HP 5300B Overcurrent

[Dekatron]

It's hard to say from your pictures what is happening exactly but it does look like the strange dip at the end of the triangular pulse corresponds to the end of the collector voltage on Q17, the other squiggle in front of the triangular pulse is harder to understand where it comes from though. The end of the triangular pulse is supposed to be a sharp cutoff so that might need to be checked further why it has that dip.

If possible check the capacitors on A2 and see if they are still within tolerance.

I found the specifications for the original transistor here: https://www.lintechcomponents.com/pr...18540487/28480 and it says "80.0 maximum collector to base voltage,dc and 0.8 maximum collector to emitter voltage,dc" & "7.00 amperes maximum collector current,dc" which is a little bit more than the BD241C you mentioned that you are using as it has a maximum current of 3A, it could be that the BD241C is not up to the task. Neither do we know if the BD241C is fast enough for this switching application.

I suggest that you try to find a better suited transistor for Q17 like a more modern MJE-transistor. Having a quick look at Mouser resulted in the MJE15028G even though it has a somewhat lower power rating and also the TIP41B which has a lower fT frequency but might still work here.

[Uncle Bulgaria]

Thank you, Dekatron! Your sleuthing is much appreciated, and your Googlefu admired.

Before I go about ordering more things I rummaged in a box and found some MJE15034G. My cross-referencing suggests they're similar to what you suggest, but perhaps you can tell me if they're similar enough. If not, I'll be back when I've replaced Q17 with one of your suggestions, and checked the A2 capacitors.

[Dekatron]

Thanks, trying my best to find hard to find data.

The MJE15034G has only 4A capability so it might be a bit weak, but you could try it to see if it works somewhat better as it is a more modern transistor.

[Uncle Bulgaria]

MJE15034G has replaced BD241C as Q17. Original axial tantalum C29 has been restored after testing good.

The waveform at Q17 appears identical, as far as I can compare with my photographs.

I've made comparisons of the Q17 Base and the Test Points marked on board A2. I'm not sure if they're useful as the waveforms should really be compared with the manual figures, but only if I had a 50-pin socket I could hard-wire to imitate the relevant test card, as detailed in the manual.

[Dekatron]

Could you please show base and collector signals of Q17 too in one photo and then a series of photos with collector of Q17 instead of base but with TP1-4 showing in the other trace just as with the first series of photos.

[PJL]

And are the output voltages OK now?

[Uncle Bulgaria]

PJL - Output voltages remain the same as they have been since CR10 stopped blowing (and when DC was applied across C6, as in post #86). -5V is still sitting at -4.5V.

Dekatron - your wish is my command.

I see there is an 'external trigger' socket on this oscilloscope. Would that help with clarity of traces, or am I going to have to add the 'scope to the repair pile as it's probably in need of a service?

[frsimen]

Any oscilloscope will struggle with this waveform. Try using 26V DC instead. That should reduce the blurriness of the display. I think the blur is caused by the ripple voltage on the supply when running from the mains. The PSU is altering it's frequency slightly as it compensates for the ripple.

Paula

[PJL]

I agree with Paula, the blur is due to the phase changing slightly and the short pulse is at the end of the discharge cycle and is going to carry minimal current. A 350V 4A power transistor, you just need to check it is not getting too hot when under load.

Hopefully the fix to the transformer has solved the intermittency problem.

So the questions now the PSU is up and running again are:
Why does it blow the Zener when on mains?
Why is the -5V rail low?

[Dekatron]

Thank you for the photos!

One more request, could you email me a high resolution photo of the A2 module, the forum compressed it a little bit to much, it looks like one resistor is missing part of its casing in the photo but it is to small to clearly see if that is the case.

If you have the possibility I'd check all of the capacitors on A2 to see if they are ok, especially the electrolytic and the three polyester ones, the mica one should be ok. These capacitors are used in the timing but also 120Hz suppression (0.033uF) and if any of these are poor the regulation will not work properly.

[Uncle Bulgaria]

Yes, thanks Paula & PJL - I remember your suggestion from post #108. I was having trouble getting it to run at all from DC from my Thurlby PL310, and the higher current Farnell unit that worked is now in storage for building work. I'll try again with the Thurlby to see if it improves the trace. It seems to have an added slant at the right end now and be more out of focus there, as if the beam isn't hitting the screen properly. Anyway, a problem for another day.

Dekatron - your profile isn't set up to allow e-mails, so either send one to me I can reply to, or a PM with your e-mail address. I previously checked the capacitors against the manual and didn't risk taking them out and stressing all the tracks even more, as I thought the types were reliable. However, as two threads and over a hundred posts have led to C6 being The Problem, I'll check the A2 capacitors out of circuit now!

PJL - it doesn't blow the zener on mains power since replacing C6 back in post #101, following frsimen's simulations showing high ESR was likely the culprit. The -5V rail is peculiar, and has been the same throughout all the troubleshooting. Perhaps these A2 capacitors will prove to be instrumental...

