Hey, I am trying to understand how the Doepfer precision adder works. I like the switches where you select if you want to add or subtract the voltage!
So far I have come up with this and I think it works:
but if I compare it to the photos of the A-185-2, some things look different! For example they use 100 ohm for the resistors that connect the opamps in the left column (inverting opamps) to the final summing opamp, but if I do that, the sum is less, precise… but they clearly had a good reason to do what they did! so… why?? xD
Also their switch setup is quite different: they have all of the “left positions” connected together into one opamp and all of the “right positions” toegether into another one. and the center looks also like they go into the opamps, similar to my version. I can not imagine how that works… I think there might be another arrangement which leads to higher input impedance, but I can not think of it… maybe someone can? I can not trace any of the inputs… so no idea how the whole arrangement works
Maybe I should just ignore that and see how far my solution takes me, but it somehow nags me!
On the picture of the PCB you can clearly see the two “buses” between the switches, and the 100K 0.1% summing resistors…
There are 3 more 100K 0.1% resistors, I can only guess they are used in the 3 non-attenuated input buffers.
Now, I don’t understand how a single trimpot can be used to adjust both + and - gains to exactly 1.
But someone else will surely chime in
I don’t know why they’d use 100R between one op amp and another, but if it were in the loop, i.e., immediately after the op amp before the feedback branch rather than after the branch, it would not alter the precision.
If you pursue your version I’d suggest a resistor in such a position after the output op amp. However, Matthew Skala has a persuasive argument that 100R is too small: if the output is shorted to ±12 V, the power dissipation would be likely to burn out a 1/4W resistor. But 1k inside the loop would be fine.
Thinking about it, it still doesn’t work… The op-amp summing the positive “side” must be inverting to work as a summer…
I’m quite sure the left side is correct, you can more-or-less follow the tracks on the PCB…
The right side is more difficult…
After starring another 10 minutes at the pictures of the PCB…
I’d say there are not enough “horizontal” traces to rout the jacks to the op-amps and back to the switches to implement the input buffers…
(Or maybe yes… at the very top ? no better resolution pictures ?)
So we have 8 op-amps after the switches…
Two are used to sum the + and the - sides of the switches, one is used to invert one of the sides, one is used to sum these two signals, one is used to invert one of the outputs. This leaves 3 unused op-amps…
There are three non inverted outputs and one inverted; I’d say the last three op amps would be buffers for the non inverted outputs.
Except I haven’t yet figured out how that’s done with only 9 matching resistors. 3 100k plus a 100k pot for the inputs, 2 for the 4-input summer feedbacks, 2 more to invert one of those, 3 more to sum those, 2 more to do one more inversion, that’s 12…
OK, 11 if you do a differential sum instead of invert and sum…
Not completely sure, but I think this is what happens! The upper part is repeated for all inputs but the first with attenuator, I think (so far I only checked the second from top ).
No idea why they always put 100 ohm into the feedback loop… also no idea if this can work, I will try to simulate it tomorrow!
Oh, and I think I mixed up inverted and normal out…
edit:
I also traced the 1V reference and it is different from what I thought before, still not sure if it is correct now, but it looks okay to me (Also uses less current!! WIN WIN!)
The output with the pot is really useful. If I had space I’d put pots on two of the inputs. You can tune it to a 5th or a 7th and just have it ready. So with no extra inputs you can transpose your sequence up/down a fifth or an octave.
When I do my footswitch thing I planned on pairing this circuit with a set of precision voltages for the intervals you want and an analog switch to trigger them.
Yeah, those internal 100Rs are very strange, I think. I don’t see the point at all.
I also find it odd they carefully buffer the inputs, carefully match them apparently, provide a trimmer to get the 1 V/oct exactly right, and then evidently connect all three non inverted outputs to the same op amp output?
I can not trace any of the inputs… so no idea how the whole arrangement works 
).