Heathkit PS-4 exhibits snarky behavior

[thorpej]

I picked up a Heathkit PS-4 (same circuit as IP-32) regulated power supply recently to serve as a secondary supply for the @kward Special I'm also putting together. It featured nearly-all original components (including Daystrom-branded GE 6L6GCs) and sort-of-worked; Voltage won't zero-out, voltage adjustments take a fair bit of time to take effect, and voltage starts to oscillate if you try to drop the B+ adjustment pot too quickly.

The original 6BH6 control amplifier tested OK, but I went ahead and swapped it with a known-NOS tube just in case. All of the caps were replaced save one that had been replaced with a Sprague Atom fairly recently. The original 6L6GCs tested weak, so I replaced them with some pre-loved 6L6GCs that previously served as output finals in my Silvertone 1484 but still tested strong-and-close on my Hickok 600A. The 0A2s seem to be perfectly good.

I replaced a couple of out-of-spec resistors, and calibrated the voltmeter movement against my Fluke 115 in both the B+ and C- settings; the original precision resistors are dead on.

Unfortunately, some of the bad behavior still exists, although it's better than it was to begin with. I'm mainly concerned about the responsiveness of the B+ control, and the oscillation.

Poorly-shot iPhone video here: https://www.dropbox.com/s/ma31xo9thzoc2v5/IMG_4884.MOV

I suspect there's a problem in the bleeder network maybe. Or maybe this is to be expected without a "real" load. But the calibration instructions in the manual don't mention adding a load of any sort. Tried hooking up a 13K 10W resistor, supply sags quite a bit (won't go above 300V on the meter) and still oscillates when the voltage is dropped.

Schematic here: http://www.rsp-italy.it/Electronics...Kits/Heathkit PS-4 power supply schematic.pdf

Anyone have suggestions for next steps in the restore / repair process?

 

[gadget73]

With my PS-3 it did stupid things when the output pass tubes were badly matched. That uses a 1619 but its not a grossly different circuit. I just swapped a few through until I came up with two that it liked.

Also worth checking the solder work. My PS-3 had a broken cathode connection on one of the pass tubes so it was only working on one. That plus the usual off value resistors, etc that comes with old gear restoration work.

 

[s-petersen]

I found the the resistor(R-7) between the negative supply and positive (almost 700 volts across it) had drifted very high.it should be replaced by 2 500K series resistors to meet the voltage spec on most resistors. My unit didn't jump like yours,but wouldn't go below 40 volts or so, possibly C7 is leaking on yours as well

 

[thorpej]

I found the the resistor(R-7) between the negative supply and positive (almost 700 volts across it) had drifted very high.it should be replaced by 2 500K series resistors to meet the voltage spec on most resistors. My unit didn't jump like yours,but wouldn't go below 40 volts or so, possibly C7 is leaking on yours as well

Ha, yah, that's a good point. I'll take a look at that one tonight.

 

[thorpej]

I found the the resistor(R-7) between the negative supply and positive (almost 700 volts across it) had drifted very high.it should be replaced by 2 500K series resistors to meet the voltage spec on most resistors. My unit didn't jump like yours,but wouldn't go below 40 volts or so, possibly C7 is leaking on yours as well

Hm, so it turns out that R7 (1M that connects the 6L6 plate to the 6BH6 screen) is one of the ones I had replaced already. It forms a voltage divider with R8 (150K). Now, the B+ is a little high (630V rather than 600V, due to the higher mains voltage), but the voltage at the 6BH6 screen is WAY off ... -120V rather than -57V (the cathode is rock steady at -150V), which leads me to believe that R8 has possibly also drifted. Anyway, if Vg2k of the 6BH6 is only 30V rather than the proscribed 93V, then the gain of the error amplifier is going to be way off.

I didn't get a chance to try the two-510Ks-in-series to replace R7 last night, but I'll should be able to get to that today and take a look at R8 as well.

 

[thorpej]

As far as the B+ being too high, it would be extremely inconvenient to constantly run the supply of a variac, and there's not a convenient winding available to buck the B+ winding. I suppose an inexpensive and relatively simple option would be to bolt on a Triad F7-120 (115VAC -> 120VAC @ 500mA) and run it backwards, but there's no space inside the case to mount it, really, unless I bolt it to the removable part (which might be a perfectly reasonable option). That would also get the 6L6 screen voltage back to where it's supposed to be (it's currently running a little high, but it's not obvious to me that would negatively impact the circuit that much).

Another option would be to add a Zener(-string)-regulated power MOSFET under the chassis to clamp the B+ at 600V. But I'd probably want to make two of those in order to fix-up the screen voltage as well.

 

[gadget73]

how much is that extra 30v really going to affect it though? Mine doesn't seem overly bothered by it.

 

[thorpej]

Possibly not at all. I'll use the variac until I get it stable, and then see if it remains stable without.

