Dying VaxStation 4000 VLC - help?

From: Mark Tapley <mtapley_at_swri.edu>
Date: Wed Nov 20 11:36:00 2002

All,
        The Good, the Bad, and the Ugly.

The Good: I got the comparator out, visibly intact. No legs lost, no big
divots chipped out of the mainboard, etc. I got its replacement back in.
One cold-looking solder joint, but I can retouch that later and it's
conducting OK for now. The system works like it did before the operation.

The Bad: It works *exactly* like it did before the operation, which means
the comparator was not the problem. Same run-a-little, hang-a-little
behaviour, same drunkenly varying voltage on non-inverting input to side A
of the comparator.

The Ugly:
        This problem is vexing me.
        Right after I got the old comparator out, I went to work with the
Ohmmeter. (Remember this is an old Archerkit Analog 20kOhms/volt on its R*1
setting, so there's probably some significant internal resistance. I did
zero the needle with the probes crossed before each series of measurements
on the same range, but...) Initially, I read on the voltage divider 24 Ohms
to +5V (across the component marked "1152") and 80 Ohms to Ground (across
"6192" and a capacitor, in parallel).
        Then I took the board off the aluminum foil I had it sitting on
while soldering. Both of those resistors started reading much higher
resistances, and their resistances were different depending on which way
round I had the probes and on what range the meter was set at. Here's the
table, board sitting on a towel:
                1152->+5V 6192->Gnd
x1 26/ manyKOhm 80/ manyKOhm
x10 18/ many K 32/ many K
x100 1250 / 9000 9500 / 1800
x1000 7800 / 9000 8500 / 10000

        Numbers before/after the slashes are with probes forward/reversed
(don't recall which is which, and not necessarily consistent entry to
entry). Table is in near chronological order, and that may be significant.
        What makes an SMD resistor start acting something like a diode?
Sigh. What's with the different readings at different scales? Do I just not
know how to use an Ohmmeter, fer cryin' out loud? Is all this the natural
consequences of trying to measure resistances in-circuit, with active
components messing things up?
        In any case, it looks like I'm headed for the merry world of
surface-mount soldering. You know, when I got this system I named it
"Sensei" figuring I'd learn something from it. Maybe I'll try to be a
little less clever naming the next system.


Toth wrote:
>I have a working VLC here that I can refer to for component information if
>there is anything specific you need to know.

        You da man! OK, yes, please!
        Sitting facing the rear of the computer, power supply unit to your
right, find the connector on the cable bundle coming from the power supply.
Just beyond it is an 8-pin IC marked LM393.
        Less important measurements: With the computer running, I'd like to
know the voltage on pins 2 and 3 (be careful not to short things!). Pins
are counted from Right (pin 1) to left (pin 4) across the far side.

Voltage pin 2 (expect about 3.5 V):
Voltage pin 3 (expect about 4.2 V):

        Please watch pin 3 for a while (30 seconds or so) to see how much
it varies (expect none).

        With the computer off, look just to the left of the comparator,
where there's a column of 4 surface-mount resistors (?), farther two
brownish, nearer two black. Please measure resistance across each of those.
I've named them from far to near, using the practically invisible numbers
written on their tops. I'm really only interested in the first 3, since I
don't know where the last one goes, but I'm hoping I might learn something
about the naming convention:

Resistance 181 (expect about 180 Ohms) :
Resistance 6192 (expect ... ummm ... ??) :
Resistance 1152 (expect ....ummm ... ??) :
Resistance 4221 (don't know) :

6192 has a capacitor in parallel, so may take a bit to settle. My meter
shows 170 Ohms for "181", except when measuring reversed-probes at X1, when
it shows 75 Ohms.

>The board itself is a 7 layer
>board, so it wouldn't be easy to draft a schematic of the 3.3V regulator
>and comparator section of the board.

        No lie. What I (think I) know about the schematic comes from
finding hard shorts between component pins with the Ohmmeter. I've
described it before, but briefly:
        Pin 2 on comparator (inverting input) connects to middle pin of the
"88W" device (black, 3 pins, just beyond the inverter) and to one end of
the resistor marked "181". The other end of that resistor goes to +5V. I
think Tony's right, 88W is a 3.5V (or so) Zener diode in a 3-pin package,
one pin of which is not used.
        Pin 3 on comparator (non-inverting input) connects to the middle of
a voltage divider formed by 1152 (to +5V) and 6192 (to ground). 6192 has a
capacitor (at the near left corner of the inverter) in parallel with it, so
that it comes up slowly. That's the trouble spot - in operation, that
voltage wobbles around from about 2.5 up to 4.2 on my machine.


Tony guessed (right as usual):

>2) The '3 terminal component' is a zener with one pin not used. Again we
>know one pin is grounded, and the other goes to the input of the
>comparator. In which case there must be a resistor from that input to a
>voltage source (probably a supply rail).

Resistor "181", about 180 or so Ohms.

WRT to capacitor in parallel with 6192:

>Aha... That would be my first suspect. Has it gone leaky?

I do not know, nor how I can find out (other than replace it - along with
(possibly) the two resistors in the divider, which will be my project for
next week,).

>If it's had significant reverse voltage, then it will now be pretty leaky
>(or even shorted).
>
>However, it's not unheard-of for the markings on the PCB to be wrong,
>just to make life 'interesting'.

Spot on again! I checked voltage on the "reversed" capacitor. Its pip end
is at ground ... but its other end is at -12V (despite the "+" marking on
the PCB near that end). So it's set up correctly after all.

>The capacitor will normally have the value marked on it. Either as
>something like '10u' or as a 3 digit number like '226'. The latter you
>read like a resistor code (22*10^6 in this case) and the base unit is
>_pF_. So 226 is 22uF, etc.

No markings on any of the capacitors that I can see. There are 3 sizes:
large, with rectangular cross section, medium, with square cross section,
small, with rectangular cross section. All just have bright orange bands,
no writing.

>I would be suprised if a resistor had failed, so leave that for the
>moment. If you can't decode the markings on it, then perhaps you can
>desolder it and measure the resistance out of circuit. Yes, I know it's
>faulty, but it works some of the time so you shuld be able to get an idea
>of the correct value.

I've had a thought. Would the following values make sense in the voltage
divider?

1152 - 1152 Ohms, to +5V
6192 - 6192 Ohms, to Ground, parallel with capacitor.

Or is it more likely:

1152 - 11,500 Ohms
6192 - 61,900 Ohms

The tap is spec'ed to have an "input biasing current" of 25 nA. Either set
would fit with the other resistor in that row having a value of around 180
and marking of "181". Current should be enough that the tap drain is
negligible, but low enough not to toast anything. Ratios are just about
right to generate the voltage I'm seeing, when the system is running.

Given either set of values, and that the machine is 25 MHz clock, what's a
sensible range of values for the tantalum cap. parallel with 6192? Is it
critical to get that value just right, or is bigger better?

I'll be on travel Thurs-Sun, so anybody not reading this gets a break from
discarding my (probably by now tiresome) messages. For the rest of you, my
thanks again!
                                                - Mark
Received on Wed Nov 20 2002 - 11:36:00 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:35:27 BST