Re: 7834 readout problems


 

On 17 Sep 2014 20:40:49 -0700, you wrote:

Hi Albert - thanks for joining in the fun! I did check those voltages, and they're both hovering at 15.04v.

The voltage at the base of Q3264 should be 3.1 volts.
got 3.12 - close enough?
If it was very far off, then Q3264 could be the cause but a value of 3.12 is
ambiguous. 0.1 milliamps if Q3264 was very bad would only cause a shift of 70
millivolts so I was hoping for something more.

The CR321x and R32xx parts are the switching network which disables various
readout elements based on which plug-ins are used and which plug-ins are
selected via the mode switches. The force readout signal from the plug-ins is
used to prevent the mode switches from disabling a specific readout so plug-ins
like counters and multimeters do not need to be selected to use the readout.
On other 7000 series oscilloscopes, you can disconnect the connector P2118 going
between the readout board and the mode switches and force readout signals to
force all of the readouts to be active.
Oh, OK, I see what you mean about the force readout - I found P43 on the main interface board which when disconnected does exactly that.
The force readout signals allow the plug-ins to override the mode switches as
far as the readout is concerned.

I considered suggesting this early on as it is likely to improve the situation
or at least reveal something but did not because I think it would fix the wrong
problem.
Yeah, that's what I suspected, as specially given that I'm seeing the correct character in one of the characters (the '<' for uncal). I'm going to try it anyway, though that may be Tomorrow Paul's problem.
If it makes for a quick fix though, then maybe it is the way to go. It is not
uncommon to trade off gain and zero errors in analog to digital converters to
minimize peak or RMS error which *is* what is needed here.

I have never before studied the detailed differences between the earlier readout
boards and the later ones like in the 7834. This game is fun.
yeah man, thanks for playing! There was an even later version on the 7834 readout that I think that drew the characters raster style, like the character generator on the DF01. What an ingeniously odd circuit this is…
I know the late 7903, 7844, 7104, and 7904A used a microcontroller driven vector
readout as shown in a change notice in the scan of the 7904A service manual that
I have. The late 7633 or 7632B or both used either the same board or
something else again. I do not have a late service manual for either of those.

The late 7934 had something that looks like a hybrid of the original readout
design and the late 7904A readout design. It does not have a microcontroller
but cleverly swapped an EPROM for the analog character ROMs.

My guess is that the readout board evolved to use up old readout parts that
Tektronix still had on hand after they stopped producing them.

The 7854 of course generates its own readout as part of a full raster display.

All of my mainframes have the original analog ROM readout design in one form or
another so I have not studied the later ones.

(*) This points to either Q3264 or Q3406 being the problem because Q3419 can
only increase the current.
Ahh, interesting. Q3264 is listed as 'Selected' so I was also suspicious that this might be the culprit.
It might have been selected for high gain.

Unfortunately, I swapped Q3234 & Q3264 and no change. That leaves Q3406. Are they in the same package for thermal matching? I swapped it with the one from my 7603. no change.
Dual transistors like Q3406 are usually employed where Vbe matching is important
but that does not make all that much sense to me here since the voltage steps at
the base of Q3406A are about 500 millivolts. I assume they needed the extra
accuracy.

Q3416 maybe?
It is worth checking since we are out of good options but I doubt it. I think
the signal into pin 9 is digital. If the voltage level is far enough off which
I believe you measured to be the case, then it might cause U3418 to behave
oddly.

I do not think there is anything special about Q3416 so your favorite general
purpose PNP should work in its place. The line which drives pin 4 goes to some
other places through CR3416 which should be checked.

I would try adding a pullup resistor to pin 9 to raise the voltage a little and
see if that changes anything.

While difficult, what I would do is carefully measure the readout voltage levels
at pin 10 compared to the +15 volt reference. TP3445 off of the timeslot
counter can be used to trigger an oscilloscope and produce a synchronized
display. Careful measurement at that point should reveal if there is an offset.

Join TekScopes@groups.io to automatically receive all group messages.