Who knows what kind of SMPS is this one ?

[ym58]

I have purchased a second hand 12V/50A (600W) SMPS .

I tweaked the output voltage to 14.6V and limited the output current by cutting off three shunts out of 5 in the output negative rail.
No OVP triggered, I was quite lucky and happy ... until I pulled 40A output current to a 50% SOC 173A LFP4 battery (40A was the max current I could get after 3 shunts were cut-off)

I rapidly toasted (less than 10 sec) the 2 x  MOSFETs, 2 x gate power resistors and 2 x PNP gate transistors !
The fuse (on the AC input) blew up not fast enough to save the previous components life.

I decided to order parts from AliBaba (it came rather fast) and replaced all above-mentioned parts.
I decided this time to reduce the output current to 20A or so (using power resistors as load instead of the LFP4 battery)
Bang ... the same components blew up again  !

I changed them all again (for the last time) and managed to raise slowly the output current (by ways of using different power resistor values) from a few amps up to 10A max ... the SMPS now works fine but I am quite sure it won't be able able to output more than 150W without blowing up again  .

What can be wrong with this PSU  ?
Who can suggest me a possible block diagram of this PSU (the 2 x small transformers in the pic below don't make me recognise a classic fly-back half-bridge design ... ) ?

Thanks for your helping me sorting this out  !

[madires]

It's not a flyback, but likely a half-bridge with the small transformers driving the MOSFETs and the two brown caps between the large electrolytics connected to one wire of the large transformer's primary winding. Search for "half-bridge SMPS" and you'll find plenty of schematics.

[Faringdon]

The capacitors may have dried out, and so you are getting overvoltages, which blow things up.
When you blow up a power mosfet in the primary of an offline smps, that smps is often kapput.
This is because the deadened mosfet can go short from drain to gate, and the gate connects to the signal circuitry...so all of that may have gotten high voltage on it....or higher than normal, and have been weakaned.
I have worked in repair shops, and when a primary fet blows...that SMPS is removed from the repair bench...and stated "beyond economical repair"

[ym58]

It's not a flyback, but likely a half-bridge with the small transformers driving the MOSFETs and the two brown caps between the large electrolytics connected to one wire of the large transformer's primary winding. Search for "half-bridge SMPS" and you'll find plenty of schematics.

Do you mean these two ones ?

[2N2222A]

Is this power supply normally current limited? Or does it just have an over current protection trip that you lowered by removing some of the current sense resistors?

But yes, a bad capacitor on the input side of the choke that is on the output side of the main transformer can cause the power supply to increase its power too much to try to maintain output.

[ym58]

The capacitors may have dried out, and so you are getting overvoltages, which blow things up(...)

Which capacitors ?
The ones @madires mentioned in his post or the two large black electrolytics capacitors 820uF/200V (in series, by the way)

[Wolfram]

That unit is a power supply with overcurrent protection, not a battery charger. Tweaking voltages and current limits might make it work like a battery charger or it might not, depending on details that don't matter for its intended use. As you have discovered, this unit is not happy charging batteries, and there can be multiple reasons for that. Control loop design for a battery charger is very different than for a constant voltage supply, so it might not be operating in a stable manner. Many supplies are also not designed to operate for prolonged time in current limiting mode, or deliver any meaningful fraction of the nameplate power rating, unless they are specifically designed as constant current supplies.

Using this power supply as a charger could involve major changes to the control side, which would take serious reverse engineering of the original design to understand where it falls short. This would be an interesting exercise in power supply design if you want to dig deep, but likely be more expensive and time consuming than just designing something from scratch. Directly mains powered SMPS are very unforgiving, both in terms of personal safety, difficulty of probing, and the potential for catastophic failures on time scales shorter than a millisecond. I would strongly recommend designing something that operates from lower voltage if you want to take that path. A buck converter operating from an off-the-shelf 24 V supply, with proper voltage and current loops compensated for battery charging, would be my approach for a problem like this, and I design battery chargers for a living.

If you want to experiment with hacking supplies for charging, and you don't have any strict requirements on longevity or reliability, then there are some good learning opportunities here. I would start with drawing a complete schematic of the unit, and characterize all the magnetics.