[frsimen]

I don't think there is a fault with your oscilloscope. You can prove it by checking against its internal calibrator. I expect you will find that the trace is well focussed across the whole screen. The blur you are seeing is purely a function of the small changes to the frequency of the power supply oscillator.

You will need to adjust the trace rotate on the back of the 'scope to cure the slant problem, but you are already aware of that!

Paula

[Uncle Bulgaria]

I've got it running from DC again, and it has made the oscilloscope trace much clearer.

The tantalum 7.5µF C5 on A2 was showing very high leakage (it was tested OK a few weeks ago) so has been replaced by a parallel electrolytic combination of approximately 7µF.

The polyester capacitors have zero leakage and measure very close to nominal capacitance. No change in the secondary spike on Q17, but it's easier to see the waveform now.

First picture is collector Q17 top and base bottom. Second picture is collector Q17 and vb1 of Q3. Third picture is what those waveforms ought to look like.

I know it seems to be working Paula, but this peculiar waveform is bugging me. Perhaps I'll fix it until it breaks again...

[PJL]

The trailing pulse is due to flyback but it carries very little power as the energy has already been transferred to the reservoir capacitor. The driver circuit detects the end of the discharge cycle and then starts a new conduction cycle but your replacement transistor is faster than the original so it is allowing it to flyback before the conduction cycle starts...I doubt it is doing any harm so ignore it

Is the instrument actually working as it will have suffered a few hiccups on the PSU rails during the repair.

[Uncle Bulgaria]

Originally Posted by PJL

Is the instrument actually working...

...ish. Hooking it up to the 5302A Universal Counter and turning it on (connected to the mains) the display lights up and sits there happily. Pressing in the 'Sample Rate' knob illuminates 'C', the decimal points come up and all the digits show '8' except for the leftmost one which shows 9 (dead segment, loose connection?). 'K' and 'Hz' remain illuminated and are so when it's idling.

Turning the 5302A knob to 'check' should show a 10MHz +/-1 count according to the manual, but the display is stubbornly at zero. Perhaps I need to move into troubleshooting the 5302A module now. Luckily Elektrotanya has the manual...

In this situation, the voltage rails (+/-17; +/-5; +3.5) show: +17.3; -18.2; +4.9; -4.4; +3.5.

[frsimen]

Check that the reference source is set to the internal clock, rather than external. The counter won't work properly otherwise! The clock source is on the main unit.

Paula.

[Uncle Bulgaria]

Good call, Paula. That little INT EXT switch has got flicked a number of times during all this work - it seems a bit loose. However, it's set to INT and no count is shown.

[frsimen]

I would check that there is a 10MHz clock reaching the counter part of the circuit before getting too involved elsewhere.

Paula

[factory]

Originally Posted by Uncle Bulgaria

...ish. Hooking it up to the 5302A Universal Counter and turning it on (connected to the mains) the display lights up and sits there happily. Pressing in the 'Sample Rate' knob illuminates 'C', the decimal points come up and all the digits show '8' except for the leftmost one which shows 9 (dead segment, loose connection?). 'K' and 'Hz' remain illuminated and are so when it's idling.

Turning the 5302A knob to 'check' should show a 10MHz +/-1 count according to the manual, but the display is stubbornly at zero. Perhaps I need to move into troubleshooting the 5302A module now. Luckily Elektrotanya has the manual...

In this situation, the voltage rails (+/-17; +/-5; +3.5) show: +17.3; -18.2; +4.9; -4.4; +3.5.

Pressing the sample-rate control tests the display, you should have all eights on the displays, five decimal points, an overflow light (upside down decimal point in MSD display) and the C. The displays are in sockets so it's possible to swap them & confirm a failed seven segment display, note some are fitted upside down.
As the displays are multiplexed, it's more likely the fault is a dead display segment.

The reference goes to the plug-in and back again via an OR gate (U15C) on the plug-in, to F2 on the mainboard, it can be switched off by certain functions via a NAND gate.
The module input should go to eventually go to F1.

To add to the confusion some plug-ins had a TXCO on them and another internal/external ref switch.

David

[Uncle Bulgaria]

Originally Posted by factory

Pressing the sample-rate control tests the display, you should have all eights on the displays, five decimal points, an overflow light (upside down decimal point in MSD display) and the C. The displays are in sockets so it's possible to swap them & confirm a failed seven segment display, note some are fitted upside down.
As the displays are multiplexed, it's more likely the fault is a dead display segment.

The reference goes to the plug-in and back again via an OR gate (U15C) on the plug-in, to F2 on the mainboard, it can be switched off by certain functions via a NAND gate.
The module input should go to eventually go to F1.

To add to the confusion some plug-ins had a TXCO on them and another internal/external ref switch.

David

Attached is a picture of the display 'at rest', and with the display test active. The modules are plugged together, and I see no second oscillator switch on the 5302A.

I have the TCXO and get a lot of 'scope hash instead of a 10MHz signal at pin 16 of the main interconnect. I wonder if the low -5V supply is affecting it.