HP 5300B Overcurrent

[Uncle Bulgaria]

Paula, you're a genius. Would you mind sharing a screenshot or LTSpice file of what you modelled? I've had the programme installed for a long time but have never got to grips with it, I guess because my electrical education is only what I've gleaned since the titbits we had to do at A-level physics. I'd be interested to see what you did.

I've replaced C6 with a 1000µF 50V Dubilier axial I had. After reforming for half an hour it had a leakage of 12µA so I think it's OK. The original Mallory has a leakage of 0.6mA at 40V and an ESR of 150 ohms on my component tester.

The lumps on the waveform at the positive end of C6 had disappeared with A2 removed and full mains applied, so I replaced A2 and applied 50% mains. C6 showed wobbles rather than the high frequency spikes of before, the collector of Q17 had some switching, so I gingerly increased the applied voltage. Voila, the whine increased then died away, CR10 sees no great spikes and there's ~22V on C6 with gentle ripple.

Attaching the 5302A counter module, the display lights and the unit sits there with full mains applied.

Voltage rails are still a bit off, as before - the values are the same as in post #86. However, it is running and not blowing up, nor making peculiar noises. Dare I say this is fixed?!

[factory]

If it really is a tantalum, it should be noticeably heavier than a regular aluminium electrolytic, due to the materials used.

David

[mole42uk]

Originally Posted by Uncle Bulgaria

Attaching the 5302A counter module, the display lights and the unit sits there with full mains applied.

Voltage rails are still a bit off, as before - the values are the same as in post #86. However, it is running and not blowing up, nor making peculiar noises. Dare I say this is fixed?!

Hurrah! I am so delighted that you’ve succeeded. You can’t believe how delighted I am!!

[frsimen]

I'm pleased you have got to the bottom of the problem. The slightly out of spec output voltages don't seem to be causing a problem, so I wouldn't worry unduly about them.

Attached are some screen shots from the simulation. Note that the simulation is a pretty bare bones version of the circuit. The mains transformer has been replaced by a voltage source plus a series resistor. The 100k resistor represents the combination of the component leakages, the 1M resistor and the 'scope probe resistance. The simulation uses the transient simulation, starting to record at 3 seconds from simulation start. The delay is to allow time for the C6 to charge up. Recording much earlier shows the capacitor charging period, which looks quite a lot like the trace you posted earlier.

The screen shots aren't very clear but you can probably see what is going on.

Paula

[Uncle Bulgaria]

Originally Posted by factory

If it really is a tantalum, it should be noticeably heavier than a regular aluminium electrolytic, due to the materials used.

You're right - it is heavier, and also bigger. Here's a comparative picture of it next to its replacement. What might be the reason for using such a capacitor instead of a normal aluminium electrolytic? It's just a reservoir capacitor for the rectifier, isn't it?

On the subject of capacitors, I replaced A1C29 (3.3µF tantalum coupling Q17 and Q16) some time back. Not having the right value, I have a couple of small electrolytics in parallel. Is its value crucial, or can I use a 4.7µF, for example? I can only find a passing reference to it in the theory of operation.

Thank you for the screenshots, Paula. It's good to see how an expert uses these tools! I take it R2 represents the ESR of C6, rather than having it defined within the C6 parameters. How did you come to the figure of 100k for R3? I guess as A2 is disconnected, T1 and T2 are only seeing DC from the rectifier so can be discounted. I see you've left out the 1M R8 in parallel with the 4700pF C5. The manual says both components help with fast regulator response - why did you include the one and not the other, when the regulator isn't running?

[frsimen]

I don't consider myself to be anywhere near an expert in using LTSpice. It is a very useful tool once the basics have been cracked. I'm sure it can do a lot more than I use it for.

In the simulation, the 150 ohm resistor does represent the ESR of C6. I could have put it in as one of the capacitor's parameters, but it may not have been quite so obvious, I thought.

The 100k resistor was chosen as the parallel combination of R8, the 'scope probe and a resistor to represent the leakage of C6 (I assumed that it had a fairly modest leakage), Q17 and the base bias network. Pure resistors in parallel will behave in the same way as a single resistor of their combined value in a simple simulation like this. If stray inductance and capacitance were to be included the result would be a little different, but at 100Hz that would be of little consequence.

