• Please review our updated Terms and Rules here

Donner 3500 Testing

billdeg

Technician
Joined
Nov 18, 2003
Messages
3,885
Location
Landenberg, PA USA
thm_donner_3500.jpg

Donner 3500 - 1959 portable analog computer.

thm_donner_3500_SN-plate.JPG

One of the earliest 3500's, early s/n and pre-Systron label plate.

Hey ho...I have started working through the Donner 3500 manual to see how well the unit I have works....looking good so far. Very simple tests, but the 3500 can at least measure voltages correctly. The tubes are all OK. The lights light.

thm_Donner-3500_voltage-test-plugs.JPG

The plugboard wiring to set the unit as a voltage meter

thm_Donner-3500_voltage-test-ok.JPG

12V unregulated supply fed into the plugboard, returns ~12v on the meter.

http://vintagecomputer.net/browse_thread.cfm?id=247
 
Although, your meter reads accurately, it isn't necessary
that it does that. I have a analog computer that reads
10 with 15 volts on the input. It is all about ratios and
scaling anyway.
The spacing between pins is the same as the twin banana
jacks. Although, expensive, they can be had from not only
Pamona but a number of other companies. They are often
used for speaker plugs.
I have a number of Systron Donner plugs but use them my
self on my analog computers.
A quick check of ebay shows dual plugs of 10 ea for 13.95.
Not Pomona but cheaper.
Dwight
 
Last edited:
For those not familiar with analog computers each came with special plugs that lets you connect two plugs with a cap or resistor embedded in the connector. I agree, I can just make something with banana clips and something on the top to hold resistors / caps in place hopefully with minimal electrical interference. The local farmers market should have used clips and posts with screw holes in the sides I can use. I have a lot of standard resistors from kits and such, but not the ones referenced in the analog computer problems guide. I will need some of these as well.

Last night I made a video of the solution (?) to the next problem in the manual. Not so sure I got this right. I was expecting the cap to go up to +100v not "up" to -100v.

https://www.youtube.com/watch?v=2HT7scFacZE
 
Last edited:
Tube analog computers tend to like resistors in the meg ohm
sizes rather than the kilo ohm. Using a couple resistors in series
is useful in many cases. I've found that you can relatively
easily put two resistors in the loop on the side for a cable.
A little hot glue to keep them insulated and protected from
flexing.
I also put some stranded wire in with the lead before tightening
the screws. That seems to work better than a screw into a
single wire.
Spray painting them when done will help to find the different
values quickly. I found that to be better than just labeling them
with a sharpie ( doesn't work well on black plugs at all ).
The original S-D plugs all used color codes.
Capacitors are another issue. You want to use capacitors with
low dielectric absorption. Polycarbonate and polypropylene
work well as I recall.
Do be careful. These old tube types would not pass today's
safety testing labs. You can get a nasty shock when they
are powered up.
Dwight
 
Sometimes it is useful to have a sine/cosine generator.
The classical thinking is to place two integrators in series
and a unity inverter completing the loop.
You'd start with a non-zero value on one of the integrators.
When you try this, you'll see that it seems to die off
in a short amount of time.
At first, you think it just needs more gain so you try
increasing the gain of the unity inverter.
To your disappointment, you find that all it does is
makes the oval instead of a circle when displayed
on a XY scope.
So, what is the problem?
It is phase error. The resistor/capacitor/amplifier never
gets exactly a 90 degree shift. It might be 89.99 but not 90.
This causes the oscillation to die off.
When I did a rose engine demonstration at the Maker Faire,
I needed sine/cosine that would last for 10 of 15 minutes
to plot on an XY plotter.
I found that one could correct for the phase loss and if balanced
well, the oscillation would nether increase or decrease
for an hour or more.
I'd take a small amount of the divided down voltage coming
from the inverting amplifier and add it to the summing
junction of the second integrator.
This would pull the phase forward a little, making up for
the lost phase.
I watched the drift on a differential scope and was able
to make it hold constant for a long time.
A better method would be to use a simple limiter
before the summing junction, where you have the
input coming from the inverter.
This could be a couple diodes and a resistor.
Dwight
 
I have been experimenting with some of this stuff, but to generate sounds that look interesting in an xy scope. I suppose I could apply this to what you are saying here. Easy to fine tune phase with synthesizers
 
follow up...picked up a lot of clips/parts locally. I have what I need now to experiment. By itself the 3500 is pretty limited these probably would have been used in parallel or with a larger Donner analog system and plotter. By itself the same-era Heathkit EC-1 has more to it built-in for desktop analog computing if you ask me....If anyone reads this post in the future I'd like some original parts just to have them, but functionally they're not really needed as long as the core system works and can be calibrated.
Bill
 
I don't recall but I think the EC-1 could only have 4 integrators with
reset values while I can see from the picture the 3500 has 5 that
could be used.
The Comdyna GP-6 and GP-10 ( closer to the 3500 ), each, only have
4 resetable integrators. The Comdyna does have the advantage of
built in capacitors with a two speed select ( I've found it quite handy ).
The 3500 also has things like diodes that can be patched.
What features does the EC-1 have that your missing on the 3500?
I would guess the auto repeat of the EC-1 is handy.
Dwight
 
Last edited:
Yes the autorepeat is very handy....

Re earlier comment...I found an ebay lot of used banana plugs for $15. More than I will need perfect and I can add caps and resistors in circuit with them..

I can't really say with authority which is better than the other, I think I just like the EC-1 because I have more knowledge of how to use it. I can I suppose say that it would be nice to have the Donner potentiometer strip option.
 
Back
Top