Systron-Donner M107 Precision DC voltage Source - Teardown video

[intabits]

My teardown of a Systron-Donner M107 Precision DC voltage Source.



I bought this unit not working, but had some success a few years ago in repairing it, in this thread:-

https://www.eevblog.com/forum/repair/seeking-help-in-troubleshooting-a-systron-donner-m107-precision-dc-voltage-sourc/msg1363041/#msg1363041

But it was short lived, and now I'm finally getting around to trying to repair it again.
 
It will be interesting to see if it's the same fault as before, but hopefully it will just be some dodgy soldering I did last time, that has gone bad.

I've made a teardown video, and when I get some repair success, I'll make a follow-up.

[intabits]

After fixing several faults, I've had some success in repairing this unit.

The 1000V range still does not work, but on the other ranges I'm getting 3-4 digits of agreement with my HP3456A voltmeter. Calibration is almost certainly needed, but I want to do some more measurements with more meters to get a better idea of the inaccuracy.

And I now have a much better understanding of how the feedback system works, and have a plan to determine where the fault lies with the 1000V range. Unfortunately I'm pretty sure it will be in the HV amplifier board, whose operation I do not yet understand. (I may be asking some dumb questions about that here...)

Repair video:-

[intabits]

Success at last!

First fault fixed was a few years ago. 3 bad FETs were replaced and the 1V, 10V & 100V ranges then worked. But there was still a problem with the 1000v range.
The unit worked for only a few days before it failed again.

Only recently was the project revisited, where the next fix was replacing a failed 7404 inverter. A loose flying lead from a trimpot was also fixed (this only affected calibration on the 1V range). Also some dicky keyswitches were replaced. But the 1000v range was still bad.

And finally I found the 1000v range problem yesterday, a trimpot was open on one side.

A new trimpot and more keyswitches are on order. A calibration will bring this unit back to full operation, except for any possible issues with the GPIB interface, which has not been looked at at all.

Repair video:-

[intabits]

I'm not sure if no responses indicates no interest, but in case there is, here's the final installment (for now at least).

The failed trimpot that affected the 1000v range has been replaced. This involved drilling out rivets to migrate its mounting frame to the replacement pot.

10 more front panel keyswitches have been replaced. I should have done all 25 up front, but the 5 remaining work OK. I should replace them as well at some stage, since they are rather "mushy".

To enable the front panel to drop away from the frame enough to work on it easily, some wires were replaced or extended. (I will appreciate this when I replace those last 5 keyswitches).
- Main power switch was a bit sloppy, so it was replaced, with longer wires, and an inline connector to enable the front panel to clear easily the frame.
- All the wiring from the analog board to the front panel binding posts, and the current limit pot, has been replaced with longer wiring. The new wires to the binding posts are only 300v rated, so they were inserted into PVC tubing to handle the 1000v output.

Unfortunately, a new fault has developed - the relays frequently dropped out, causing the 1v range and/or standby mode to be selected. This behavior now seems to have subsided considerably.

A crude calibration was done, and it's now good to about 4 or 5 digits. There's a "sidebar" step in the process that looks like it needs to be done as well. This is indicated if the zero adjustment on the 10v & 1000v ranges is hard to get right. It was not done because the range dropping fault described above was getting in the way.

Final video:-

[Electrole]

Congratulations with a job well done!
Let's hope no more errors pop up...

I wonder if you plan to do some measurements of the performance of the unit, once the unit behaves confidently?
The accuracy and the stability over time would without doubt be of interest to a large audience, but I am in fact more curious to know how the M107 peforms with regards to noise. How much RMS and/or peak-to-peak noise does it have in a certain bandwidth, like from 1 Hz to 10 Hz, and perhaps from 10 Hz to 20 MHz? The specifications provided by Systron Donner were not very concise. In the service manual they state max. 20 µV below 150 Hz in the 10 V range, but it's not clear what the lower cut-off frequency of that measurement is. Also, they state less than 100 µV RMS "wideband" noise in the 10 V range, though there's no definition of "wideband". In order to make comparisons with other references, it would be cool to have figures for the noise of the M107, but using a setup with a well-defined bandwidth so that we may compare...

[intabits]

Thanks for the kind words.

Given that the newest issue of ranges dropping out seems to be much more infrequent, I can have another go at calibration, including the extended "sidebar" steps. I've only got old HP345x voltmeters as references, none of which are calibrated, but I'll use the one (HP32456A) I trust most for the calibration.
There's quite a bit of drifting about as they warm up over a couple of hours, which I'll try to monitor and record after I've calibrated the M107 to the 3456.

And yes, the stability over long periods will be interesting to monitor. I can't leave them on continuously, but firing them up for a few hours weekly or monthly and seeing how consistent they are will be interesting.

As to noise, my knowledge of measuring and characterizing it, is at the layman's level.
Connect the output to an oscilloscope via AC coupling and wind up the gain?
I have no idea how to assess the frequency distribution of the noise.
Are some sort of filters needed at the input? Or use the CRO in FFT mode?
I'm not looking for answers to those questions here, I'll do some research myself, but if any useful links can be provided, that would be great.

but using a setup with a well-defined bandwidth so that we may compare...

Any suggestions on what that setup might look like would be most welcome.

[Electrole]

Hi,
One article to start with when it comes to measuring noise from references is https://www.analog.com/media/en/technical-documentation/application-notes/an124f.pdf
This is just for inspiration; In general, it's about:
1) AC-coupling, with a sufficiently low cut-off
2) Sufficient gain after the AC-coupling and before your measurement device, using an amplifier with sufficiently low noise
3) A filter to define the frequency span of the measurement
4) A suitable device used for the measurement: A digitizer with data processing applied in software, an RMS voltmeter, an oscilloscope, etc.
5) Compensating for the contribution of the noise of the stages in your measurement system
6) Making sure that the levels you measure are correct (= calibration)

Lots of fun ... :-)

[intabits]

Thanks for the link and the description. Looks like a good read, and it contains further references to explore.
As I suspected, this area can get very involved, but it's one that I've been wanting to get a better understanding of, and to experiment with.
It's in the queue...