-----Oorspronkelijk bericht-----
Van: cctalk-bounces at
classiccmp.org
[mailto:cctalk-bounces at
classiccmp.org] Namens Tony Duell
Verzonden: zaterdag 4 april 2009 22:05
Aan: cctalk at
classiccmp.org
Onderwerp: Re: HP 9825 16K 09825-66523 memoryboard failure
I have a 16K memoryboard for my HP9825A it suffered from a bad -12V
and
I assume you have schematics.
Yep, thanks to your excellent work.
destroyed all the rams TMS4080.
This puzzles me a bit. The -12V rail is regulated down to -5V
using a resistor/zener circuit,. and it's the -5V that's used
by the RAMs. I assume you've checked this voltage.
Or did the -12 disappear, removning the bias from the DRAMs
and cooking them?
The -12V was defective (not on this machine but on another witch
I fixed
(uA723 replaced) and put a 8k board in it.)
I measured the voltage and there was little or no noise on the power lines.
I did put some extra decoupling caps (100nF and 2.2uF tantalium) on the
power lines just to be sure.
That wasn't the problem, the errors are a bit erratic always the same bit 9
but not always the same address.
Today I replaced the address buffer ('LS368) but that wasn't the solution
too.
I bought from ICI 32 pcs. nos uPD411AC-2 replaced
all the
rams but I'm
getting an error when testing the board.
The HP9825 isn't under suspect because it works okay with
another 16k board.
The systemtest (from the tape) says : 60K board
failes Address
60100,60300,60400 not always the same sometimes only 60400.
Bit 9 sense 0 or 1 (not consistent sometimes 1 sometimes 0)
I swapped
the
LB9 and UB9 chips on the board with other 411's(new ones) But the
error stays I replaced the refreshcounters decoupling caps but
nothing.
First check the power lines with a 'scope. Noise on power
lines causes all sorts of problems with DRAMs.
I've had vias fail on old HP boards when
soldering/desoldering components. My next task would be to
check that all pins on all RAMs go to the rignt places and
that none are left floating.
I'll think I'm going to do that I got a current tracer, so I'm going to use
that first.
And see if something is coming up.
And then do make sure that all address, data, etc pins are
being driving nad that none are floating around.
-tony
Thanks,
-Rik