Tektronix 576 curve tracer with periodic screen blanking

[Apfelmus]

Hello all,

I was digging out a Tek 576 curve tracer that did work correctly some 10 years ago before it went to the attic. I had some trouble restarting the unit. It finally did. After some cleaning and wiggling of all connectors and transistors it seems to work quite allright.

Except for one major fault: after some 10 minutes running the screen blanks, eventually reappears and blanks again after some seconds. This process repeats forever, without any distinctive noise or smell inside the unit.

I have the original service manual, good scans are readily available on the web. I attached an extract about the power supplies.

My diagnosis so far:

- power consumption is approx 80W when turning on, rising to 100W when the symptom appears, than lowering back to 80-90W, rising again etc.

- just before it happens, there is absolutely no sign of anything on the screen, i.e. no wobbling, increase or decrease in luminosity etc. The image on the scrren just disappears.

- checking the power supplies revealed that several voltages drop (100V and 12,5V in particular) when this happens, then slowly rise again. Its hard to tell which of the supplies actually drops *first*, as they all seem to be interlocked with each other. Meaning, the 100V supply has input from the 12,5V supply and vice versa!

- measuring current on the 100V line revealed a strong current increase before the screen blanks. Apparently, the current limiting device is shutting down the 100V supply.

- tracing the 100V line in the unit leaves the high voltage supply as major suspect. Step generator and horziontal/vertical amps also use 100V supply, but do not seem to malfunction. I disconnected the step generator, without changes to the symptom. I did not disconnect the horizontal/vertical amps as they are identical and indipendent, so if one of them has a fault there should be at least some sign on the screen.

- tracing the current in the HV supply, R850 and R851 both clearly show the increase in current. The oscillating frequency is around 28-30kHz, which seems correct. R850 gets very hot - I measured up to 180°C with an infrared thermometer. But there is no smell, and the resistor looks good.

- since the problem seems heat-related, I placed a fan in front of the HV supply, hoping to stop the symptom or at least to modify the durations. But this did not change anything.

- on the 248V line (C866), everything looks stable, I could not see any change in voltage during current increase.

- tracing current through Q866 via R868 shows very few current, and it does not change.

So now I'm wondering if
1. since the HV supply does not react on cooling, am I looking at the right place?
2. where the hell is this current increase going to? The only remaing current sink is either the transformer having a problem when getting warm, or somewhere in the picture tube circuitry, or the picture tube itself.

What I would do next:
- measuring the resistances, especially the high ones.
- mesure the HV (but I need a probe for that)

If anybody has another idea of where to look, I would be very pleased to hear about.

cheers
Martin

[n_r_muir]

Hi,

I would try isolating the CRT by removing the socket from the back of the tube base. If the periodic increase in current goes away, the problem is in the crt. If not, the the problem is elsewhere. What sort of temperature is the mains transformer operating at?

[Apfelmus]

I'll think about that. But the HV supply will go in an off-scale mode that maybe is not very representative.

Meanwhile I checked HV voltage directly, it remains constant throughout the cycles. But the voltage on the base of Q855 rises to about 3V. This voltage depends on the position of the Intensity control. When I turn Intensity full CW, I get approx 2.5V, fully CCW its -1.5V. When the fault appears, voltage goes beyond 2.5V.

Your question about the temperature of the mains transformer: after 15 minutes its getting slightly warm, thats all.

cheers
Martin

[maninashed]

If you have a manual, check any zeners in the hv regulation circuitry, also any tantalum decoupling capacitors, they have a habit of failing short in Tek gear.
Bill

[mhennessy]

I have a 576 that behaves in exactly the same way. It's been locked away in a cupboard for about 10 years now, as I have just been too busy to look at it - it's not a tool that I would use much. I haven't attempted to perform any analysis of the problem as yet, beyond assuming that I'd be starting with the power supply.

I know that this might not immediately help you, but I thought it was worth mentioning because it tells you that your fault might not be totally unique/obscure/esoteric/whatever.

The power supply is pretty conventional for its time -I'm very familiar with a simplified version that appears in a 'scope that I've done a lot of work on over the years (the Telequipment DM63). I've just skimmed the circuit description, and it's not the clearest explanation. Essentially, the -75V regulator is the "master" for most of the others - so this needs to be rock-solid and ripple-free.

