1 Nov
2015
1 Nov
'15
6:37 a.m.
2015-11-01 15:04 GMT+01:00 Robert Jarratt <robert.jarratt at ntlworld.com>:
Well. I guess that anything is better than nothing.
The control logic is feed by some low voltage. This can happen in many
ways. Either there is a simple voltage divider network with a zener from
the rectified mains voltage or there is some small transformer providing
the bias voltage. Then when the PSU is up and running the bias voltage can
be feed from one winding of the main switch transformer (it is more
efficient than the voltage divider network).
If there is a voltage divider network, remove some part of it and feed the
control circuit from the bench supply. If there is a small transformer then
connect that to mains voltage but make sure that the rest of the supply is
not connected to input AC.
Then connect the variac to the input rectifier network so that you can
control the voltage over the input filtering capacitors and the main switch
transistor.
The reason for why it normally isn't a good idea to use a variac is that
there a usually several circuits in the PSU that are designed for a certain
operating voltage. Like a voltage divider network for the bias supply or a
small transformer. Then of course if you apply heavy dummy loads on your
supply and are using a variac the current through the transistor can be
quite high (hopefully there are some kind of over-current feedback signal).
Thus my suggestion to not use dummy loads at all when powering it with a
variac just to see how the switch works.
Ok, no SMPS IC. Somewhere there has to be a circuit that generates a basic
clock signal for the supply. I would try to find that signal and follow it
through the circuit.
/Mattis
-----Original Message-----
From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Mattis
Lind
Sent: 01 November 2015 12:46
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Testing H7864 (MicroVAX II) PSU With No Load
s?ndag 1 november 2015 skrev Robert Jarratt <robert.jarratt at ntlworld.com
:
up
Some of you may recall I have a faulty H7864 PSU,
which failed a while
ago with a loud pop, but no obvious physical damage. I replaced the
blown transistor (on the primary side of the large transformer), but
when I power it on, the transistor does not switch and there is no
output, so clearly there is still a problem somewhere.
I have been spending some time drawing schematics for almost the whole
thing. I am now at the point where I intend to compare it with a
working one, probing each one side by side, to see up to where it
appears to be working. It would be awkward to have two dummy loads,
just for lack of suitable equipment. Does anyone know if it is safe to
run these PSUs with no load? Would my testing be valid without a load?
Maybe you can publish your schematic somwhere. It is much easier to come with ideas to pin point problems if we all view
the same schematic. I
think it
would be possible to identify the problem without
comparing the two PSUs
at
least in this case where there is some kind of
fundamental problem.
I do intend to publish the schematics, but right now I am pretty sure they
are:
a) full of mistakes
b) not drawn logically
c) there are a couple of areas I couldn't trace without major surgery on
the board.
I would start with trying to decouple the bias voltage powering the
switch logic
so that it could be run from a bench supply while
powering the main
switch
transistor and power transformer from a
protection transformer and
variac. If
you run like this you could start without any
dummy load at all.
When you get to a higher input voltage from the variac it could be
useful to
have some small dummy load.
I am not a PSU expert, and I am not sure what you mean by a "bias voltage
powering the switch logic". As for decoupling it, again I suspect that is
way beyond my knowledge. I do have a variac though, and I believe it is not
a good idea to power a SMPS from a variac. I guess the decoupling you
mention would avoid the problem of using the variac, but I don't know
enough to do that.
What kind of chip is controlling the PSU? With bias power applied is
there any
switching activity output from the chip? The RC
network that usually
make up
the time constant should have some kind of
sawtooth signal I would
guess. If
not it can obviously be broken or some feedback
signal has caused it to
shut
down, for example due to over current feedback.
There are no chips in the PSU (apart from a couple of comparators). Quite
a while ago, I did put a scope on the base of the transistor I mentioned
(using an isolating transformer), and could not see a signal (or rather a
very small signal), whereas on the known working one I did see a signal. It
has been my aim to discover why this is, but it is too complicated for me,
with no real PSU experience, to understand.
/Mattis
> Thanks
>
>
>
> Rob
>
>