Late 1970s SMPS Design Query

If you mean the villard cascade section, it's already been disconnected.

I reckon it's maybe dilectric breaking down in one of the caps, maybe one of the polyester ones, but it's just an uneducated guess. Not sure if probing around with a scope would assist in tracking it down; probably not enough separation between components, though.

Are those options handled automatically by the newer scopes then? Seriously I haven't a clue and freely admit to being well out of date on the subject.

Does the reading give a credible resistance for the expected component path?

With HT doubler disconnected, apply power slowly, look at the transformer low voltage secondary winding waveforms. Triggering on one, look at the ratio and shape of the others.

If the control chip on the primary thinks its regulating (3V7 on pin3), you may need to provide a load, by plugging into the scope using the low voltage connector harness. Do not apply meter leads directly to chip pins. A 10K resistor lead wrapped around a probe tip allows the second resistor lead to function safely as a pin prope.

RL

Have to admit to being a bit confused here. What is the purpose of the

On Monday, January 25, 2016 at 9:55:48 AM UTC-8, Cursitor Doom wrote: ...

How are you powering the unit and grounding the monitoring scope?

You do realize that since there is a bridge rectifier from the AC input the whole circuit is "live".

To test this you need to run it off an isolating transformer and then you can connect the scope ground to the negative output of the bridge rectifier.

kevin

One complication is that you can't really check a diode with a so-called "diode test" function. Diodes can fail in a way that reduces their breakdown voltage.

If you don't have an ESR meter to check the capacitors and a curve tracer (or equivalent homebrew test jig) to check the diodes, you may be better off preemptively replacing them all.

-- john, KE5FX

Not 10K exactly - that's not the point. 10K because it's still small enough not to seriously effect the low frequency DC meter accuracy of a 10M input Z. If it was a scope probe, I'd recommend 1K to do the same thing, until the circuit was running 'normally'.

Until things are working normally, it won't be obvious, but the meter input capacitance, combined with the probe lead lengths and loop area, can act as an antenna to inject noise into the test point of the high impdance analog error amplifier.

It might not be important in some situations, but in the control circuit of an operating SMPS, it can disturb operation sufficiently to make low voltage DC measurements meaningless and stable function impossible. Meter leads should normally be twisted to reduce unneccessary loop area. Scope grounds located locally.

(EMI generated locally by the SMPS under abnormal operating conditions can also be sufficient to temporarily knock out the measurement and display of some cheaper digital meters. I've also seen it upset the display and change the settings on some digital scopes....)

RL

I am running this off an isolation transformer; no worries, m8.

Can't say I'm too surprised.

Yes, well I have an ESR meter but I'm told that they're not infallible, either. Maybe homebrew test jigs which subject parts to higher voltages are the way to go...

I'm going to have to carry this out tomorrow as it's late here now, but just wondering what this check is intended to confirm/eliminate? Are you suspicious of the condition of the main transformer? If not, what exactly are you getting at here? Anyway, enough for today already. :)

It's looking for abnormal loading on one winding, or output winding/solder joint faults. This will also give you some idea of what the primary waveform looks like, in reflection, without having to use any specific HV technique.

Although the long term waveform may be chaotic, triggering to a higher amplitude peak that repeats at an audible frequency, should give visible traces for amplitude and shape comparisons.

If there's no issue, and the chaotic condition results in a quasistable DC voltage on the regulator chip's pin3, then you'll need to connect the low voltage output connectors to get any further in troubleshooting done. These things aren't always designed to be nice under no load conditions, in dedicated equipment use, where no load conditions do not occur, though a resonant or flyback circuit can more tame than others in this respect. There are ~ no output bleeders.

If/when the load is attached, you'd repeat the output winding exercise.

RL

The trick is: usually when they die, it's a small piece melted straight through from P to N. The rest of the diode behaves normally, but now it's got this resistive "short" across the terminals. If it measures the same

Obviously this doesn't always work in circuit, where such resistances might be present already.

This goes for pretty much all semiconductors, by the way. A common MOSFET failure is gate breakdown, which shorts the gate to the channel in one spot. If your gate driver supplies enough DC to drive that short, the transistor will go on working perfectly fine!

Tim

On a sunny day (Mon, 25 Jan 2016 18:56:00 -0000 (UTC)) it happened Cursitor Doom wrote in :

On the transformer side?

Transformer internal flash over would do too, I have seen that in TVs (quite common). In fact I did it myself on a coil I wound, just a few month ago. Posted about it here.

Well, let's put it this way. The two little contacts you can see poking out of the top of the large transformer? Those are the feed to the Villard so I'd say it's been disconnected, yes. I have only 1500V to avoid now. :)

That's my real fear. Everything else is replaceable/fixable.

Well I obviously needed that clarification! Thanks, I'll report back later today on my findings.

In fact I should really focus on testing that large transformer first and foremost, because if it's toast, the whole psu is junk and tinkering around with caps and semis is just wasted time.

On a sunny day (Tue, 26 Jan 2016 08:42:00 -0000 (UTC)) it happened Cursitor Doom wrote in :

If this is designed by a TV designer, then it likely is just about 1 turn per volt. Some old horizontal output transformer, remove coils, start winding. But that indeed comes close to a re-design.

You reckon it comes apart? That might be fun - and instructional.

Actually the Tandy one examined in detail in AoE 2nd Edition has some secondary windings of only 2 turns per winding to give 5V outs.