Edit: The 50H165R and one of the small transformers is likely an auxiliary flyback supply to power the control section. The main PWM controller could be on the small riser card next to the MOSFETs. The presence of an output inductor suggests that this is not a flyback converter, but a (two switch) forward or half-bridge topology.

[madires]

Do you mean these two ones ?

Yes, but check if they are really connected to one wire of the primary winding.

[ym58]

@2N2222 :
yes, I think it's a current limiting-type SMPS for it has shunts on the output negative rail and also because I haven't noticed any 'hiccup' behavior ... (well, so far).
you wrote :

a bad capacitor on the input side of the choke that is on the output side of the main transformer

... could you be more specific ?

[ym58]

Yes, but check if they are really connected to one wire of the primary winding.

I will

[Faringdon]

Any of the big electrolytics on the primary side going dry could cause the problem...esp the two big cans you said were in series.

[ym58]

That unit is a power supply with overcurrent protection, not a battery charger. Tweaking voltages and current limits might make it work like a battery charger or it might not, depending on details that don't matter for its intended use. As you have discovered, this unit is not happy charging batteries, and there can be multiple reasons for that.

Not quite sure ...
As I mentioned in my OP, an escalating output current yielding a BBQ frying of the mosfets and transistors also occurred with a strictly resistive load (power resistors) ...
I have modified several 'current-limiting' Chinese SMPS (as well as hiccup ones, by the way) so as to charge LFP4 batteries and never met such an uncontrolled escalation of the ouput current ... so IMHO, there ***MUST*** be something wrong with ***THIS*** unit 

[ym58]

Any of the big electrolytics on the primary side going dry could cause the problem...esp the two big cans you said were in series.

Sounds like a right track, I will try to swap those four (if I can source the brown ones) ...

[2N2222A]

you wrote :

a bad capacitor on the input side of the choke that is on the output side of the main transformer

... could you be more specific ?

This may be incorrect and not apply to your power supply But electricity comes out of the main transformer, in to a diode or diodes. The diodes go to a choke, but there is a capacitor there to absorb the pulses coming out of the diode. That big toroid choke in your picture is the choke. I believe it's this first capacitor that often fails, resulting in reduced output, and excessive Voltage spikes due to a bad low ESR capacitor. This causes the power supply to increase the PWM until it is destroyed from too much Voltage flyback on the primary. This may not apply to your particular power supply. You check by connecting an oscilloscope to it while it is operating to check this low ESR capacitor. I'm referring to C19 and the others next to it.

The big black capacitors on the input side are rarely the problem. They are normal capacitors not the special low ESR type that are used in the high speed switching. If the power supply has a 110/220 Voltage selection swich, it is these capaictors that make up the Voltage doubler when set to 110. If these capacitors go bad then line buzz or humm will enter the output of the power supply, or it will malfunction due to low Voltage. This is what keeps the power supply working for a fraction of a second after it is unplugged. These capacitors are also the most dangerous as they together hold over 340V DC after it is unplugged for a while.

The brown capacitors are a dry type that don't ever fail. They're used to give the big black capacitors some low ESR properties and reduce noise in the AC supply.

Take a marker and draw a line going through the middle of capacitors to separate the primary side from the secondary to make things easier to work on, if there isn't a line already.

I haven't noticed any 'hiccup' behavior ... (well, so far).

Because the power supply burns out before it reaches its rated 40A output.

[jonpaul]

I feel that the OP is not qualified nor has the equipment or capabilities troubleshooting or modifying an offline SMPS, especially with no manual nor schematic.

this kind of " fooling  around" can lead to damaged test equipment, injury or death.

I advise the OP to drop this project and spend some time to learn about SMPS, design theory, topology and safety precautions eg isolation transformer.

