26 Sep
2014
26 Sep
'14
1:08 p.m.
On Fri, 26 Sep 2014, Robert Jarratt wrote:
On Fri, 26 Sep 2014, Tothwolf wrote:
I had a couple of these PSUs within easy reach. The machine this one came
out of hasn't been powered in a number of years, but was working fine last
time it was powered up. I have another identical PSU in a companion
expansion chassis which I can also pull if needed.
The board in question can be removed from the main PSU chassis without
desoldering any parts. There are 2 screws in opposite outer corners plus 5
more accessible through holes in the board (only the 5 with black plastic
bushings, the others don't attach to the main heatsink but just hold the
semiconductors to the two aluminum plates), along with the 4 shoulder
screws holding the self-aligning connector board to the rear of the PSU's
chassis. Once those are removed / loosened, the board can be lifted out,
but be careful not to tear the huge silpad (use a nylon spudger to
carefully separate it from the two aluminum plates). You don't want to
know what a sheet of silpad material that large would cost to replace...
Looking at the connector head-on from the mating side, the two left-most
pins are connected to '+', and the next pin to the right, along with the
next two pins towards the right on the other side of the connector are
connected to '-'. The pin furthest towards the right seems to be a
different supply or sense line of some sort. (Anyone have a pinout for a
H7874?)
Compensating for lead resistance, with a Fluke 177, I'm measuring 12.1
ohms between the '+' and '-' contacts on my PSU's board.
Does your board also use a bunch of Nippon Chemi-Con / United Chemi-Con
LXF capacitors? I've had so much trouble with various series of those
brown Chemi-Con capacitors leaking or popping that I tend to replace them
wholesale on sight.
Which of the small capacitors was it that you broke on your board?
The extra aluminum plates on the bottom of the board wouldn't make repair
of this supply easy unless you have a vacuum desoldering tool and a long
reach nozzle since you would have to desolder some of the semiconductors
to fully access the solder side of the board.
Which outputs are measuring a short? I just
pulled one of these PSUs
from one of my systems so let's see if I can follow along. I doubt that
large square Sprague capacitor would fail short. It looks to me like a
custom designed flat-style snap-in electrolytic capacitor. The failure
mode for electrolytics is usually high esr and open. Ceramic and
especially tantalum capacitors are much more likely to fail short.
Comparing a known working PSU would be great. The back of the PSU has
two connectors. One has two sets of three power sockets, and one has two
sets of four power sockets. It is the one with two sets of three that
seems to have a short (when powered off and the subassembly carrying the
connectors removed from the PSU). Looking at the connector with the
component side uppermost it is the set of sockets on the left that seems
to be the problem. The two leftmost sockets appear to be shorted to the
rightmost of the three sockets. Looking at the backplane, the markings
say that +12V is shorted to GND.
From what I can see the components that connect the tracks between these
sockets are: three capacitors (two ceramics and one small aluminium
electrolytic), three 2W resistors, and the large Sprague capacitor
(although it is hard to see this for sure). I am not sure why there
would be resistors across the 12V output, unless it was to provide a
dummy load to allow the PSU to work when removed?
If you take the board out you I hope you will see that it is very hard
to remove the heatsink to be able to see the tracks.