Solartron Vari-Pack SRS153S renovation

[trobbins]

I picked up a cheap, clean, complete power supply last year, and recently got to renovate it. I came across a few interesting issues, so since I've seen this supply and the 151 come up in various threads I thought I'd make a post on it.

The 0-350Vdc, up to 100mAdc, general purpose supply has a product datasheet and schematic attached, and a Service Manual is available on-line.

The supply uses a Parmeko Neptune 6000/73 series transformer and /55 series choke, with M8091/CV4044 half-wave rectifiers and two parallel 5B/254M operating in triode-mode for output voltage setting.

Given the supply uses a choke-input filter, I measured the choke inductance at 20Vrms at dc current from 5mA to 130mA, with inductance drooping from 30H to 22H, so it's not a swinging choke. 20Vac is a lot lower than this choke-input application with its 625-0-625V secondary, so I was mindful that the actual inductance presented could be somewhat higher.

The default critical inductance was not enough to avoid rectified voltage rise at no external load, with the service manual showing a graph of the voltage rise below about 10mA external load.

My aim was to check how well a parallel R-C ripple trap across the choke would work. A ripple trap not only has the benefit of reducing the output rms ripple voltage, but it also effectively increases the inductance to allow a lower level of load current to achieve choke-input regulation, and also suppresses transient voltages at the choke input should they occur.

A parallel capacitance of 82nF was needed if a likely higher inductance of 35-40H was present. I was able to step through capacitance values from 50 to 78nF, and then 100nF, and voltage regulation and ripple voltage measurements indicated a capacitance value between 78nF and 100nF was likely optimum (although I ended up using 100nF for now).

With the ripple trap in place, the internal loading effectively achieves critical inductance conditions with no significant rise in no-load filtered voltage, and the ripple level was substantially lower.

Unfortunately I could only test to 50mA loading as the top-cap connection to one of the 5B/254M was stuck, and my efforts to rock/pull it off failed miserably. You don't think you apply too much force, until you do, and then you castigate yourself for not giving it a spray with contact cleaner/lube.

All capacitors initially tested in good condition up to their rated voltage, although one of the TCC Visconal 1uF 600Vdc was lightly leaking, so was replaced after I checked it used mineral oil and not PCB.

I did make a few other subtle changes. I don't like mains wiring bundled and laced with secondary side wiring, so I separated them. I added 3x series 1N4007 in series with each valve rectifier, as a form of belts-n-braces protection. I also reconfigured SW1-B so that C4-C5 now discharge when power is off and no output tubes are in.

The front panel meter has a 600V FS, and all parts etc seem fine to have higher voltage than 350Vdc output at rated current, so I may try and modify the triode-mode operation to pentode-mode to lower the minimum voltage drop across the output valves. I think an isolated low power screen voltage can be easily generated from doubler rectifying the S-R heater and using a small ebay switchmode inverter to generate circa 200-300Vdc.

Ciao, Tim

[Silicon]

Thanks for the documents.

When they state a DC resistance of 700 Ohms, are they referring to the output resistance?

[trobbins]

Yes, output resistance, so nominal 7V change for 10mA step.

I had made three spot measurements that align: circa 610R at 32mA, 420R at 26mA, 560R at 15mA, and increasing to 1900R at 6mA, but with some error tolerance.

700R also aligns with known choke supply path series resistances of 130R for PT secondary half-winding, 170R for M8091 at 100mA, 10R for fuse, and 240R for choke DCR. Then there is the series pass stage.

I am just installing a HV shunt regulator ( https://dalmura.com.au/static/HV%20D...unt%20load.pdf ), with the aim of seeing how well that suppresses the choke supply voltage variation from circa 0 to 50mA of output current, as a simple way to improve output voltage regulation. Not that I have an application that needs regulation, but more for interest. Of course a more typical path to take would be to add some form of modern output voltage feedback and reference comparison and gain to control the series path stage using say an opamp and FET, but that would be for another day.

[trobbins]

I installed the shunt regulator and set it to regulate the raw supply from the choke-input filter, which it did at 542V with < 0.1V variation for external load variation from zero to about 35mA.

I am one valve down on the following series pass regulator, and measured somewhat similar output voltage variation with load change to before with no shunt regulator, so may put this effort on the shelf until I come across another 5B/254M.

[stuarth]

Just out of interest, how would a variable HT supply such as this Solartron, with a simple pass device to control the output voltage and no regulation, compare to a variable supply using a Variac into the HT transformer?

With the Variac, the low voltage windings on the HT transformer would not be useful, but I think the load regulation would be a bit better, the supply would be smaller and run much cooler, and there’s no series pass device to fail catastrophically and damage the load.

I’m currently considering various options for a variable HT supply, my least preferred option being to use silicon series pass devices, with or without regulation!

Stuart

[trobbins]

Yeh the SRS135 sort of needs sufficient heater volts for its rectifier diodes, series pass valves, and its front panel 6V3 CT supply. If the positive HT diodes were ss, and the series pass valves and the front panel heater weren't needed then a variac in front of the unit would provide a choke-input regulated variable HT. For that type of supply, the load regulation is pretty good if you have sufficient load current to keep the choke in continuous current. If you wanted consistent regulation down to zero external load then you can fit dissipative loading, or a HT shunt regulator like what I installed, and/or add ripple trap to enhance the choke's performance.

'Regulation' as per what is typically shown in valve datasheets for choke input filtering, does depend on the choke. I must admit the Parmeko P480 is not too shoddy at about 8-10 ohm (DCR) per Henry.

But you sort of have to justify (imho) whether a simple capacitor input filter type supply, with up-front variac, is not going to meet your needs.

[stuarth]

I have in my component stash a couple of small 0.7A Variacs and various HT transformers which I’m considering using to build variable HT supplies which would go down to zero volts with no range switching (I hate bench supplies with range switches or coarse/fine voltage controls), or current derating at low output voltage.

Expected use would be valve testing or powering valve equipment (including battery valve radios). I don’t feel the need for a regulated supply, but the ability to ramp up the output starting from zero, and no risk of the pass device going short circuit and over stressing the load, appeal.

I don’t really like the idea of semiconductor devices operating at hundreds of volts, and valve regulated supplies going down to zero volts get rather complicated. I did consider an unregulated valve (cathode follower) design, but I think I’ll go with the Variac based design, although the heater winding being effectively useless offends (slightly).

Stuart

[thermion]

Iv’e got one of those. Is that why i could never get it to go to zero volts off load?

[trobbins]

Yes, I'd anticipate that the zero volt setting (likely to be the reason for RV2) was made at some level of output load, so wouldn't accommodate a significant rise in raw positive DCV due to choke not being in 'continuous current mode' - ie. no external load. I haven't tried to assess the change in min output voltage setting versus loading so far - I'm hopefully getting a replacement 5B/254M in a few weeks, so can check that then.