Help me get started on a middle-aged HP 'scope power supply?

[willray]

Greetings BadCaps gurus,

I come seeking wisdom. It seems I've managed to go through life so far, getting lucky simply making guesses at bad components in power supplies based on appearance, smell, heat, discoloration etc, but I'm now trying to resurrect an end-of-the-CRT-era HP 'scope, and its power supply is proving that I'm but a dilettante.

I've done the obvious - pulled and tested, or tested in place the electrolytics, and even replaced common suspects that didn't test bad, and gone over the board for obvious signs of overheating and such. So far, none of the caps seem out of spec, and I can't find anything else suspicious.

Still, the power supply isn't working right.

On the bench, with no load, it'll come up and its on-board diagnostics say that it's producing normal output (or at least they do until it's been run for a while).

On the bench, _started_ with no load, a small load can then be attached, and it'll carry it.

Cold, it won't start with a load - even just a 12V fan.

Started with no load, a load attached and let run for a minute or so, it'll then start with the load -- for a while -- then eventually it'll start registering outputs out of spec and won't start again until it's rested.

Most of the output voltages look to fine on a DVM - I haven't stuck a scope on them yet to see what the ripple looks like. One's putting out zero V, but I don't know for sure that it's doesn't require feedback from the rest of the 'scope before it comes up.

Would anyone care to point me at the likely best next steps or likely suspects, or, aim me in the direction of a quick education that'd help me get there? By trade, I'm a Biophysicist, so the world of the EE isn't /completely/ foreign to me, and I can get a good ways along basic theory just from the physics, and I have a reasonably decent collection of diagnostic tools, but I don't _think_ in power supplies, and this one is giving me fits.

Many thanks for your time,
Will

[Khron]

Re: Help me get started on a middle-aged HP 'scope power supply?

A good first step would be trying to find the service manual for it. There's usually detailed schematics of just about everything, as well as adjusment procedures and whatnot.

[PeteS in CA]

Re: Help me get started on a middle-aged HP 'scope power supply?

Standard request for this forum ... pictures please?

Based entirely on your description and speculating a lot, if the Pulse Width Modulator is of the UC384x family, you may have a bad start-up cap. When powered on, the cap receives a trickle of charge until the voltage on it is high enough for the UC384x to try to power on (once it powers on the UC384x is powered by a winding from the main output transformer). If the start-up cap is to low in capacitance (unlikely) or its Equivalent Series Resistance has gotten too high (a usual-suspect-grade problem), that attempt to start up will take the voltage to the UC384x low enough to cause it to shut down again. If my speculation is correct, find the UC384x, and near it should be a small 35V or 50V electrolytic capacitor, usually 47uF-100uF. Its ESR may be too high for the UC384x to start up. This application can be a bit stressful, so replacing it with a Nichicon PW or United Chemicon LXZ series part (both are 105*C, low impedance/ESR parts), if possible, would be a good idea.

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

A good first step would be trying to find the service manual for it. There's usually detailed schematics of just about everything, as well as adjusment procedures and whatnot.

So far I'm not impressed with what I've found, though I'm doing better than the last time I looked for service docs for this one. My previous experience with HP service documentation was that they were sufficiently detailed for even me to be able to follow and fix most anything. In this case the diagnostics make the assumption that if the self-test of a single voltage is good, then you're done - less helpful than I expect from HP...

I've managed to find a scan of a service manual for a same-series model that appears to use the same power supply as my reluctant patient (mine's a 54200A, Agilent has docs for the 54201A - appears to be just a difference in max sampling frequency), but the trouble-shooting section for the power supply is long on theory and description, and short on testing procedures.

http://cp.literature.agilent.com/lit...4201-90902.pdf

It has gotten me far enough to know that the error that the supply throws, when it does recognize that something's wrong, is a flag on current limiting to the switching-supply transformer primary (measured on the downstream side) -- which leaves me now wondering whether the current limiter is insane, or if not, where the current's coming from.

