Altair-what do I do first

From: Ross Archer <archer_at_topnow.com>
Date: Tue Oct 1 18:52:01 2002

Joe wrote:
>
> At 12:23 PM 9/29/02 -0700, you wrote:
> >Bob Shannon wrote:
> >
> >> Thanks for catching my typo Joe. The MTBF (mean time between
> >> failures) is greatly REDUCED by unsoldering chips, often very
> >> dramatically so.
> >
> >
> >Hmm, I never even thought of this one before.
> >Yet another general argument for sockets. :)
>
> Yes and no. I used to be all in favor of sockets but what I've found over the years is that sockets generally cause more problems than they fix. Yes they make repairs easier but electrolysis bewteen the IC leads and the sockets cause problems in time, even to the point to where the leads will corrode completely through and fall off or break when you try to remove them. In addition, ICs tend to creep out of the sockets and have to be periodicly reseated. Using GOOD quality sockets (gold pins) and ICs (mil-spec parts) minimizes problems but are EXPENSIVE (unless you're a good scounger like me and can get them free :-). My personal opinion is to do like most of the good manuafacturers have done and to install SSI ICs directly on the board and install EPROMs, CPUs and specailized parts in sockets. That seems to a good compromise between cost, repairablility and reliability.

I've definitely seen the case where reseating chips was
necessary with sockets.
There's no doubt you're right; the correct approach is to
socket any chip that
you might reasonably replace during the product's service
lifetime (EPROMs),
parts that can be user-populated, or any chips that are
either exceptionally
likely to fail or exceptionally difficult to remove. :)

Still, I love getting a board with sockets, because it means
it's easier for me
to scavange (at worst) or repair.

>
> >
> >I would think if you had an adjustable DC supply,
> >you could gradually ramp up the voltage on the unregulated
> >input and watch the output with a voltmeter. If it ever got
> >to 5.25 volts, you'd not want to ramp any higher and replace
> >the regulator.
>
>
> Of course you can. All the arguments about removing ICs to check the regulators is rediculous! Who in their right mind would remove 50 to 100 ICs to test one part with only three leads! If they're THAT worried about worried about the regulators it would be MUCH easier to remove it and test it by itself instead of removing all those ICs!

OR even better, wire a brand new 7805 or whatnot in a
circuit with a resistive
load to verify it works, then solder it in place of the old
one, since regulators
are so vital and yet so cheap.

>
> >That wouldn't help with re-forming caps though.
>
> Actually it would. You're just powering the circuit from a variable DC supply instead of a variable AC supply and you're bypassing the transformer and retifiers.

I meant it wouldn't help if your variable supply was on the
input side fo the regulator,
rather than the output side.

I'm tempted to experiment with a 7805 by applying a voltage
on the output side
and nothing on the input, to see if this is safe. If so,
your idea of using
a bench supply with current limiting, and possibly ramping
up voltage gradually
to re-form caps, is a zero-desolder, zero-tamper solution
which ought to work
well.

>
> Maybe
> >you'd start at the output of the regulator and work up from
> >a very low voltage?
>
> Assuming that the circuits are designed to operate off of 5 VDC, you should be able to safely start at that voltage or maybe slightly less and then increase the voltage to the rated voltage (8 volts or so for S-100 type cards) while monitoring the output. (In fact, you can take the voltage all the way to the input limit of the regulator as long as the regulator can dissapate the heat generated.) If you want to reform caps at the same time then start lower, say 1 VDC.
>
> Another advantage of using a bench type power supply is that you can set a current limit in case a cap or anything on the board is shorted and the chances of doing any damage are greatly minimized. The current limit setting is sort of a SWAG. On a fully populated board I'd probably start at 1/2 of the rated current of the regulator and probably have to raise the setting to get it to operate. On a board that's not fully populated I'd reduce the current in proportion to the percentage of ICs that are installed on the card. I generally keep notes on how much current various devices draw, If you ever have to test a similar device those notes give a good starting point. I use bench power supplies like this ALL the type, IMO they should be a repairman's second investment after a good meter, even before buying an O'scope.
>
> Or would undervoltage hurt some
> >components due to a mysterious process I'm not aware of?
>
> It's POSSIBLE that some IC or circuit might latch up and cause a problem but I'd say the chances of that happening is between slim and none on a TTL circuit. I have seen that kind of thing happen on CMOS devices such as HP calculators but even there it doesn't do any damage, it just causes the device to lock up. Bear in mind that at this point we don't care if the circuit is operating, only that there aren't any shorts or things of that type.
>
> Joe

Nice ideas.

If one were really seriously into collecting, you could
really make an argument
for needing a good quality bench supply.

Wonder what percentage of folks on classiccmp have one? I
bet it's a fair number?
(Me: 25 Mhz scope, logic pulser, logic probe, numerous VOMs,
an RCL meter, an
impedance meter, is it. I *knew* there was something neat I
was missing!) :)

-- Ross

> >
> >-- Ross
> >
> >
> >>
> >>
> >> Ever notice the soldering specifications for TTL devices, like 300
> >> degrees C for not more than 10 seconds? This limitation is given for
> >> the parts to meet their rated MTBF, not because 300 degrees C for 11
> >> seconds will destroy the parts right away.
> >>
> >> Resolder the parts, and you may be throwing away well over half their
> >> service life. Clearly not a professional way to restore a machine.
> >> For some repairs, we have no other option, but melting solder is a
> >> last resort.
> >>
> >>
> >> Joe wrote:
> >>
> >>> At 10:38 PM 9/28/02 -0400, you wrote:
> >>>
> >>>
> >>>> If you think this does the least dammage, your grossly in error. As
> >>>> a test engineer, I can direct you to any number
> >>>> of volumes that will show you the dramatic increase in MTBF
> >>>>
> >>>
> >>> I think you mean dramatic DECREASE in MTBF. But I doubt many
> >>> people on this list even truely understand what MTBF is. I worked in
> >>> reliablility, logisitics and maintainablity so I'm prpobably one of
> >>> the few that would catch this.
> >>>
> >>> Joe
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> for
> >>>
> >>>> resoldered parts. This is known, for-sure dammage, not some risk of
> >>>> dammage from a theoretical regulator failure.
> >>>>
> >>>> Care to defend this position?
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>
> >>>
> >>>
> >>
> >>
> >>
> >>
> >
> >
> >
> >
> >
Received on Tue Oct 01 2002 - 18:52:01 BST