 

[kward]

The responsiveness of the B+ control may have to do with that 20 uF cap strapped across the output. Look to see if it's unresponsive on turning down the B+ voltage only, or if it happens when both turning down and turning up. I suspect that cap can charge up a lot faster than it can charge down. Notice the charge down path--its either going to be through the connected load (if there is a load connected), or through the adjust circuitry of the supply itself if there is no load connected--same issue I faced on my home-built bench supply.

The spikes in voltage when turning down the supply may very well be oscillation as you noted. Have you scoped the output yet under those conditions? Since the supply is essentially a feedback amplifier, it may be necessary to tune the feedback loop in a similar way as you would tune the HF response on an audio amp with global feedback, that is, to install a step response filter. I'd have to think a bit more about where it would be installed if it's needed. Perhaps across R6, so that it shaves down the error amp gain at higher frequencies.

Also it would be interesting to check when a resistive load is connected, does the supply produce that voltage spike (oscillation?) also when the B+ control is turned down?

===

 

[kward]

One other thought. Could the voltage adjust pot have dirty contacts?

 

[thorpej]

The responsiveness of the B+ control may have to do with that 20 uF cap strapped across the output. Look to see if it's unresponsive on turning down the B+ voltage only, or if it happens when both turning down and turning up. I suspect that cap can charge up a lot faster than it can charge down. Notice the charge down path--its either going to be through the connected load (if there is a load connected), or through the adjust circuitry of the supply itself if there is no load connected--same issue I faced on my home-built bench supply.

The spikes in voltage when turning down the supply may very well be oscillation as you noted. Have you scoped the output yet under those conditions? Since the supply is essentially a feedback amplifier, it may be necessary to tune the feedback loop in a similar way as you would tune the HF response on an audio amp with global feedback, that is, to install a step response filter. I'd have to think a bit more about where it would be installed if it's needed. Perhaps across R6, so that it shaves down the error amp gain at higher frequencies.

Also it would be interesting to check when a resistive load is connected, does the supply produce that voltage spike (oscillation?) also when the B+ control is turned down?

===

That 20uF cap is actually the one cap I didn't replace because it was previously replaced with what appears to be a Sprague Atom (tell-tale blue jacket); the orientation of the cap is such that it's hard to read the label. I think I will go ahead and just replace that one anyway with a known-good JJ 20uF cap.

The main lack of responsiveness is on the way down, yes. However, it seemingly takes a long time to go from 350V to 400V as well. I wonder if that cap across the output is just leaky.

I tried putting a 13K 10W resistor across the output... that would only be about a 30mA draw @ 400V, but I wasn't able to turn the B+ all the way up to 400V -- I suspect that might be a due to the issue with the error amplifier screen voltage being so out of whack.

 

[s-petersen]

It shouldn't be necessary to change any values, because they work fine for most people, mine is stable as a rock

 

[thorpej]

Finally had some time to play with this a little more. First of all, C6 (the one cap I didn't replace) was completely messed up. A previous owner not only used a higher capacitance value (47uF instead of the original 20uF), they used the incorrect voltage rating (400V instead of 450V). I pulled the cap, tested it on my IT-11, and it was leaky as sh*t at 250V. Replaced it with a 20uF @ 500V JJ.

I also verified R8 tested within tolerance (it came in at 147.5K vs the spec'd 150K), and replaced R7 with 3 330K resistors in-series so that each resistor only had ~220V across it.

With those changes, I was able to get the voltage at pin 6 of the 6BH6 to right around the -57V specified, but I'll be damned if it takes a LONG time for the unit to settle in there (> 15 minutes). It behaves moderately better, but still can't come up to 400V with a 30mA load attached. *grumble*

 

[s-petersen]

I would check the string of 27k bleeder resistors, My PS4 settles in less than a few seconds.
There is an adjustment for 400v , I actually set mine to allow 450v to form caps at full voltage.

 

[thorpej]

I would check the string of 27k bleeder resistors, My PS4 settles in less than a few seconds.
There is an adjustment for 400v , I actually set mine to allow 450v to form caps at full voltage.

I checked them, and they seem fine. But I might just replace them in any case.

Yes, I have run the calibration procedure with no load. Perhaps I should calibrate with the load attached? (But the manual doesn't mention needing a load during calibration.)

I suspect there's something else I've missed.

 

[s-petersen]

It is a kit, there is a possibility of an error. My PS4 does not need a load to maintain voltage.

 

[gadget73]

My PS-3 doesn't need a load either. Mine has a vacuum rectifier so it may take a little bit longer to warm up and settle than the PS-4 but within 2 minutes I have adjustable stable voltage within the supply's current limits. The first minute it tends to jump wander as the tubes begin to conduct.

 

[thorpej]

I'm beginning to wonder if there's a problem with the original rectifier diodes. I may pop them out of their holders and jumper in some 1N4007s as a test.

 

[s-petersen]

That could be tested by connecting a probe to the 600v going to the plates of the 6L6, then putting it under load.

 

[thorpej]

That could be tested by connecting a probe to the 600v going to the plates of the 6L6, then putting it under load.

Yah, I guess I'd be looking for the output of the voltage doubler to sag way down under load.