Also, other rails provide voltages for other regulators in order to save costs - for an example, look at R780, which provides base current to the "pass transistors" Q787 that generates the +5V rail. Ordinarily, this resistor would be between the base and collector of Q787, but ripple on the incoming supply would find its way through to the output - to fix that you'd need to split the resistor in two and decouple the midpoint with a large-ish capacitor - or replace the resistor with a current source. But by simply feeding the resistor from a regulated hum-free rail, that problem goes away. Very neat.

Other examples of this are R810 and R796 - both of which are fed from the 100V rail when on the face of it, there is no need to do so. But as well as the ripple point, these resistors end up being high values to achieve the small currents needed, which increases the gain of the error amplifiers (Q810 and Q795 respectively) - gm.RL and all that . Or, to think of that another way, the collector loads are looking more like current sources.

Obviously, with all this interconnection, they've had to think carefully about what happens at start-up, and during overload protection. But once you understand these points, the design makes a lot more sense.

I think that my first move would be to work out if the main power supply is jumping or being pushed. Is the unit "cycling" because of an overload on one of the output rails, or is it unable to supply the required current? I suspect the former. Monitoring the voltage across each of the current-sense resistors is probably the best way. I guess I'd start with the 100V rail (R846), as that is where the HV generator draws the bulk of its power from. A DVM with a fast-response bar graph would probably be enough; otherwise a differential measurement with a 'scope would do. Or perhaps monitoring the emitter-collector voltage across Q837 - as that will sharply fall as the unit goes into current limit - might be easier?

EDIT: Having re-read your first post, it does sound like the problem is in the HV PSU

If problem is an overload caused by the HV supply, things get tricky. There are a lot of high voltage diodes and capacitors that could be breaking down. And the transformer itself could be suspect. But I note that it generates a 225V rail that is used elsewhere in the 'scope, so an overload on these could be the problem.

I'll watch this closely, as it will hopefully help with mine. Perhaps this thread will persuade me to make a start on mine

Mark

[Apfelmus]

Originally Posted by mhennessy

If problem is an overload caused by the HV supply, things get tricky. There are a lot of high voltage diodes and capacitors that could be breaking down. And the transformer itself could be suspect. But I note that it generates a 225V rail that is used elsewhere in the 'scope, so an overload on these could be the problem.

Thanks, Mark, for the insight. Indeed, I'm pretty sure the HV supply is drawing too much current. The increase is measurable from the very beginning, a steady increase that accelerates towards the end. It still takes some seconds between, say, the usual upper limit and the moment the current limiter comes on (approx. 25% above).

I could not find any fault on the 225V rail so far:

- monitoring the current through the series regulator revealed a very low number, suggesting the 225V current is not more than some mA. And it remained steady.

- Except in the HV supply itself, I only found the 225V in the output stage for the deflection plates. I have absolutely no sign of anomaly on the image on the CRT, so I cannot see what could be wrong there.

I noticed the HV transformer getting quite warm at the end, but not hot. Something like 50-60°C according to my thermometer. Wonder if this is OK.

I certainly will check the capacitors and diodes in the HV supply. Measuring while the Tek is running is tricky, and measuring while switched off may not reveal the problem.

What puzzles me is that although it seems to be a thermal problem, it does not react at all when I give the HV supply an extra cooling...

cheers
Martin

[John_BS]

Martin:

I had a similar (but not identical) problem on a 453 scope. Normal display etc on switch-on, but after 20 minutes or so it would fail. I traced it to the EHT oscillator taking progressively more current until the current limit kicked in.

In this instance there was a clear thermal correlation, in that if the 'scope was allowed to fully cool, it would take the same time to fail. If re-powered before cooling, the failure was much more rapid.

There are two likely candidates:
1. One of the diodes/caps on the EHT side
2. the transformer.

In my case it was the latter. As a first step, try disconnecting the two EHT secondaries and see if the 100V currrent ramps up or not.

John

[Leon Crampin]

Before going too far in, check C850 which decouples the supply for the HT oscillator. On several Tek 'scopes I've looked at, the component in this position has been a Ta electrolytic which is killed by the ripple current. A modern Al device makes a good replacement - fit the physically largest device you can, to limit temperature rise.

I used one of these extensively for component evaluation years ago. It was a good instrument, but not very reliable...

Leon.

[MotorBikeLes]

MHennesy replies in post 5 that circuit is similar to TQ DM63. It seems to have the same or similar Tek mfgd EHT transformer. Whilst I have not had trouble, I believe there have been problems with the transformers paxolin insulation breaking down near the HV pickoffs. Examine carefully for smell/ bulging of the paxolin.
Les.