Is it possible for an xformer drive transistor to be failed /nearly/ shorted? I'd expect them to let the magic smoke out, and also to be incapable of producing the correct output voltages under any circumstance, if they were shot - "almost, but not quite dead" is outside my envelope of experience for semiconductors...

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

Standard request for this forum ... pictures please?

I know - my apologies - I'm only near the 'scope occasionally (it's in my shop at the farm), and was hoping that the "stands up good voltages with no load, but can't make current" behavior was a completely obvious clue to those better educated than I.

So - I went and had a chat with the 'scope - it's a HP 54200A (a detail I'd managed to forget previously), and snapped a couple photos of the power supply board. I'm not sure how high a resolution that badcaps will let me post, but, for reference the PWM looks to be a TL594IN.

I don't quite grasp how they're using the overvoltage sensors (Motorola 1826-0468s that appear to cross-reference to MC3423P1 ) as over-current sensors for the switching xformer. There's no proper voltage divider, and there's a 10K resistor in series with the 1826-0468 input - seems quite a lot of opportunity for drift to muck up whatever calibration was intended there, though that's probably just my naiveté showing...

Let's see if I can make a couple pictures stick -- the whole board, and the PWM and control corner...

[PeteS in CA]

Re: Help me get started on a middle-aged HP 'scope power supply?

Holy cow! That thing's from the mid-80s! That OV crowbar chip was made by Signetics. I noticed a small round bridge rectifier made by Varo. Those pot cores are Ferroxcube 3C8 ferrite. Varo is long gone, and Signetics and Ferroxcube were bought by Phillips back in the late 80s or early 90s; Ferroxcube now looks to be a division of Yageo. Maybe the crowbar chip is tripping, triggering the OC protection which is a different part.

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

Holy cow! That thing's from the mid-80s!...

Ain't she a beauty? Who has the real estate to lay out a board like that anymore?

Simple circuits for simple minds like mine - my antiques and I go well together.

Maybe the crowbar chip is tripping, triggering the OC protection which is a different part.

That region of the board is where things are leading me at the moment. I don't know if the one that's tripping is tripping because there's really an overcurrent in the primary, or if it's just screwed up sensing - either way it's easy enough to test.

As an aside, they aren't using the crowbars as proper crowbars - the crowbar tied to the primary switching transformer isn't shorting the output and producing an overcurrent, it's being used to sense an overcurrent and shut down the PWM when one happens. They've attached it by stuffing a 60-turn winding around a leg of the switching transformer primary, feeding that to a bridge, and attaching the + output to the crowbar sense input.

The other is doing nothing but watching the thermal breaker on the big heat sink.

Neither perform a traditional "short the output" crowbar task, instead they feed into a comparator in the PWM and turn off the oscillator when their output goes high. The logic for using them in the circuit escapes me (especially the one that's just watching a switch), but I have to assume that there is elegance here that is lost on me at the moment...

Will

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

Just an update for anyone watching the idiot bumble around...

The scope is still not running. The overcurrent sensing circuitry is definitely seeing a brief voltage transient that's getting clipped off by the crowbar. Since the crowbar isn't directly sensing the current, and since it's shutting down the PWM rather than _causing_ the over current, I'm still not sure where the current surge is coming from.

With the bare power supply, it doesn't happen all the time, and it's so brief I haven't managed to capture a good look at it on the scope (I'm going to have to break down and take a good scope to go talk to it I guess). There's nothing drawing power though, and with no load the PWM is putting out tiny pulses, so I'm currently completely confused as to where the power is going. There's just not much downstream that could be leaking... All the components in the near vicinity of the crowbar seem to be well within spec, so it seems unlikely that it's just component drift deciding to catch the startup surge and call it an overcurrent event.

I do have one dead full-wave schottky rectifier on the 2.4v supply, but I don't think that's the source of the overcurrent problem. I think about the only place I've got left to look is the soft-start circuitry for the PWM. The service manual calls it out as existing to prevent a misinterpretation of the startup surge, so I suppose it could have gotten marginal and be letting the PWM come up too fast.

