HP 5300B Overcurrent

[Dekatron]

Have you checked the 1Meg resistor and 4.7nF capacitor (I think they are named R8 & C10)? You could also try to increase the smoothing capacitor C6, perhaps doubling its value, and see if it works better.

You can also try a 1Meg resistor and 4.7nF (or lower capacitance) capacitor directly across the zener CR10 to see if it lowers the spikes and protects CR10.

If CR10 is the only component affected and everything else runs perfectly across the range of load that the power supply sees you could increase the zener value as long as it stays below the necessary range to protect the transistor. You could also try a modern TVS-diode at the same voltage as they are usually beefier and withstands overload better.

[PJL]

I think we are back to the question of why it fails on AC but is OK when powered by the external PSU.

[Dekatron]

Originally Posted by PJL

I think we are back to the question of why it fails on AC but is OK when powered by the external PSU.

I agree. The components I mentioned above will affect the AC situation much more than the DC situation (the 1Meg/4.7nF are only used with AC as they are switched out of the circuit in DC as the power switch disconnects them.

I’ve built quite a few blocking oscillators, like this design uses, but I’ve never used a zener diode in the position like here as I’ve used beefy transistors/mosfets that would withstand the peak voltages involved.

[PJL]

I assume the transformer is properly fitted to the board now?

28V DC did not cause the Zener to fail so it is a bit of a mystery why it should fail on AC. Not sure the 4.7nF across the Zener is a good idea as it could kill the driver transistor.

Might be worth going back to the variac and monitor the voltage across C6 with the scope whilst bringing up the AC voltage slowly.

[frsimen]

Oh, that's not what I was hoping to hear. You only checked operation with the DC up to 26V. David (Factory) carried out tests at 28V too.

Could you report your measurements with more detail and measured values, rather than using descriptive terms, as it's not easy to judge how good or bad the situation is otherwise.

PJL has already covered the main things to check, you may have a problem with the mains transformer and PJL's checks will allow you to determine if that is the case. Check the mains selector switch position is correct too. The transformer you've repaired has to be mounted properly on the PCB, do do check again that you have that right. Check the components suggested by Dekatron, although I don't think the resistor will make a huge difference here.

The low output voltage on the -5V could be an excessive load on that rail, which has yet to be found. The simplest thing to do is to remove the module and link pins 25 and 50. If the -5V is still low when running from the 26V DC, there's probably nothing wrong with the module but there may be something wrong on what's left.

Paula

[Uncle Bulgaria]

A new day, a new test: less venting, more data.

No module attached. Contacts bridged on A1J1. T2 soldered in place (board holes have been repaired with eyelets, continuity measures OK) with continuity measured through winding at repaired pin 4. 28VDC applied across C6.

First picture shows waveform at collector of A1Q17 (now a BD241C) with front panel power knob switched off, second with it on. Both at x10 1V/div., 10µs/div. I make that a peak of 55V at a frequency of 31.25kHz.

This all looks very similar to factory's results, except I now have a secondary peak on the pulse at the end of the waveform.

Voltage rail readings (the same whatever the position of the on/off knob):
3.5V = 3.75V
5V = 4.98V
-5V = -4.54V
17V = 17.34V
-17V = -17.43V

It's been sitting like this with the waveform absolutely solid on the 'scope for the last 20 minutes while I've been collecting these data and writing this post.

The mains transformer has been set at 230V all the time.

[frsimen]

Your measured voltages are about right, the -5V is perhaps a little low, it shouldn't be below -4.75V but the +5V is towards the low end of the expected range (the nominal value should be 5.2V, 4.95V minimum) so might account for it.

The waveform doesn't show any sign of damaging the Zener diode when running from 28V and, as you say, the peaks are much the same as those reported by Factory. I've no idea what is causing the odd glitch just after Q17 starts conducting. Check around the base in case there is a bad joint, maybe.

None of this explains why feeding from the mains causes the Zener diode to burn up. Try PJL's test, powering from the mains via a variac. Increase the voltage on the variac until you get the same waveform as you have reported in your last post. Note the voltage output of the variac at that point and report back with your findings.

Paula

[Dekatron]

Originally Posted by frsimen

I've no idea what is causing the odd glitch just after Q17 starts conducting. Paula

If you hook up the second channel on your oscilloscope to the base of Q17, while keeping the first channel as it is now, you should be able to see the waveform switching Q17 on and off. Then you can determine if the short spike at the end is because of Q17 switching or if it is due to the magnetic flux in the transformers being dumped in the circuit.

[PJL]

Have we done any checks on the mains transformer primary to secondary insulation?

[Uncle Bulgaria]

No, not yet PJL.

Dekatron - the 'hooky' piece of one of my probes got melted when Q17 failed. Thus I only have one left at the moment so can't monitor two channels at once and take a picture unless I can get a photographic assistant!

Watching the waveform on Q17's collector as I brought the variac up showed a very messy load of spikes. I thought perhaps it needed a little more to get it going, but turned it up too high and there goes CR10.

I thought I'd go back to basics and measure what's happening on C6 with CR10 replaced. There seems to be a problem! Here is the trace with contacts bridged, board A2 in place and 50% mains in on the variac (123V at the moment). The noise is 100Hz, and at 2V/div (x10) it can be seen that no wonder CR10 is feeling overwhelmed!

