DEC H7104-D debugging...

Josh Dersch derschjo at
Wed Mar 23 23:21:37 CDT 2016

On 3/22/16 12:29 AM, Mattis Lind wrote:
>> such good luck.  No smoke or fire, but now I get nothing at all out of the
>> supply.  The whine is gone, but there's nothing output at all.  The
>> fuse/resistor didn't blow (it's still got continuity across it) and the
>> transistor I replaced is still fine, but there must be something else in
>> the supply that's causing issues...
>> Blargh.  I hate working on power supplies.
> In this case you have an excellent technical manual and a good schematic
> which should help a lot. Working with big SMPSU without schematics or
> technical manuals is a not fun.
> So here is my piece of advice directly from my head.
> 1. The PSU has really two AC inputs wired together in the input harness.
> You can separate the startup supply part from the SMPSU part. Connect the
> startup PSU to a separate AC input and the SMPSU part to a insulation
> transformer, a variac and lightbulb in series.
> 2. Check that the startup PSU, that uses a normal 50/60 Hz transformer
> gives the correct voltage. +11 if I remember correctly.
> 3. There is a circuit that monitors the input rectified 300VDC voltage and
> enables the relay when it has reached a proper voltage. It is a soft start.
> Disable it for temporarily. Good idea to check that is working though. If
> not the soft start resistors will become overheated when trying to run it
> at full load.
> 4. Now you need to have the SMPSU section connected to AC inlet. On one of
> the daughter boards there is a switching bias supply. Check that it
> generates the +/-12 V and +5V.
> 5. Is the main SMPSU switching logic making a nice square wave signal
> output. The control board is yet another daughter board.
> 6. I disconnected the output terminals from the H-bridge to the transformer
> and connected a dummy load to be able to have a look at the output waveform.
> 7. Reconnect the transformer. The main switch produces +5V and +38V. The
> 38V is then used on the daughterboards to create all the other voltages,
> +/- 12V, +/- 15V, +12V SB, +5VSB (depending on which supply). Is there 38V
> here?
> 8. There is a crowbar circuit on the +5V output terminals. Check that it is
> not tripping.
> Please note that when working with the +2.5V supply it gets the bias
> voltages from the +5V supply. So either get that one working first or
> supply bias voltage from a regular PC supply.
> /Mattis

So I had a few minutes tonight to investigate a few things:

- The startup PSU is running, and the relay in the 5V supply is firing 
and stays energized until the supply is powered off.
- The +5V, and +/-12V bias voltages are being properly generated. This 
makes sense, since the 2.5V supply is working fine.  From my 
understanding of the supply, this means the 300VDC is also present (also 
validated by the relay firing.)
- The 20Khz clock generated on the bias board is present (measured from 
pin 8 of J1 on on the bias board), at the proper rate and duty cycle.
- 5V AC and DC LO signals (measured at J1) are .3 and .9V, respectively.

So far, so good.  I also measured the 5V Overcurrent and 5V Crowbar on 
J1 (pins 2 and 19) and they're at 12V and 8V respectively.  I'm not sure 
I understand what this indicates; the Overcurrent lamp is not 
illuminated on the power controller, for example, and per the tech 
manual (section 4.4.2) if an overvoltage occurs and the crowbar kicks 
in, the supply should shut down for ~1 second, which I'm not seeing any 
indications of (I'm not seeing the voltage reach anything over .2V with 
a small load attached).  The overcurrent protection appears to work 
similarly.  I don't see an indication of what the levels for the signals 
on J1 are supposed to be, however.

I'll hopefully have more time to debug later this week, thought I'd 
report my findings and see if anyone has any bright ideas.

Thanks as always,
- Josh

More information about the cctech mailing list