Did the replacement C6 tidy up the odd little oscillation on the collector of Q17 when it turns on?

Paula

[Uncle Bulgaria]

I see, so it's possible to adjust individual components to simulate ESR etc. I think my hurdle is probably the assumption part! I would have very little idea of what to include and what to leave out, but I understand the Kirchoff principle.

The Q17 collector trace now looks as follows (mains applied, no module attached). It's very fuzzy (it's not the focussing), and looks even more impenetrable (picture 2) when the module is attached. Peaking at less than 50V, apparently. However, the oscillation is still visible. The signal on the base is more obviously a small square wave, but also fuzzy.

[frsimen]

The beauty of a simulation is that you can easily try different circuit arrangements to see if the result is the same.

When running from the mains, there will be a ripple voltage across C6. The regulator circuit will try to control the switching period to give a ripple free output on the 5V line. In so doing, it makes the waveform rather difficult for the oscilloscope to display multiple pulses cleanly. With a digital storage oscillosope, you can select single shot to get a clear display. Some oscilloscopes have a hold off control and adjusting that might help, if it is fitted.

It not too easy to be certain but there may be some disturbance on the rising edge of the base drive waveform. Try using a faster timebase setting, so that only the first part of the pulse is displayed, trigger on the rising edge of the base drive waveform. That should help to give a clearer view of what is happening there. You're only interested in the first part of a pulse, so the blurriness will be diminished with the faster timebase speed. If you have two probes, you can display the collector waveform at the same time.

Paula

[Uncle Bulgaria]

Yes, it seems very handy. I'll try making a corresponding circuit if I have a similar conundrum! I attempted to use it to model a possible valve amplifier at one point, but it was beyond me. However, since I've now made two circuits in KiCAD, I might be at a better stage.

Here's the base pulse with no module connected (picture 1). I can't get the timebase any faster (it's at 5µs) as then only the falling edge is shown, with the rising edge off the screen and trace. There's a latching button with "+" for out and "-" for in, but operating that makes it worse. I would get out my more modern Tektronix but it's inaccessible with the current building work.

Picture 2 is the fastest timebase and clearest picture I can get with the collector waveform. At 1V/div (x10) it is <50Vpk. The latter part of the waveform (picture 3) still demonstrates the secondary 'blip'.

Do you have an opinion on my capacitor value in post #105?

[frsimen]

The photos are much clearer and show that there is nothing wrong with the base drive waveform. I think you can leave the capacitors as they are, as the waveform looks fine. Fitting a larger value probably won't cause upset the performance, but I'm can't say for certain.

The blip in the collector voltage is caused by something else. In an earlier post, Dekatron asked for the base and collector waveforms. Perhaps Dekatron can suggest what is causing this and whether it is something that needs further attention.

Paula

[Dekatron]

Having separate traces is not enough as we might miss what is going on at the base when the spike is shown on the collector. We don't know if the spike on the collector is there for every driving pulse on the base or not. It must be checked with two probes connected at the same time with two traces shown and also possibly checked over some time of the sweep to see if it is always there or just on some pulses.

But if we assume that the spike is always there it could be there because a resonant circuit forms from the components involved when th etransistor is switched off - the driving transistor parasitic/stray capacitances (possibly also from the zener diode) and the transformer inductance which produces the pulse. There can also be other parasitic/stray capacitances and inductances involved which creates the spike. It could also be there as a "topping-up" extra pulse from the control logic topping up the energy in the T2 transformer as it has been checked by the current transformer T1 as necessary to keep regulation.

So until we get two simultaneous traces from the base and the collector that has been checked over a wide portion of the sweep we really can't say what produces the spike.

[factory]

I'm guessing you missed the part where the probe tip melted, that is why we haven't got both traces at the same time, ATM.

David

[Dekatron]

Originally Posted by factory

I'm guessing you missed the part where the probe tip melted, that is why we haven't got both traces at the same time, ATM.

David

No I didn't, he commented on my post about that, I was just explaining why two separate traces aren't good enough here.

[Uncle Bulgaria]

I lashed up a couple of crocodile clips which enabled me to get the traces shown. The vertical scales are different between each channel, but I guess you're more interested in the timebase alignment. The second picture is a faster sweep with the same vertical scales as picture one.

