with an LCR meter and it had 9.8uF with 4.5 Ohm series resistance. A somewhat high value, but I don't have any of these capacitors to replace it.

The ESR value of C10 is too high. The MOSFET inside the PWM IC was probably destroyed via high back emf voltage from the switching.

The datasheet of FD9020D calls for 2x 10uF 450V filtering capacitors but the SMPS vender skimmed on budget and use only a single 10uF 400V capacitor. You should replace it by two 10uF 450v in parallel or a single 22uF 450V capacitor instead. Consider mounting the capacitor horizontally if space is insufficient.

I attach a scope measurement of the back emf when the filtering capacitor is faulty in another similar SMPS I have repaired earlier.

[wraper]

The ESR value of C10 is too high. The MOSFET inside the PWM IC was probably destroyed via high back emf voltage from the switching.

Such ESR can be normal for some brand new capacitors, especially miniaturized. No way it's high enough to kill the IC. Not to say OP measured at 1kHz which is not correct and will show higher ESR value.

The datasheet of FD9020D calls for 2x 10uF 450V filtering capacitors but the SMPS vender skimmed on budget and use only a single 10uF 400V capacitor. You should replace it by two 10uF 450v in parallel or a single 22uF 450V capacitor instead. Consider mounting the capacitor horizontally if space is insufficient.

[rteodor]

It may be the optocoupler.

[MathWizard]

So when it burned out, did you have a secondary load hooked up ? I see in the datasheet that the FB voltage gets divided down and compared to 3.7V. So if the secondary or opto had some problem, the chip should reset or shutdown.

But yeah I would try and scope the input drain of the chip, surely some overvoltage or over current into that pin or the StartUp voltage pin, or even on the low voltage Vcc winding, if that get too high for some reason, or it's cap fails, would kill the chip pretty quick.

[johnmx]

What is the resistance between pins 1 and 8 on the board with FD9020 removed?  I'd be suspecting D7 or D5 if you measure low resistance without IC soldered in.
(...)
EDIT: My bad I see you said it is 'pratically' the same.  Could you possibly trace the schematic out a bit?  Maybe notate some more of the differences? That might help people to better help you

The resistance between pins 8 (+) and 1 (-) is 2.3 MΩ and >50MΩ between 1 (+) and 8 (-).
The differences on the primary side are: C11 does not exist, C1 is 22 uF and in parallel with a ceramic cap of 100nF, R11 is 10Ω.

Its possible the transformer core is cracked, or shorted turns.

The primary coil resistance is 2.15Ω and 0.58Ω for the auxiliary coil.

Personally, I would never rule out anything bought from Ali Express wasn't fake or reject.

What makes me not suspect this is because the original controller also broke in the same way.

I want to replace all capacitors, but it may take a while to buy them. Additionally, I wanted to check the circuit with an oscilloscope using a new controller, but for that, I want to power the board with lower voltage (< 100Vac) to avoid damaging the scope. I don't have high-voltage probes. I have a Variac with an Isolation Transformer for this type of measurements, but it's in another location and I may only have access to it next month.