Comparatively, the second picture shows the ripple on C6 when board A2 is removed, so the switcher is not running. The scale is different to show the ripple is 2.5V on 26V at 100Hz.

I remember a previous post suggested replacing C6. Next step?

[frsimen]

Are you saying that you are measuring 26V across C6 with the 123v applied? If so, there is something badly wrong with the mains transformer, or someone has wired the
110/230V switch incorrectly.

Before changing anything, can you measure the AC voltage at the input to the bridge rectifier with A2 unplugged but the link pins 25-50 link in place. Looking at the photo you posted, the voltage to measure is where the two orange wires connect to the PCB. You need to measure the voltage that you are applying at the mains input socket too. If you are using your variac, don't let the voltage at the input to the bridge rectifier go above 28V otherwise C6 will be damaged. The voltage at the input to the bridge should stay below 22V if the transformer is connected correctly and not faulty. As the load is quite light when A2 is unplugged. the transformer output will be a little higher than when it is loaded. The loaded voltage will be no more than 20V I expect.

Paula

[Uncle Bulgaria]

Rats! In checking to make sure I had written the division scale of the second picture correctly, I failed to mention that the test was made with 100% of mains when A2 was unplugged, as I knew CR10 would not be affected. The C6 voltage shown in picture 2 is with full mains but no switching occurring.

[frsimen]

That's good, I would have been more concerned had that been what you were seeing with 123V applied!

There are a couple of things which seem odd with the second trace.

1) The voltage on the left hand side of the trace is about 3 volts lower than at on the right hand side. You need to use a longer timebase setting, or wait a little longer until the voltage reaches its final value. It may be that there is some sort of longer period sawtooth waveform present, so do check with a longer timebase base setting to see if that's what is happening.

2) Are those 2V lumps 10ms apart? There should only be a very small current (perhaps 30uA plus any leakage in Q17, CR9 and C23) being drawn by the circuit with A2 removed and I wouldn't expect to see any ripple at all once C6 has charged up. Have you connected your 'scope probes directly across C6?

It's possible that you are seeing some sort of earth current upsetting the reading. It's not a fault but can cause odd results when you try making measurements. Try connecting the scope probe and its earth clip both to the same point on the negative side of C6 if you still have the odd display when connecting directly to C6. If you still see the same pips on the display, it's likely to be due to earth currents flowing via the 'scope lead.

Note that the mains earth connection in the 5300B doesn't connect to the negative terminal of C6.

Paula

[Uncle Bulgaria]

Thank you for yet another considered comment, Paula.

1) The trace is off-axis, and the things on top of the 'scope prevent me from getting behind it to tweak the 'trace rotation' adjustment. This one was an emergency present from a very generous member after my old 'scope blew up, and the horizontal alignment is slightly variable.

2)Yes. The probe was clipped to the positive end of C6, and the ground clip to the negative end. This is connected to the chassis, and the mains earth is also directly connected to the chassis, so there is continuity between C6 negative and mains earth.

[frsimen]

Ah, that's not really a problem, thankfully.

You could try disconnecting the ground clip of the probe (make sure it can't touch anything when you do this). That might tidy up the display that you are seeing.

Paula

[Uncle Bulgaria]

With A2 disconnected, the 'lumps' are the same with the probe connected to C6+, and the ground lead connected or disconnected.

[frsimen]

One final check is to connect the probe to the negative side of C6. I'm not sure at the moment whether there is something odd about C6 or you're just measuring some odd potential difference between your oscilloscope and the 5300B.

If you see the lumps with the probe on the negative side of C6 too, that suggests the lumps are nothing to do with C6. It that's the case, some oscilloscopes have an earth terminal which you can connect to the chassis of the 5300B with a short length of thick wire. That will reduce or possibly eliminate the problem.

The original C6 is a large value tantalum capacitor. I suspect that you will need to be careful in choosing a suitable electrolytic capacitor to replace it, should it prove to be faulty.

Paula

[Uncle Bulgaria]

OK - hooking the probe to the negative side of C6 shows a flat line with no lumps, and no DC deflection. Full mains applied and no A2 board. Reading is the same with the earth clip attached or not.

I checked the manual after you mentioned C6 was a tantalum, as I was surprised as I'd assumed it was aluminium electrolytic! I've never seen a tantalum like it before. It's the big blue Mallory in the photograph.

[frsimen]

C6 is beginning to look like it could be the culprit, quite what is wrong with to cause that odd looking waveform is beyond me.

Leaving the set up as it is, that is no A2 board in place, you could try a 1000uF electrolytic in place of C6 to see if the odd waveform is still present. If it is, try replacing the diodes in the bridge rectifier. If the odd waveform has gone C6 is faulty. You'll need advice from someone more knowledgeable as to a suitable replacement. Maybe Radio Wrangler could suggest something?

Paula

[frsimen]

The mystery waveform's source is solved, I think. I ran an LTSpice simulation of the circuit as you have been testing with A2 removed. If the ESR of C6 is very high (100s of ohms), you get the waveform that you have measured. That would also explain the horrible waveform that is present when A2 is back in place.

Based on the simulation, I think changing C6 is the best option. Finding a suitable axial leaded capacitor may prove to be challenging.

Paula