[Dekatron]

It's not easy to see what is going on as the photos are somewhat blurry.

But there are some things that stand out - they are not that easy to spot since the forum software compresses photos and they are of low resolution to start with.

In the upper trace there can be seen a positive pulse which I have tried to show inside the red rectangle, the first edited photo shows the rectangle only and the second the pulse is also drawn with a red line. This pulse has not triggered properly so I am not sure when it happens and what effect it has.

When it comes to the negative spike on the collector it can be seen that there is an effect on the base drive at that moment too, a short dimple, again look at the two photos at the left double arrow line inside the red boxes.

When it comes to the end of the pulse on the collector it can be seen that it corresponds to the end of the base drive as they correspond to each other in form - the red double arrow at the right show their connection.

The extra pulse could be due to the A2-unit sending a regulating pulse to Q17. The only way to check this is to probe the A2 unit, possibly at the base of Q2 - that trace has a test point #3 marked which corresponds to one of the traces in the document. Since there are a few capacitors in the A2-unit whose value is part of the regulation it might be a good idea to check if they are ok.

[MotorBikeLes]

Dekatron, you could be reading too much from those traces. The Crotec is a low bandwidth scope and may not be properly adjusted for fast rise pulses. Many of those effects could just be the scope's limitations, and expect to see a pulse on one trace having an effect on the second trace, even on a flat line.
There obviously is something odd with that fast dip and return towards the end of the top trace's positive "flat line", but I have no idea what it is.Its shape may be partially due to the scope's limitations, but more than that, I have no idea.
Les.

[Uncle Bulgaria]

Thank you Dekatron, and MotorBikeLes for questioning.

I'm sorry for the blurriness of the traces - it's the 'scope rather than the camera. With the mains applied the traces are much fuzzier whatever I do with the focus and triggering than the much sharper lines I was getting with DC applied across C6 (see post nos. 58, 68 etc.).

Post #68 shows the trace I was getting with the old Q17 with DC applied - a single pulse, just with the excessive spike on the rising part of the pulse, and sharper than the ones I've been getting with mains. I think frsimen had an explanation for this to do with coupling with the 'scope.

Perhaps you've spotted why I'm now getting a second short pulse, Dekatron. That shows experience! I think in the first of my pictures in #114 it's possible to see the little dip in each of the base pulses, which correspond to the start of the second peak in the collector pulse.

I'm rearranging my apology for an office/bench to an even smaller corner of another room as we need to dig up the floor where I was before. As soon as my 'scope's up and I've digested my Christmas fare, I'll probe the test point #3 and look around A2. I have gone over A2 with a magnifying glass, but there are a few capacitors that could feasibly have reacted poorly to all my poking about...

[Dekatron]

Is it possible to put back the original Q17 and compare the traces?

I am not worried about the leading oscillation on the trace in post #68, that is normal for switching pulses.

The strange dip however is something completely different and it shouldn't be there in my opinion. It doesn't really look like it is due to a proper signal from the A2 unit unless any of the capacitors has become poor on the A2 unit, it could however be due to a transistor that isn't working as the original one does.

@MotorBikeLes - I agree, the scope is not a fast one but it did at least show some corresponding parts of the traces unless they are artifacts of the scope itself.

[Uncle Bulgaria]

Alas, the original Q17 burned up as the cause of the melted probe tip.

I wasn't referring to the initial leading oscillation spike in #68, rather that it was one continuous pulse instead of the 'strange dip' that's now making that single pulse into one long and one short.

All the A2 capacitors are original.

[Uncle Bulgaria]

I've done a bit of probing around as suggested by Dekatron.

Picture 1: Test Point 3, base of Q2. Some ringing with a just discernible wiggle at the end of the pulse. I wondered if this corresponded with the wiggle seen at the base of Q17.

Picture 2: TP2. Upper trace is emitter of Q3, lower is base of Q17. I think the Crotech needs some adjustment itself, as one side of the screen is always out of focus. Anyway, it can be seen that the wiggle on the base corresponds with the more obvious wiggle on Q3.

Could this simply be down to improper filtering from a borderline A2 capacitor? I don't recognise any on the board as being of an obvious failing type, but then I thought C6 was a normal electrolytic...