On Tue, 21 Feb 2012 09:11:09 -0700 Jim Thompson
wrote in Message id:
:

On Tue, 21 Feb 2012 06:45:08 -0500, JW wrote:

A heated "discussion" between JL and JT prompted me to post this. I often
get these HP 8116A generators in for repair that have processor reset
problems. CR3 and CR4 in the reset circuit become leaky which pull the
LM324 input negative and seems to drive the LM324 outputs into unexpected
states.

Is it that leaky?

CR4, driving + input of U27A would have to be ~2Meg or less to upset
U27A.

Of the 7 or 8 failures I've seen, they ranged from ~150K to 600K or so.
I'm used to seeing diodes short low resistance (or open), but like that,
not so much. I've been replacing them with 1N4148 since I have loads of
them around, and their reverse voltage spec is higher (although I don't
think that is what's killing them)

LM324, like LM339, on positive input only, will reverse phase at the
output, if + input exceeds Vbe below negative rail. So CR4 is
explainable.

OK. That was what I was wondering when the discussion came up about the
LM339 in SED.

CR3 goes to a negative input, so the output could possibly go
positive, but that would a normal behavior for the LM324.

So what are the "unexpected states"?

U27 pin one goes positive. That part of the circuitry eventually goes to
the CPU, holding it in reset.

1N4150 is rated at 50V reverse. Has "leakage" bee verified?

Only in terms of resistance out of circuit with a DMM, and in-circuit by
observing a negative voltage anywhere from roughly -1VDC to -4VDC at U27
pin 3. I guess the only mystery left is why they fail so often. Thanks
Jim.

On a power up the voltage at CR3 and CR4 swings from about +15 to
-15V or so. They are driven positive first by a LM348 through a diode,
then pulled down through a 100K resistor to -15V. See the two jpgs I
posted for the schematic. My questions are 1. Is it expected that the
slightly negative voltage at the input would do this? and 2. Is there a
reason why CR3 and CR4 fail so often? The diodes are 1N4150. Obviously
there isn't much current flowing through the diodes... Maybe there was a
bad bunch of parts?

Thanks.

...Jim Thompson