Fun fun.

Will

[Longbow]

Re: Help me get started on a middle-aged HP 'scope power supply?

Since the crowbar isn't directly sensing the current, and since it's shutting down the PWM rather than _causing_ the over current, I'm still not sure where the current surge is coming from.

From your description, it sounds like the input current IS being sensed.

I do have one dead full-wave schottky rectifier on the 2.4v supply, but I don't think that's the source of the overcurrent problem.

Yes it definitely could be. Also, there are very few cases where a crowbar circuit actually shunts the output of a supply. Your circuit is typical in that it shuts down the supply which will provide adequate protection without major drama.

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

From your description, it sounds like the input current IS being sensed.

Ah, I meant as in, the crowbar chip is a voltage sensor, so they're going through some effort to get it to trigger on current. With the (to me) strange non-voltage-divider input they're using to present the voltage to the crowbar, it's not obvious what current/duration profile actually produces the 2.6V input that's the actual sensed trigger.

I originally thought the fault might just be false triggering, but although I can't get a clean picture of the signal when it triggers, I'm now sure that the crowbar really is seeing something approaching 2.6V, so at least that variety of false triggering is not the problem. Can't rule out false triggering entirely though, as I've no information as to what the actual "too much current, for too long" value is, and we still could be getting to that 2.6V trigger voltage illegitimately.

Yes (the dead 2.4V shottky) definitely could be (the source of the over current).

Unless I'm missing something obvious, all that it's doing, is shorting a couple turns of the switching transformer secondary. Since that's only a minuscule percentage of the secondary, I'm not seeing how it'd induce much extra current in the primary?

Clearly, if my thinking's completely sideways on this, I require some edumacating :-)

Also, there are very few cases where a crowbar circuit actually shunts the output of a supply. Your circuit is typical in that it shuts down the supply which will provide adequate protection without major drama.

And see, I need education. Somehow the existence of crowbars has completely escaped my attention. The chip documentation certainly suggests that the intended use is to shunt a supplies' output on some overvoltage, and to let some other current-sensing mechanism shut down the supply on overcurrent.

Clearly just shutting down the PWM is the less dramatic solution, but since someone as unfamiliar as I might look at the crowbar docs and say "well duh, of course there's an overcurrent when the crowbar trips", I wanted to make it clear that the crowbar was sensing it, and not causing it.


It's not obvious to me why one would use a purpose-built voltage-sensing crowbar chip for overcurrent sensing, and another for watching a thermal breaker in one part of the circuit, and then use simple comparators for faults in other parts of the circuit, all tied to the same clock-defeat on the PWM.

Using the crowbars to debounce and latch the clock-defeat seems like one possible excuse, but I don't see any reason why the other faults don't need debounced/latched, and that all could have been accomplished with a single latch on the clock-defeat input.

It strikes me as highly unlikely that I'm thinking of anything that HP's engineers didn't though, so my guess is that there's rationale and elegance here that is currently lost on me.

Will

[Th3_uN1Qu3]

Re: Help me get started on a middle-aged HP 'scope power supply?

Unless I'm missing something obvious, all that it's doing, is shorting a couple turns of the switching transformer secondary. Since that's only a minuscule percentage of the secondary, I'm not seeing how it'd induce much extra current in the primary?

At least take that diode out and test without it. It's never a good idea to test a power supply with shorted rectifiers, no matter how insignificant you think they are.

[budm]

Re: Help me get started on a middle-aged HP 'scope power supply?

Since T5 is the main transformer (the Primary is driven by Q1. Q2) that produce 5 output voltages, if any one of the 5 is shorted out, it will shutdown the circuit.
-2.4 is produce by CR19 Dual Diode with Common Cathode.
The Primary current sensing is done by T4 current sensing transformer.
Pin3 of U5 is fed by OR GATE Diodes (CR16, CR17, CR9, and CR8) setup, if pin 3 goes HI it U5 will go into shutdown.

