This forum has been very helpful to me. I've had it
bookmarked for years but only joined recently. I'm
looking for some wisdom from your collective experience.
This weekend I recapped two FIC FB11 motherboards.
One had been running my office server for eight years
and became unstable last spring (2009). It had
developed bad RAM once before that. The second is
a spare that started failing a couple of weeks ago
with ATA errors and freezing up once a day or so,
though it had been running only since May.
This spare had a bulging 1000uF cap that seemed
suggestive to me, so I decided to recap both in
hopes that they would then be fine. (I had contacted
Topcat about recapping with Rubycons but got anxious
this weekend because I have one last spare running
now and that board seems to be starting to generate
Adaptec 2940UW SCSI timeouts, which is what the
others started doing before they failed. Perhaps I
shouldn't have been so hasty!)
One board had ten 1500uF 6.3v I.Q. caps on the VRM
and ten 1000uF 10v Teapo SEK caps, and a number of
330uF and smaller Teapo SEK and SS caps. These
all seem to be general purpose 105 C rated caps.
The other board had the 10 I.Q. caps but the rest
were S.I. brand. I don't know the ESR rating of the
I.Q. caps, but I replaced them by some Nichicon PWs.
I replaced the 1000uF caps by Nichicon HMs, and for
good measure I replaced the rest of the smaller caps
with Nichicon P- or H-series caps too.
The recap seemed to have gone well in that memtest86+
ran fine for several hours on each board. However,
when I try to run my OS, NEXTSTEP 3.3, on the test
bench, both boards are similarly unstable. They both
develop ATA errors with two different HDs and two
different cables. These errors are not present on one
board at boot up but start after the system has been
running for a while, say 15 minutes. In addition, the
original board does not recognize the PS/2 mouse.
NEXTSTEP puts great stress on the hardware and I've
found that some boards that are stable under Windows
are unstable under NEXTSTEP, so it doesn't seem odd
that a program like memtest86+ would run fine while
NEXTSTEP does not.
I'm hoping that someone on this forum can give me a
hint on what might be going on. The common hardware
between the two test setups consists of the P/S,
the CPU, a stick of SDRAM, the monitor, keyboard, and
mouse. It doesn't depend on the hard drive or the
cable, though both cables could be bad. I'm not sure
that the P/S is all that good; it's a brand new AGI unit
that came with the server case, which I swapped out
for a better unit when I originally built the server. It
does test OK with a small ATX P/S tester but I've not
put it on a scope.
I'm wondering what the problem could be! Perhaps
I have some solder joints that are cold even though
they look good, or perhaps, despite being careful,
I've damaged my boards, or maybe low impedance
caps outside of the VRM make the ATA controller
unstable, or maybe I need lower impedance caps
on the VRM than the Nichicon PWs. I could
try a different P/S, RAM, and CPU, too, since
they are common to both boards.
It's been a while--about thirty years--since I've
repaired delicate electronic equipment, and the
last time I had a full schematic! It's somewhat
uncomfortable working without full information.
Any ideas?
bookmarked for years but only joined recently. I'm
looking for some wisdom from your collective experience.
This weekend I recapped two FIC FB11 motherboards.
One had been running my office server for eight years
and became unstable last spring (2009). It had
developed bad RAM once before that. The second is
a spare that started failing a couple of weeks ago
with ATA errors and freezing up once a day or so,
though it had been running only since May.
This spare had a bulging 1000uF cap that seemed
suggestive to me, so I decided to recap both in
hopes that they would then be fine. (I had contacted
Topcat about recapping with Rubycons but got anxious
this weekend because I have one last spare running
now and that board seems to be starting to generate
Adaptec 2940UW SCSI timeouts, which is what the
others started doing before they failed. Perhaps I
shouldn't have been so hasty!)
One board had ten 1500uF 6.3v I.Q. caps on the VRM
and ten 1000uF 10v Teapo SEK caps, and a number of
330uF and smaller Teapo SEK and SS caps. These
all seem to be general purpose 105 C rated caps.
The other board had the 10 I.Q. caps but the rest
were S.I. brand. I don't know the ESR rating of the
I.Q. caps, but I replaced them by some Nichicon PWs.
I replaced the 1000uF caps by Nichicon HMs, and for
good measure I replaced the rest of the smaller caps
with Nichicon P- or H-series caps too.
The recap seemed to have gone well in that memtest86+
ran fine for several hours on each board. However,
when I try to run my OS, NEXTSTEP 3.3, on the test
bench, both boards are similarly unstable. They both
develop ATA errors with two different HDs and two
different cables. These errors are not present on one
board at boot up but start after the system has been
running for a while, say 15 minutes. In addition, the
original board does not recognize the PS/2 mouse.
NEXTSTEP puts great stress on the hardware and I've
found that some boards that are stable under Windows
are unstable under NEXTSTEP, so it doesn't seem odd
that a program like memtest86+ would run fine while
NEXTSTEP does not.
I'm hoping that someone on this forum can give me a
hint on what might be going on. The common hardware
between the two test setups consists of the P/S,
the CPU, a stick of SDRAM, the monitor, keyboard, and
mouse. It doesn't depend on the hard drive or the
cable, though both cables could be bad. I'm not sure
that the P/S is all that good; it's a brand new AGI unit
that came with the server case, which I swapped out
for a better unit when I originally built the server. It
does test OK with a small ATX P/S tester but I've not
put it on a scope.
I'm wondering what the problem could be! Perhaps
I have some solder joints that are cold even though
they look good, or perhaps, despite being careful,
I've damaged my boards, or maybe low impedance
caps outside of the VRM make the ATA controller
unstable, or maybe I need lower impedance caps
on the VRM than the Nichicon PWs. I could
try a different P/S, RAM, and CPU, too, since
they are common to both boards.
It's been a while--about thirty years--since I've
repaired delicate electronic equipment, and the
last time I had a full schematic! It's somewhat
uncomfortable working without full information.
Any ideas?
Comment