Power supply board - industrial type from 1990's [solved]

[Grossel]

Hi.

Due to a ton of time thie easter, I've spend some time on this PCB board and draw up what the schematic should be. I put it up here if someone wish to discuss this design and my comments on it.

It takes 110V input, and use a UC2844N. What is spechial about this is the use of a BC557 PNP transistor in the feedback loop, I have some struggle to figure out about this because the datasheet for UC2844N doesn't include a similar use of a bjt transistor (ok, TI has an example design using a NPN transistor).

Thing is - those power supplies tend to stop working after approx 20 years in use, and the goal here is to find what component that is likely to fail at first. Maybe an electrolyte capacitor. The design doesn't seems to be designed to use be as power efficient as possible.

I do suspect those board may not be properly designed to handle when input voltage slowly decreases below ~85V (110V battery not being charged).

Components sharing same hetsink:
Primary: V6 and V22 (both mosfet's)
Secondary: A3 and V18, A2 and A4

I've not tried to put input voltage so I don't know what component are failing. I'm considering if I'd wait until I have a 110V DC source or just pickout some caps and use my ESR meter to test them.

 

[KeepItSimpleStupid]

C34

 

[Diver300]

The schematics have a few mistakes which makes reading them difficult. The +12 and +24 V output labels are swapped over. The main transformer power inputs V+ and V- are the wrong way round. It isn't clear where wires cross and where they join. It is easier to read if there are no points where 4 wires join. They can be staggered to have two T-junctions. For instance, I think that R23 and R24 are in parallel, but I'm not sure. The voltage feedback seems to have been forgotten.

I suggest that you look at R7 and R10, and at R9 and R12. I had similar resistors in series fail. It is outlined in this thread:- https://www.electro-tech-online.com/threads/failed-resistors.158347/

They were also failures of old SWMPs, after years of service, where resistors were put in series to keep the voltage across each resistor in limits. The exact failure mode isn't clear, but the consensus is that putting resistors in series is a bad idea.

 

[Nigel Goodwin]

The schematics have a few mistakes which makes reading them difficult. The +12 and +24 V output labels are swapped over. The main transformer power inputs V+ and V- are the wrong way round. It isn't clear where wires cross and where they join. It is easier to read if there are no points where 4 wires join. They can be staggered to have two T-junctions. For instance, I think that R23 and R24 are in parallel, but I'm not sure. The voltage feedback seems to have been forgotten.

I suggest that you look at R7 and R10, and at R9 and R12. I had similar resistors in series fail. It is outlined in this thread:- https://www.electro-tech-online.com/threads/failed-resistors.158347/

They were also failures of old SWMPs, after years of service, where resistors were put in series to keep the voltage across each resistor in limits. The exact failure mode isn't clear, but the consensus is that putting resistors in series is a bad idea.

Standard TV repair diagnosis technique - look for any resistors in series, VERY often that will be the problem - you can even predict it before the set is launched, just from the circuit. You can tell the manufacturers over and over again (and we did), but it makes no difference.

I'd think that the designer used a proper single resistor, and that it was changed in production (for cost reasons), but as a lot of designers seem to start straight from University, with no practical experience, I'm not so sure?.

I'm sure I've mentioned before?, I twice changed the mains transformer in a nice Sony amplifier (100W+100W - proper RMS watts) - same fault both times, the heat fuse in the transformer had blown. Now the transformer was a puny little toroid, sat in a space for a MUCH larger toroid - and comparing t's physical size to others, it was only rated at 60W. The customer was driving it hard, and it was overheating and blowing the heat fuse.

I spent months and months trying to get a sensible answer out of Sony, their only response was "we've had very few failures" - sorry, not the point Sony!.

So the point is, manufacturers VERY often make bad design/manufacturing decisions, causing failures which should never have happened.

 

[schmitt trigger]

Nigel; based on bitter experience, I fully agree with you.

My most bitter professional experience was as a FA engineer for a company which manufactured hi-rel power supplies.

The units were indeed very well designed and ruggedly built, but there would still be some low level, but consistent failures which would reduce the unit's MTBF below what was contractually obligated.

The view of most, if not all, design engineers was: Blame the vendor. Blame the manufacturing site. Blame the FA analysis.
Their design was perfect! It had been fully analyzed using Mathlab and PSPICE! It met design specification 12345, revs A thru Z, what could go wrong?

Only when an honest engineer would take a very close look at the failure, it was discovered that there were secondary, very obscure failure mechanisms.
This would only happen when the customer was fed up and was threatening significant monetary penalties.

But in consumer electronics.....good luck finding a company which acknowledges a design error.

 

[Grossel]

C34

?

The +12 and +24 V output labels are swapped over.

Thanks. That's obvious but I didn't spotted it first.

The main transformer power inputs V+ and V- are the wrong way round.

Not sure I get what you mean. You mean easier to read if I swap pins 2-6 ?

It isn't clear where wires cross and where they join. It is easier to read if there are no points where 4 wires join.

Ok, I can "put a dot" where all intersections are to make it more readable.

And yes, I'll definitely look into the series resistors. Thank you very much for looking into this one.

 

[Diver300]

The transformer inputs are wrong, I think. It's not about the ease of reading.

On the ease of reading, the schematic should be arranged so that dots are never needed.

If I've understood the left drawing correctly, the wire coming down to R24 joins the horizontal one, but I'm not sure. In the right diagram, I've moved the resistor to so that the dot isn't needed.