[John_BS]

If it's the Paxolin on the transformer, then Maninashed is your expert regarding this failure! The breakdown happens between the lamina of the Paxolin, so is hidden from view.

John

[Apfelmus]

Thanks for all further advices.

So far I made the test proposed by Mark, unplugging the CRT. Not a difficult task. To my surprise, everything was nearly as before even without the load of the CRT (which I assumed should be the major part).

Looking at my thermometer (actually a thermal imager, a wonderful tool when debugging circuits) I noticed the current limiter always goes on when the HV transformer has reached 60°C. The temperature then settles somewhere at 65°C.

I also notices the HV diode D870 not getting warm at all. Once turned off I checked the diode, and its open circuit both ways, at least with my Fluke handheld. I wonder how the circuit can make the correct HV if that diode turns out to be broken...

I'll check that again with the CRT plugged in. I will also check C850 as advised by Leon. And I'll make a picture of the transformer, to check the "paxolin".

cheers
Martin

[John_BS]

Martin:

Do try disconnecting the secondaries: that's the only way to seperate the transformer from the rest of the EHT. Taking the CRT out of the equation is not that helpful. On my 'scope, the transfomer must have had some partial insulation break-down causing shorted turn(s): this leads to extra heat generated in the winding, and as the ferrite core gets hotter, its losses increase. Eventually the whole thing spirals out of control.

John

[Apfelmus]

Originally Posted by John_BS

Do try disconnecting the secondaries...

John, do you mean disconnecting pin 7 and/or 8, or 5/6 as well?

If 5/6 is open the control loop will maximise the current through Q851.

cheers
Martin

[John_BS]

OK, try just disconnecting pin 8, and if the current still ramps up after switch-on, disconnect both pin 8 and pin 5, remove Q855 and lift one end of R856, then repeat.

John

[Apfelmus]

Here is an update.

I tested the diode D870 with a higher voltage, it seems OK. Its probably several diodes in series, so the voltage drop in pass direction is something like 20V.

No measurable heating of the diode when running with everything connected.

I let the transformer run with open secondaries pin7/8, as suggested by John. It gave very distinctive whines, changing while heating up. Power consumptio of the whole unit remains the same, and the transformer heats up as fast as with the secondaries connected.

Attached is a picture of the transformer, and also an IR picture of it when the current limiter comes on. The parts with the brightes yellow are at 65°C.

If its a problem with the transformer, like temperature-dependent shorted windings, shoundn't the current increase appear rather suddenly and not as progressive?

cheers
Martin
Martin

[John_BS]

Martin:

I can't fully explain the mechanism for the thermal dependence, but in the case of the 453 transformer, this was the pattern of current consumption (from cold) leading to complete failure. Some ferrites change their magnetic properties as they get hot: the winding resistance clearly increases, and possibly the potting compound dielectric losses on your transformer??

John

[mhennessy]

Originally Posted by Apfelmus

If its a problem with the transformer, like temperature-dependent shorted windings, shoundn't the current increase appear rather suddenly and not as progressive?

Yes, I've wondered about this...

Could there be a problem with the control circuit? I've had problems with the Telequipment circuit - caused by high-value resistors (less likely here, as I'm sure Tek would have used HV types) and also by the transistors in the feedback loop. If the transistors are socketed, quickly checking them by substitution is easy.

If you haven't already, it might be interesting to monitor the 248V rail as the unit starts to trip. If the input current rises while the 248V stays constant, that would suggest that the loop is correctly trying to regulate in the face of something trying to change. If the 248V rail starts to rise just before it trips, perhaps that points to the regulator loop?

Always difficult separating out cause and effect

[John_BS]

I'm 90% certain that it's the transformer, sadly........

John

[MotorBikeLes]

For what it is worth, that transformer is quite different to the TQ ones to which I referred earlier.
Les.

[Apfelmus]

Originally Posted by mhennessy

If you haven't already, it might be interesting to monitor the 248V rail as the unit starts to trip.

OK, I did that:
when cold, intensity min: 254,3 V
when cold, intensity max: 259,8 V

when warming on normal intensity, voltage starts with 254,7V rising towardds 255,3V, then decaying to 254,7V just before shutdown.

Sadly (to quote John), that seems to validate the control loop, doesn't it?

BTW, I found C863 not soldered in place on one side (you may see it on the picture). I wonder why... I soldered it back in, no change.

Is there a way to directly determine the "aging" of the transformers isolation? What does the dielectric value has to do with it? I guessed dielectric is relevant for capacitors, not for inductors.

cheers
Martin