Just the ramblings of an old retired EE (HV and SMPS since 1970s)

Jon

[ym58]

@2N2222A : thanks for your academics reminders, that will surely help those whose EE knowledge is poor to dare tackling SMPS issues !
I will swap the output stage electrolytic caps (i.e. C19 and the likes) from this board with newer ones (maybe 25V besides, as these are 16V, which is kinda marginal as I tweaked the SMPS to 14.6V) but honestly I feel that the area I should rather investigate is 'upstream' ... I think I will have to dig out my scope !
In the past years, I hacked numerous power SMPS to convert them into LFP4 or Li-Ion chargers (from 4S to 10S) like in this thread but this one really sucks ... its particular topology with the two tiny transformers (see pic in OP) isn't really familiar to me (not a flyback, nor a half-bridge ...  )

[madires]

As already suggested, one of the small transformers could be part of a flyback SMPS (standby or for SMPS control) while the other one is driving the MOSFETs. The circuit isn't complex and can be reverse engineered without much effort.

[f4eru]

culprit could well be the AUX winding supply of the regulator putting a too high voltage to the chip, having all sorts of bad effects.
It could also be the regulation being marginal, and getting unstable with the higher loop gain....

Etc....

[ym58]

(...) The circuit isn't complex and can be reverse engineered without much effort.

Yeah I know, you're right @madires ...  but I've been 'reversing' so often that I'm kind of weary now , very time-consuming for what it's worth !
But anyway, some good news at least for tonight, it looks like the HV caps (the ones in series) were a bit weary too
I replaced them with newer caps (470uF/200V) I had handy somewhere back on my shelves and the PSU now works much better and bears 25A without frying the Mosfets and gate components !
( @2N2222A : I think @faringdon hints and my feeeling were right, the output stage caps were not the culprits )
The SMPS is now under test connected to an inverter + water heating resistor (250W) ... so far so good.
---
Next step : I will try to output 50A to charge LFP4 batteries ... but for the time being : no rush (I don't have any 20N60C3 left) !

[2N2222A]

But anyway, some good news at least for tonight, it looks like the HV caps (the ones in series) were a bit weary too

So you changed those too, meaning you also changed the output stage capacitors in addition to the HV mains input ones?

[ym58]

So you changed those too, meaning you also changed the output stage capacitors in addition to the HV mains input ones?

Nope, as I wrote I only replaced the two HV electrolytic capacitors : 820uf/200V (replaced temporarily with 2x 470uF/200V, the only ones I had handy).
But I also took this opportunity to purchase and order a good ESR meter and will definitely replace the output stage caps if they turn to show a high ESR as you hinted.
(I also purchased HV 820uF caps coz my 2x 470uF are too low in the long run)

[ym58]

Yes, but check if they are really connected to one wire of the primary winding.

Yes, I confirm they are.
Those two caps are OK though, I checked their capacitance and resoldered them in place.

[Faringdon]

When you power the circuit up .
Best is to not use the bias supply if it doesnt work...just switch in 12v, (or whatever ) from an external supply.
Also.....make an external board with a 1A (or low value) fuse, and an heavy  inrush resistor, eg 47R, 50W.
Always switch it on via the inrush resistor....then you can enable the PSU after you switch out the external inrush resistor.
Read the chip datasheet to see how to disable it
Or just disable it by switching out its vcc with the external vcc supply

Be very careful not to touch....everything, including the external boards is mains live and can in theory kill.

[ym58]

@faringdon :
Thanks alot for your following-up but the SMPS now works ...    !

I replaced the two HV caps with 2x 470uF/200V as the original ones (820uF/200V caps) showed very high resistances on the ESR meter.
I will later replace them again with higher capacitance values (820uF-1000uF) but for the time being these two 470uF do the job.

But your suggestions and advice are quite interesting though, at least to improve my own knowledge :
- what do you exactly mean with don't use the bias supply (12V), what do you refer to ?
- also where would you insert the 1A+inrush resistor in the circuit ?

FYI, a basic diagram of this SMPS half-bridge topology can be drawn like that (very roughly ...) :



And the driving of the two Mosfets gates is based on a piggyback-mounted TL494/KA7500 PWM controller :




(as shown above, I managed to control the SMPS Iout by ways of adjusting R24 value)
---
Oh and by the way, I am very sorry for what happened to your country yesterday !

[Faringdon]

Oh and by the way, I am very sorry for what happened to your country yesterday

Thanks, i reckon Lloris payed his spurs team mate Kane to fire that penalty over......(OK, not really)

Anyway, you have a capacitor in series with your half bridge...there is much  to speak about that...i will come back later on it.