CR16 is for -2.4V Isense
CR17 is for +5V Isense
CR9 is for Primary Isense
CR8 is for Thermal sense.
If any one of the Anode of those Diode goes HI, that will trip the shutdown.
There are two more internal 2 or Diode in the U5 that also feed pin3.
So you need to monitor which one of the OR Gate Diode on the Anode side has gone HI to trip the shutdown.
I see alot of 1uF caps around that circuits which usually lost its value when it get old. I would check them out closely since they are used for filter around the inputs of U5.
U5 shutdowns by stopping the drive to the Q1, Q2.
The output of the MC3423 just goes HI when overcurrent is sensed, the output is not used for driving the SCR which will latch on until the DC voltage is removed, but the scope is not using that SCR Crowbar setup.

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

At least take that diode out and test without it. It's never a good idea to test a power supply with shorted rectifiers, no matter how insignificant you think they are.

Yeah - it's out now. I only realized that it was shorted right at the end of my last discussion with the power supply. Silly me assumed for some reason, that either the 2.4V regulator had gone, or that the rectifier failed open, and so it took a long time for me to get over to that corner of the board.

Replacement is winging its way this way from Digikey.

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

Since T5 is the main transformer (the Primary is driven by Q1. Q2) that produce 5 output voltages, if any one of the 5 is shorted out, it will shutdown the circuit.

I agree that that should be the case, but this isn't obvious to me from the design of the circuit. Of course, I'm nowhere near close to a professional, and I'm long on theory, and what practice I once had is decades decayed, so something not being obvious to me says more about what I know, than about the thing being discussed.

But... A dead short in CR19 just turns -2.4VRet and -2.4V into an AC circuit, which is going to be largely filtered towards 0V by LC T6 and C48. A significant circulating current is going to be induced in the T5_4->CR19->T5_5 loop, but there's nothing there to sense it, and it's just 2 turns on the secondary, so it shouldn't be changing the current in the primary much.

The 2.4V current limit, while I've more difficulty intuiting what's going on there, doesn't directly monitor the T5 2.4V secondary, and looks to only be able to trip on a short to ground on the -2.4V line itself.

Am I missing something here?

-2.4 is produce by CR19 Dual Diode with Common Cathode.
The Primary current sensing is done by T4 current sensing transformer.
Pin3 of U5 is fed by OR GATE Diodes (CR16, CR17, CR9, and CR8) setup, if pin 3 goes HI it U5 will go into shutdown.

CR16 is for -2.4V Isense
CR17 is for +5V Isense
CR9 is for Primary Isense
CR8 is for Thermal sense.
If any one of the Anode of those Diode goes HI, that will trip the shutdown.
There are two more internal 2 or Diode in the U5 that also feed pin3.
So you need to monitor which one of the OR Gate Diode on the Anode side has gone HI to trip the shutdown.

We know, at the moment, it's U4->CR9 that's shutting off the PWM, because we've got the diagnostic DS1 LED.

We don't currently know whether the -2.4V current limit might also be tripping (ever) because there aren't onboard diagnostics for it, and I haven't probed that circuit yet.

Haven't probed it yet, because with some reasonable frequency, the supply comes up to what it thinks are OK voltages, and provides stable power - at least enough for a fan. This, even though CR19 was still sitting there as a dead short - which implies that even if the -2.4V current limit is capable of seeing that short, it doesn't do it consistently.

The supply hasn't so-far shut down with any _reported_ fault other than (U4) primary overcurrent.

I see alot of 1uF caps around that circuits which usually lost its value when it get old. I would check them out closely since they are used for filter around the inputs of U5.

Everything I've checked so far, has been well within spec. Still plenty more to go, but notably everything associated with U4 appears to be good to about 1% of nominal value.

Since that R13,14,15,C19,U4 package is worse than I can solve in my head, I'm not sure what the design current/duration that's /supposed/ to trip U4 looks like, but given the quality of the components, I'm pretty sure that it's seeing it.

