On Tue, 21 Jul 2015, Lyle Bickley wrote:
On Mon, 20 Jul 2015 23:14:36 -0600 Eric Smith <space...@gmail.com> wrote:
On Mon, Jul 20, 2015 at 4:46 PM, Rich Alderson <ri...@livingcomputermuseum.org>
wrote:
industry white papers with tables of decay rates for the aluminum
electrolytics that indicate that, *no matter what*, they lose
capacitance over time, until c. 14 years from manufacturer date they
are at 10% of rating.
That's very interesting. I haven't seen those white papers, but the "no
matter what" must in fact depend on something, since on the PDP-1
Restoration Project we found that most of the 40 year old aluminum
electrolytic capacitors still met their original specifications,
including capacitance within rated tolerance. Of the few electrolytic
capacitors that had failed, the problem was a catastrophic failure, not
the capacitance being outside the rated tolerance.
In the PDP-1, we preferred to keep the original components as much as
possible. Had there been a capacitor, the failure which would have
caused extensive damage to other components, we would have given
serious consideration to replacing it. However, that was not the case
for any of the capacitors in the PDP-1.
Had our analysis indicated any expected benefit to replacing all of the
electrolytic capacitors, we would have done so, and bagged and tagged
the originals similar to what we did with failed components, so that
they could be replaced if it ever was desired to return the artifact to
its pre-restoration condition.
I'm not recommending against LCM's policy, but I also wouldn't
necessarily encourage anyone to adopt it, nor to adopt the practices of
the CHM PDP-1 Restoration Project, without studying the issue.
As Eric, I'm a member of the PDP-1 Restoration Team. The PDP-1
restoration was completed in 2005 - and annually we check the power
supplies for voltage, ripple, etc. Not one of the re-formed capacitors
have failed in the ten years since the completion of the restoration.
I also re-formed all P/S capacitors in my PDP-8/S in September, 2013.
Not one has failed since...
How often is CHM's PDP-1 powered up and operated?
If LCM's computers are going to be powered up and used routinely, it
actually makes a lot of sense to go to the trouble to replace really old
aluminum electrolytics, even if they seem to test good, since doing so is
going to increase the reliability of frequently operated equipment.
This is the reason why /I/ replace aluminum electrolytics when I'm making
major repairs or fully reconditioning electronic equipment...I want said
electronic widget to be as reliable as possible because it is never good
when something breaks down while you are using it, especially with gear
which needs to run 24/7/365 for years and years at a time.
One example I can give are some Pentium P55C architecture (Socket 7)
systems which I've been running with minimal downtime for ~15 years. The
original power supplies with their original (and relatively low quality)
capacitors lasted about 15 to 17 years (I think the manufacture date code
stamped on the oldest one was 1998) before the systems began to develop
stability issues, requiring me to rebuild the power supplies with new
capacitors. I fully expect that the replacements would last even longer
than 20 years, however I rather doubt I'll be running those computers by
then.
