• Please review our updated Terms and Rules here

PDP-8/A power supply problem returns...

D

DrCharles

Experienced Member
Joined
Sep 2, 2014
Messages
101
Location
West Plains, MO
The power supply problem on my 8/A has come back worse than ever, after two "got it fixed for sure this time" episodes, and now I have actually tracked it down. (Power OK going off and on, which halts the system).

It appears that the secondary of the CVT (ferroresonant transformer) is providing a marginal voltage, which during the failure is causing the linear regulator to drop out at a 120 Hz rate, thus signaling the Power OK circuit to stop everything. Replacing the suspected 110k uf filter cap with a newer 180k uf unit seemed to have solved the problem for a few hrs of run time. Now I know the increased ripple wasn't the filter cap failing, but the transformer voltage dropping instead.

There is no visible, audible or olfactory evidence of a bad connection, and at 20 amps on the +5 line I would expect some indication. Line voltage is 121 volts.

The interesting thing is the secondary waveform on the transformer, driving a full-wave rectifier into a large filter cap.
It swings in a somewhat distorted sine, as expected from a CVT, from -10V up to only about +7 volts where it flattens out for the half-cycle driving that arm of the FW rect. The other side of the secondary looks exactly the same, but 180 deg out of phase of course. That's pretty poor voltage regulation for a CVT!

Although the linear regulator has a PNP pass transistor, there's obviously not enough headroom any more once you factor in the rectifier, ripple and wiring drop to maintain +5 nominal. So there are only two potential (pun intended) problems... I2R drop from a bad connection (which as mentioned should be pretty hot), or a problem with the ferroresonant circuit itself.

It's colder in my computer room now with fall well under way, and this failure mode is now present sometimes for minutes at a time, and more frequently than ever. Something is gradually worsening and appears to be provoked by the lower ambient temperature.

Wondering what would happen to the AC output if the oil-filled cap on the resonant tank were failing?
These caps don't fail often but a few people have reported failures due to their advanced age (the cap, not the people) and the high circulating current.

Either way I have to pull the heavy chassis from the rack to reach the backplane or remove the transformer box which contains the CVT & its resonating cap. :p

Any thoughts?
 
Did some more quick reading:

http://dictionary.ieee.org/definitions/449-1998/fig4.gif
http://www.oltronix.nl/en/ferroresonant-principle

The waveform should indeed be a nearly-square wave unless the transformer is being grossly overloaded.

The resonating winding has continuity and lots of inductance, and the 6 uf, 660V resonating cap is just about exactly 6 uf (60+ second time constant, discharging into my HP DVM).

Not only that, the damn thing is performing perfectly on the bench with a resistive load of 17 amps at 17 VAC!
At low line voltages the tops of the square waves become more rounded under load, as expected. Regulation is very good, about 17.4V (indicated, don't know if the HP is true RMS) unloaded and about 17.1 under a 1 ohm load.

It does have a very odd habit, when the load is suddenly connected and disconnected, of entering some other resonant mode which I can't easily describe except to call it "motorboating" in a bimodal waveform, very audibly in the core too, until the line voltage is increased or decreased, then the waveform returns to a nearly-square wave. But that never happens under real world operating conditions so I will just file that for later...

And, of course, now it's working in the real hardware. Sigh.

I think there may just have been poor contact on the 1/4" quick-disconnect tabs that they run the 20+ amps through from the low-voltage secondary through... not a great design. On very close inspection, the CT connection at the transformer pigtail looks just a little bit "heated" (browning of the yellow crimp sleeve, slight shortening of the wire insulation).
I crimped it harder and soldered it for good measure.

The waveform at the input to the rectifiers (on the regulator board) is now nearly square and symmetric :)
I may have fixed it this time. But I've said that before... :confused:

This design (in the interest of using linear and not switching PS while still minimizing heat) really has marginal headroom for any slight increase in voltage drop with age. Even a volt at 20 amps may not get hot enough to make it smell obviously burning (20 watts) but this design has no headroom at all for that extra 0.05 ohm.

Will post back after I let it burn in for a while (hopefully not literally!)
 
I have seen lots of the caps in DEC ferroresonant power supplies fail. They are motor-run caps, and are easy to find and cost just a few dollars. Motor-run caps are different from motor-start caps, so get the right type.
 
m_thompson said:
I have seen lots of the caps in DEC ferroresonant power supplies fail. They are motor-run caps, and are easy to find and cost just a few dollars. Motor-run caps are different from motor-start caps, so get the right type.
Click to expand...

And check the voltage. Most motor run are usually 370 VAC. The DEC ferroresonant I replaced have been 660 VAC. And they have been filled with PCB oil. A good reason to replace them anyway.
 
My cap appears to be fine. It seems the problem was just a tiny bit of extra resistance in the secondary center-tap connector, which naturally would cause the same problem on every half-cycle as each full-wave rectifier conducts in turn.

Re: PCB, why would that be a good reason to replace the cap? Unless the cap is failed or leaking, the PCB is safe in the steel can. If you remove it, the toxicity problem just moves to a different location. I have a few GE "Pyranol" caps from the 1950's or even earlier and they are intact. No need to call the EPA with their hazmat suits!

Same for lead and asbestos. It is far safer to contain it in-place (paint, coatings, encapsulation) than to kick up a cloud of toxic or carcinogenic dust that will inevitably settle on everything, spreading contamination over a far wider area.
 
DrCharles said:
Re: PCB, why would that be a good reason to replace the cap? Unless the cap is failed or leaking, the PCB is safe in the steel can. If you remove it, the toxicity problem just moves to a different location. I have a few GE "Pyranol" caps from the 1950's or even earlier and they are intact. No need to call the EPA with their hazmat suits!

Same for lead and asbestos. It is far safer to contain it in-place (paint, coatings, encapsulation) than to kick up a cloud of toxic or carcinogenic dust that will inevitably settle on everything, spreading contamination over a far wider area.
Click to expand...

I replace them since if it goes bust there will be PCB spread and if the damn thing catches fire it will spread Dioxine, which a very nasty set of chlorinated compounds. And removing the PCB filled capacitors is not any harmful since the PCB is contained. Hopefully the environmental station know how to deal with PCB (burning it at very high temperature)

Asbestos on the other hand will do nothing harmful when used normally unless I try to remove it. So, no, the ventilation ducts in my house which have some 5 % asbestos in them will not be removed. But I will remove all PCB filled capacitors and replace them as a precaution.
 
I don't think the caps in the 8/a has PCB in them since it is mostly manufactured after the PCB ban. I replaced PCB filled capacitors in my PDP-8, PC04, DF32, TU55, PDP-9 but not the 8/a. They were clearly marked "No PCBs"
 
Fair enough... it's up to you I suppose. PCB's are nasty stuff when burned or ingested.
http://www.cdc.gov/niosh/docs/86-111/

But at the energy available in that circuit it is very unlikely that the steel case would rupture, nor that the contents would catch fire (which also requires a large enough opening in the case for oxygen to support combustion). The usual catastrophic contamination scenario involves a building or substation transformer with tens of gallons of partially burned PCBs.

I never said that removing them is harmful. I would not want to depend on "hopefully" anyone knowing how to deal with them.
 
DrCharles said:
But at the energy available in that circuit it is very unlikely that the steel case would rupture, nor that the contents would catch fire (which also requires a large enough opening in the case for oxygen to support combustion).
Click to expand...

The ferroresonant cap in one of the RICM's PDP-8/s systems bulged and leaked oil.
 
You must log in or register to reply here.
Back
Top