Late 1970s SMPS Design Query

Thank you, legg. One big problem is that I only get about 15 seconds to probe this board when powered up before R1814 (20ohms) starts to get too hot. It would burn out if left on for much longer than that. It's making fault tracing on this one a real PITA having to keep stopping so it can cool down again. It has a large amount of noisy voltage across it and is trying to dissipate about 5x it's rated power. Not sure if this is a separate issue or just another clue to the same one.

ditto.

nah....no PICs.......unless they're smoking downstream somewhere......;)

RL

On a sunny day (Tue, 26 Jan 2016 15:29:45 -0000 (UTC)) it happened Cursitor Doom wrote in :

Yes that happens.

That driver transistor V1812 gets its supply initially from that zener via the BAW52 diode , then after that via R1839 from the output transformer rectifed by V1816 and V1817, and filtered by C1821. The ripple on C1821 should be low, worth checking anyways.

The decoupling by C1817 is not that great, not sure that side of driver transformer is clean. So yes, there may be funny waveforms there. The base of the BU208 will / make a nice on / off from it I am sure.

On a sunny day (Tue, 26 Jan 2016 10:33:36 -0500) it happened legg wrote in :

That was not the point, you want the _whole_ load removed. Decoupling caps and other things on the other boards may als overload the circuit.

On a sunny day (Tue, 26 Jan 2016 15:38:53 -0000 (UTC)) it happened Cursitor Doom wrote in :

Do you have a capacitance meter to check the tuning caps X1806 and C1807 ? I mean the 30 nF ones.

On a sunny day (Tue, 26 Jan 2016 10:45:16 -0500) it happened legg wrote in :

I remember in the seventies going 3 weeks on an Ampex training course in Germany for ONE video machine and came back with 11 kg books (weighted at airport) full of circuit diagrams. Most just logic, and some power. No 'processors'. And mechanics.

But then again AVR1 was a beast... Was always kaput too. :-)

There are people who set out to 'design something' and never ever look at what has been done already. And it is the same with software actually, and that is why there is such a mess.

The scopes in were Tek actually.

I'm not so sure. It really looks fuggin' 'orrible. And remember this is a current-hungry old-time BJT we're feeding. Perhaps I'll try to post some photos of these signals, if not today then tomorrow.

How about pulling the primary feed to the tranny off where it meets those

What is the voltage on pin3?

RL

Still getting a *terrific* amount of noise across the b/e junction of the chopper transistor; hard to see how that device can possibly function correctly. Somewhere after those pwm pulses leave the chip, it all gets very messy indeed.

Looking at the rework on the solder side (image tag ending'a8_o'), it appears that some attempt has been made to resecure contact in the input power train by linking some of the solder joints to the bulk storage electrolytic assembly. Or is this the normal interconnection?

A cut track and link are also present, using the same wiring materials. A link on the diode bridge lead does not look original.

In the comp side image of this area(tag ending 'd8_0'), one of the resonant caps (30n1 G) appears to be darkened at its internal schoopage. If this is not just an artifact of the photograph or if the discoloration is not on the external surface, then this part should be replaced, just from visual inspection.

Both of the original BY208-1000 rectifiers appear to have been previously replaced by a GI part with the number 4 in its ident. What is the part number of the replacement being used? This may be a good oportunity to lift one end of V1808, in order to ensure the Vf and basic properties of parts in this circuit.

These parts should probably not be replaced by generic slow or even generic ultrafast rectifiers. The original parts are still available, though not from Philips/NXP.

RL

Then the supply is doing it's job under no load conditions.

Review notes on images posted elsewhere from visual inspection. If no issues from visual, return to test with loads attached and recheck output winding waveforms.

RL

Can you also please ensure that the original switch ident is BU208 from a proper vendor? Just loosen and remove, then reattach the heatsink. Do not desolder.

If this part is installed in wave-solderable socket/pins, make sure that pin/socket contacts are clean.

RL

Thanks for that. Yes, the chopper is the originally spec'd one as per that number. I'll have to check if it's been socketed or not; all the small signal transistors are, fortunately. I'll address the points in your other post either tonight or tomorrow. thanks again.

On a sunny day (Tue, 26 Jan 2016 16:20:00 -0000 (UTC)) it happened Cursitor Doom wrote in :

So then you would need some calculation (for that inductor value). I was thinking that if the tuning is incorrect (HV caps was as a stated before a common problem) then the transformer is a low impedance at the 20 kHz, and that could, apart from a winding short or flash-over, also explain the large peaks that heat that resistor, as the energy must go somewhere.

WOW what a sentence..!

On a sunny day (Tue, 26 Jan 2016 16:18:00 -0000 (UTC)) it happened Cursitor Doom wrote in :

Remember you can drive the base a bit negative relative to the emitter, this is just to remove the charge there. That may look bad (negative part waveform), but it will switch OK. To measure it you need a diff probe I guess.

Would that value matter? After all, the purpose of that tuning is to transfer maximum energy to the secondary side which is undesirable during this test, anyway.

Yes, very impressive. ;-) I got impulsive and broke the connection that feeds the primary. I just left it like that; didn't bridge it with any inductor and turned back on. The transformer hissing was gone (not surprising!) but the chopper b/e junction was *still* a terrible mess, which tends to suggest there's nothing wrong with the transformer. Unfortunately, further testing was curtailed at that point by the dramatically rapid heating of the 3k3 resistor R1804. I'm sure I'll see why this would have happened when I examine the schematic again in a while.