On another point, I can't see what R14 and R13 are doing. R6 is 500 times smaller than the R13/R14 pair, so R6 is the only one that matters and R13/R14 will only make 0.2% difference. The same applies to R5. I assume that R6 is used as a way of limiting or controlling the peak primary current, and there doesn't seem to be a good reason for such accuracy in the R6 value.

 

[ChrisP58]

This is a good thread about making clear and readable schematics.

https://www.electro-tech-online.com/threads/rules-for-drawing-readable-schematics.144863/

 

[KeepItSimpleStupid]

Why C34? I've seen way to many instances where ripple on the negative supply causes all sorts of instability problems. I've read it elsewhere too.

 

[gophert]

The bridge rectifier is not exactly a bridge rectifier...
Some of those diodes are bass ackwards.

 

[Grossel]

Thank you for reviewing.

The transformer inputs are wrong

Thank you. I oversaw some thing, but when I re-checked it turns out you've right. Initially I didn't got the diode name because the printing was facing down toward the pcb, so first when I got another similar pcb I was able to put the name on, but I totally forgot to see the error.

Why C34? I've seen way to many instances where ripple on the negative supply causes all sorts of instability problems. I've read it elsewhere too.

Thanks. Put it to my short list oc components to desolder and perform quick esr reading on.

Some of those diodes are bass ackwards.

Thank you (I did learn new english phrases on dumb stuff, how cool is that?)

Another thing I found wrong - some of the resistor color bands was silver, not grey/white (R5/ R6). Corrected now.

Also as pointing out, I'v put the dots around on connection ponts to make the schematic easier to read.

 

[dr pepper]

The bc547 transistor on the rt/ct is a trick you can do for slope compensation if the duty cycle goes over 50%, you mix some of the oscillator ramp into the feedback, your schematic doesnt look correct in this area too.
Sometimes the feedback pin is not connected, this is sometimes done for speed, feedback is taken directly to the o/p of the error amp on the compensation pin.

 

[Grossel]

The bc547 transistor on the rt/ct is a trick you can do for slope compensation if the duty cycle goes over 50%, you mix some of the oscillator ramp into the feedback, your schematic doesnt look correct in this area too.

Thank you for feedback. Yes, it turned out I'd swapped the base and emitter pin.

 

[dr pepper]

No problem.
A better way to measure transformers is to measure the inductance instead of resistance, inductance is often used instead of turns ratio for design of smps's (except perhaps for push pull).

 

[Grossel]

No problem.
A better way to measure transformers is to measure the inductance instead of resistance, inductance is often used instead of turns ratio for design of smps's (except perhaps for push pull).

Already done, but haven't put it onto schematic.

 

[Grossel]

Also, on my last upload I forgot to remove internal labels (I'd made those to try to keep track while reading the pcb, but exept from that it make no sense on the schematic itself).

 

[Grossel]

After measuring ESR (capacitance and D on all electrolyte capacitors) on 3 similar pcb boards, the conclusion is there is one single component that have failed on all three pcb cards - C10.

That is branded "SAMWHA" and rated 105 deg.celsius.

 

[Nigel Goodwin]

After measuring ESR (capacitance and D on all electrolyte capacitors) on 3 similar pcb boards, the conclusion is there is one single component that have failed on all three pcb cards - C10.

That is branded "SAMWHA" and rated 105 deg.celsius.

There's your problem then - fitting of a cheap crappy electrolytic in the first place. That exact schematic location (reservoir on the IC supply) was a common failure on PSU's in domestic electronics as well.

You should replace all the electrolytics, and use good quality 105 degree ones (I quite like Panasonic ones) - just because the original was 105 degree doesn't mean it was any good, a crappy 105 degree capacitor is still crap!.

To be 'fair', it's like they did eventually pay some attention - and upgraded from 85 degree to 105 degree, but it didn't really help, as they still bought the cheapest crappiest ones they could find.

 

[Grossel]

just because the original was 105 degree doesn't mean it was any good, a crappy 105 degree capacitor is still crap!

Very true indeed

 

[gophert]

There's your problem then - fitting of a cheap crappy electrolytic in the first place.

Ha, SAMWHA is Korean manufacturer of capacitors since the 1950s.
I had a research project with them when I was in the chemicals industry (specialty electrolytes). They use their excess capacity to make caps for other companies (e.g. Panasonic caps with "made in Korea" labels). Good companies like Panasonic don't use suppliers that make "...cheap crappy electrolytic in the first place". Private label production is common to balance surge capacity and avoid tariffs with "swaps".

Grossel
The electrolytic In the C10 position have been strained for its 20-year lifetime and it was simply a PCB design error rather than a "crappy electrolytic". It is just soooo easy and simple-minded to declare the ROOT CAUSE of a failure is the cap. Look deeper snd you will see the datasheet specifically warns about PCB design and component selection to avoid trace inductance they cause resonant voltage spikes (which can cause capacitor failures - even it if took 20-years). Read section 11 of the datasheet for the IC. It is completely possible that the SAMWHA capacitor lasted longer than other brands would have under similarly stressed conditions.

Just because the name is not common in the US or UK, doesn't mean it is a crappy brand. They are a small company, and happy to be small, they are quite busy and profitable - even if people like Nigel want to call SAMWHA products "Crappy".