One of the places I haven't looked yet, is C21, which if it's lost capacity, this whole adventure may be the result of nothing more than the PWM coming up too fast. Somehow I'm not thinking that it'll be that easy, but hope springs eternal...

U5 shutdowns by stopping the drive to the Q1, Q2.
The output of the MC3423 just goes HI when overcurrent is sensed, the output is not used for driving the SCR which will latch on until the DC voltage is removed, but the scope is not using that SCR Crowbar setup.

Right. I'm puzzled why they used a pair of crowbars where they did, and simple comparators elsewhere, when everything feeds that PWM clock defeat.

[ben7]

Re: Help me get started on a middle-aged HP 'scope power supply?

Haven't probed it yet, because with some reasonable frequency, the supply comes up to what it thinks are OK voltages, and provides stable power - at least enough for a fan. This, even though CR19 was still sitting there as a dead short - which implies that even if the -2.4V current limit is capable of seeing that short, it doesn't do it consistently.

It is much more complex to have a short-circuit current monitoring circuit connected directly to the secondary of the transformer (AC), instead it is after the voltage is rectified and smoothed(DC). Usually, for lower currents (like under 100A), there is a shunt resistor - a precision, low-ohm resistor - in series with the load. If you know the value of the resistor, then you can tell how much current is flowing through it by the voltage that appears across it (Simply ohms law!).

Now, if the -2.4v rail doesn't have undervoltage sensing, then the only 'overcurrent' protection would be the current sensing on the primary side of the SMPS!

The filter capacitor will pass all the AC current (capacitors pass AC and block DC) from the shorted diode, and either the primary side overcurrent limit is reached, or the capacitor explodes. In this case, the filter capacitor could be bad, with high ESR - then there would not be enough current flowing to trip the primary overcurrent. Since the capacitor hasn't blown up, I suspect that it is dried out.

I don't know if I am going in the right direction - just trying to help

[ben7]

Re: Help me get started on a middle-aged HP 'scope power supply?

Actually, I just noticed on your picture of the board, there are no high voltage filter capacitors for the power supply input! I see spaces where they would go too. There are missing ones on the output side too. That could be the problem...

[budm]

Re: Help me get started on a middle-aged HP 'scope power supply?

"We know, at the moment, it's U4->CR9 that's shutting off the PWM, because we've got the diagnostic DS1 LED." that is how it detect the short due to shorted diode (same as shorting out one of the output winding of the transformer, the T4 current sensing transformer (primary I sense, it basically detect the total current draw) will see high current draw and trip the shutdown. The -2.4V Over current detection is for detecting how much curren is being drawn AFTER the output of the power supply, it dose not detect the short of the winding of the transformer due to shorted rectifiers.

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

"We know, at the moment, it's U4->CR9 that's shutting off the PWM, because we've got the diagnostic DS1 LED." that is how it detect the short due to shorted diode (same as shorting out one of the output winding of the transformer, the T4 current sensing transformer (primary I sense, it basically detect the total current draw) will see high current draw and trip the shutdown.

Right - except the primary overcurrent (U4->CR9) protection is only tripping sometimes, and when it does, it's dues to a brief excursion on the rectified T4 output exceeding 2.6V. The rest of the time (even though the diode was still shorted), it's only seeing a fraction of a volt coming from T4.

I certainly can't rule out that it's right on the border of tripping every time, and it's just my dumb luck that it comes up sometimes, and fails others though. That'd be about the average speed of things in my world...

Parts are in from Digikey, so we'll know if it's really that easy soon enough!
Will

[willray]

Re: Help me get started on a middle-aged HP 'scope power supply?

Actually, I just noticed on your picture of the board, there are no high voltage filter capacitors for the power supply input! I see spaces where they would go too. There are missing ones on the output side too. That could be the problem...

No, no, those are out because I had them out for testing. They're all quite happy and back where they belong now.

Will

[budm]

Re: Help me get started on a middle-aged HP 'scope power supply?

If it false triggering, that can be due to poor filtering of noise or spike in the sensing circuits.