[Geoman]

so the 2nd new old stock PSU was a delivery of initially 3 packages in November 1989. The one that is now powering the Crimson is

Code:

MMAX95-11Y0Y2Y2Y/115-230AT SW

this (marked 1/3) is powering on once it gets power without any load and delivering the voltages right away.

The spare I got on Friday is marked 2/3, and has no "SW" its model number. It powers on with fan spin, but no voltages at all.

On Jan-Jaaps image he had his MITE on a testbench.
There is a DB15 adapter marked as "J1", is it true, that PIN11 is connected to +5V and PIN9 is connected to -V5?
   

---

So of course I took it apart to get some pictures. It's as new as new can be, a part of some minor rusty screws. I won't take it further apart, since I'm lacking the time and abilities to troubleshoot it. Perhaps the seller someday will find the third and last one of this batch too...

Compared to the one that is working in Crimson now which has grey capacitors, in this PSU there are three big blue capacitors marked "Aerovox M".

[jan-jaap]

There is a DB15 adapter marked as "J1", is it true, that PIN11 is connected to +5V and PIN9 is connected to -V5?

Not -5V but GND.

What you're looking at is the remote sense feedback. It is a mechanism to compensate for voltage drop over the cables from power supply to the load (the system). If there's a lot of current there will be a lot of voltage drop over the cables, and the PSU can compensate for that by increasing output voltage slightly so there's still exactly 5V to the system.

Here you see this phenomenon in action: the PSU (Powerone SPM5) is seeing a load of 40A. Output voltage of the PSU is 5.02V, but the load only sees 4.88V. So even this short length of 16mm2 cable already causes a voltage drop of 0.14V.



In the LH Research PSU from my Professional IRIS there's something wrong with the feedback compensation circuit because if I connect the sense feedback *and* there's significant load (more than the 40A I can consume with my electronic load), then the feedback causes a fairly significant oscillation of the output voltage at ~ 2KHz frequency. This in turn causes the PSU to make loud and terrible noises.

In the end I opted to simply leave the sense feedback disconnected. The configuration of my Professional IRIS is fixed and so is the current it consumes, so I manually adjusted the PSU output voltage until I had exactly 5.0V on the backplane of the system.

So, the relevant bit of information for you: connector J1 is irrelevant to your problem. The PSU should fire right up and put out the correct voltages. It has a couple of power resistors internally on the 5V module to sink some current and do proper regulation even without any load attached. You can see them under the copper coils to the right of the blue output capacitors.

NB: the PowerOne PSU has sense feedback capability as well, using the white 6-pin connectors on the output modules. This is P3 of the wiring harness of the Crimson/PowerSeries.

[Geoman]

Thank you! Then perhaps I take it further apart some time and check the individual modules for failures. BTW I did't connect the feedback line in the Crimson, it works fine as it is.

But! what do I see here? RIFA!! so there is a starting point at least

[jan-jaap]

But! what do I see here? RIFA!! so there is a starting point at least

Doubt it. They are metalized paper capacitors, for high frequency filtering. They can suck up moisture which cause them to swell up (and crack) and they can short and explode much like tantaliums but these power supplies are such beasts they won't even notice.

It's not impossible that the power supply requires a control signal on J1 to start up, much like what's needed in the PowerOne PSU for the 4D series. In that case, good luck finding documentation ... there's some on http://bitsavers.org/pdf/lh_research/ from the guys who rebuilt the Digibarn Alto.

[jan-jaap]

A quick recap of what I remember working on the similar unit from my Professional IRIS:

Near the top of the photo, left to right: the two large coils are part of the EMI filter, under the heatsink are your rectifier diodes, the two grey capacitors are the reservoir of 300VDC.



Below two modules, more or less identical in construction. My LH Research PSU had one, but had one less output also.

The main 5VDC output is composed from two PCBs. The upper one with all the white potentiometers does PWM, control and 300VDC chopping, and the lower one has the isolating transformers, rectifier diodes and output filter caps.



Behind to the right on the heat sink you can spot four 2N6678 transistors. Those are the 300VDC choppers. In my LH Research PSU they were 'tired' -- the amplification factor was well below spec and the fear was that the control circuit was compensating for this by increasing the PWM duty cycle to get enough energy into the transformers to keep up with output demands. This PSU was 'squeeling' under system load, but not on the bench, and the idea was that the noise was coming from the transformers, even though they were filled with resin to avoid vibrations.

You can see that there's writing on these transistors. They may be matched. I replaced them anyway, first with 2N6678 from another manufacturer, then a 2nd time with NOS Motorolas (as used by LH Research). This changed the noises but never made it go away entirely.

Like I said before, in my unit the sense feedback is absolutely involved in this problem and I disconnected it in the end because the oscillation on the output is simply unacceptable, regardless of the noises. The LH Research is an older design than the PowerOne units. The PWM circuitry is built from many analogue components and calibrated with those white potentiometers. Without a service manual I wouldn't touch any of it. This makes it much harder to service/less futureproof than the PowerOne units.

Finally in the lower photo: power from the choppers enters the transformers in the middle and comes out on the red wires to rectifier diodes on the right. From there it goes via filter coils (copper color) to the blue filter caps and the output terminals.

If you look carefully, you can also spot a red/black twisted wire pair going from the outputs back to the control board. That's got to be (voltage adjustment?) feedback

[Geoman]

As always thank you so much for this analysis! It was a delight to read!

They can suck up moisture which cause them to swell up (and crack) and they can short and explode much like tantaliums but these power supplies are such beasts they won't even notice.

So I'm going to leave the Rifa-cap as it is.

It's not impossible that the power supply requires a control signal on J1 to start up, much like what's needed in the PowerOne PSU for the 4D series. In that case, good luck finding documentation ... there's some on http://bitsavers.org/pdf/lh_research/ from the guys who rebuilt the Digibarn Alto.

Yes - I already downloaded everything I could find on the net, + your thread on reviving your LH PSU.

Those are the 300VDC choppers. In my LH Research PSU they were 'tired' -- the amplification factor was well below spec and the fear was that the control circuit was compensating for this by increasing the PWM duty cycle to get enough energy into the transformers to keep up with output demands.

I opened the PSU today from the other side to do a further visual inspection - literally there isnt't s single spec of dust inside this PSU, to I think these choppers still have a lot of life in them, like they were never used.

Without a service manual I wouldn't touch any of it. This makes it much harder to service/less futureproof than the PowerOne units.

Yes, I better leave them as they are.

If you look carefully, you can also spot a red/black twisted wire pair going from the outputs back to the control board. That's got to be (voltage adjustment?) feedback

Yes I see this wire pair


So the plan is to do further research concerning control signal on J1 to start up. The other model starts up right away, and has the "SW" designation on its model number.

[Geoman]

I dared to completely disassembly the PSU - it was not that hard, because it is quite modular. Next step will be a thorough visual inspection.

[Geoman]

As I took the whole thing apart, a replaced the burst RIFA-capacitor on the primary section as well as a small bulged(?) capacitor on the control section. As Jan-Jaap predicted, replacing the RIFA did not help: no voltage output.

I scoped the DB-15-adapter (see pictures). The waveforms were also measurable on the outputs themselves. Pulling any of these to ground did not help either to start the PSU up. 

I checked the voltages inside the PSU as well: on all posts to the modules were high voltages measurable (wore thick leather gaunlets while doing this). So the primary side is indeed working. Both fans are spinning.