Late 1970s SMPS Design Query

On a sunny day (Sun, 24 Jan 2016 22:03:05 -0000 (UTC)) it happened Cursitor Doom wrote in :

So scope the voltage on all filter electrolytics for spikes,. If it is 'noisy' as you say, and cannot maintain voltage, then it is very likely just bad electrolytic caps.

If the caps are good you do not see any sudden voltage changes on those (like spikes up and or down), but just some ramps. The other way for a quick test is short the cap with a new one, and see if the ripple on it improves a lot, if spikes go ways.

And do not forget those Spraques, use scope on AC. In fact Philips elcos are OK, I still have some I use for test circuits that are from the seventies. All electrolytics deteriorate over time though. The hotter the faster, and the fact that your board seems 'well done' is one more reason to check them all.

Biggest problem at the moment is the very limited time I get to probe around during live testing. That 4W resistor gets too hot to touch within about 20 seconds (accompanied by burning smell) so I have to keep turning it back off to allow the thing to cool down again before probing some more. It's a serious PITA. I took the little pwm transitor out of circuit (all the small signal transistors are socketed - yay!) to keep that resistor from overheating but then the 3k3 power resistor got even hotter even faster so it seems I can't win. :( Going to carry out some passive checks for shorts/low-res parts over the next couple of hours...

There's way too much noise all over the board, Jan. Can barely see any underlying signal at key points in the circuit for the overwhelming noise

- and I don't get much time to probe around (15s tops) before things start to overheat. It's a big problem.

Which 3K3 power resistor? R1804 is present to supply logic start-up power. It is disconnected internally when the scope power rails stabilize.

If you want to check out primary logic, disconnect input power completely and apply bench lab low voltage DC across C1821. Local regulation control effects (on/off) can then be simulated with a resistive connection between this C1821 voltage and R1839.

If you are getting main power switch function, then this is likely a waste of time.

Unpowered circuit - opens , shorts and junction drops......

RL

On a sunny day (Mon, 25 Jan 2016 11:37:30 -0000 (UTC)) it happened Cursitor Doom wrote in :

Should not be a problem, I have used that method in very noisy environments, caps and power circuits are low impedance.

scope on AC, look at the waveform oon the fileter elcos relative to ground. If you see spikes or other strange things replace. There IS no other reliable way.

I expect the rails to be NOISY as hell, tbh. Not convinced that the bulk storage caps will be totally responsible for it, though. I'll report back later...

R1832 IIRC. If I've identified that wrong I'll correct it shortly.

I did think about hooking up a pulse generator to the chopper's b/e junction after disconnecting the normal pulse train. 50kHz@ 50% duty cycle and see what that produces. Not sure I'd have enough base drive, though.

In-circuit be ok? The board's a bit delicate!

Don't.

You already know that the power switch and drive function. Without it there'd be no 'hissing', noise or limiting circuitry heat.

Hissing is a ~white noise spectrum audible mechanical reaction to random protection. The lower and louder the pitch, the more severe the fault and the closer its proximity to the power switch. Luckily the switch is isolated by the resonant transformer - pointing to secondary-side misbehavior.

When things start working, everything but unstable operation goes above audible frequencies.

In-circuit.

If you find a short where there shouldn't be one, then you might start lifting leads to track it down.

RL

It's NOT R1832; my apologies.

Useful to know, thanks.

All the 12 diodes in the "function secondary rectifiers and ripple elimination" section just checked out fine. But elsewhere I've come across one that's reading 0.011V both ways round. Gonna have to lift it...

Jan, elsewhere you suggested I scope line to ground. There isn't a "ground" in the usual sense as I'm using an isolation transformer for safety's sake. So I scoped directly across the two big caps each one separately. It looks really clean. I was surprised. One cap has 150V across it; the other 158V, but other than that small anomaly not really anything amiss at all.

Oh, no point, it's in parallel with a low value resistor and an inductor so the reading is totally unsurprising. Resistance readings "Looking back into the output" and checking for anything odd has shown nothing unusual at all. I wouldn't be surprised if it's just a buzzing cap somewhere.

On a sunny day (Mon, 25 Jan 2016 16:02:24 -0000 (UTC)) it happened Cursitor Doom wrote in :

OK, that is good, but for for the record, I asked you what was wrong with it, and you said 'to much noise'. If that is the _only_ thing, then now check the ripple on all those filter caps. (secondary side). All this assumes that scope is working. If it is working no need to check diodes etc.

Great explanation!

joe

My Tek DPO2024 runs Linux, and it's slow and buggy. Our $50K LeCroy runs Windows; it's basically an ugly XP computer with some A/Ds inside, but it does some very impressive things, like display eye diagrams without a trigger.

Whatever's inside the Rigols (RTOS? Bare metal?) is fast and apparently bug-free.

Hardly anyone makes analog scopes any more, and the old ones are dying. So, get used to digital.

It's *trying* to work. There are low duty cycle pulses coming regularly from the pcm chip at around the expected working frequency - but they're swamped by noise. Same story at the b/e junction of the chopper transistor. You can't even *see* the pwm pulses without using the scope's HF reject filtering! I'm guessing it's this noise that's triggering the chopper, causing the unit to run hot and making that transformer sing. That's my best guess anyway, although I'm totally switcher noobie.

Sorry, no can do. Stickin' with boat anchors here.

Agreed! We've had some very useful insights here from a number of people in fact.

Triggered or repetitive sweep? AC or DC coupling?

(just kidding)