For my VCA (see my earlier topic) I would like to mix two input control voltages and have an attenuator for both of these. The easiest solution would have been to just mix two voltages passively and feed them to the current bias input of the LM13700. However, since this input of the LM13700 is current controlled rather than voltage controlled, mixing voltages is definitely not the way to go and I need to buffer my inputs with an opamp to get a reliable current.
When I look up op-amp amplifier circuits, I mostly get inverting summing amplifiers, but I obviously do not want to invert my CV before passing it on. However, when I look at non-inverting summing amplifiers, they typically have a gain larger than 1, like shown below:
Since I would prefer not to have any amplification, I think I should just get rid of the RA resistor shown in the picture, making the numerator zero and thus getting a gain of 1. But I’m not sure if I can get away with that. I have seen people use an inverting amplifier and then invert the signal again using a second op-amp. That would suggest that my shortcut would not work.
Can anybody comment on this? I would be interested in finding the solution with the fewest components that works well.
Referring to this diagram:
what you have on the left is a voltage divider. With R1 = R2, the voltage at the summing junction is the average of the input voltages, (V1+V2)/2. Since the op amp input is high impedance and nothing’s forcing it to ground or some other voltage, that doesn’t affect the voltage divider.
Then what you get at the output is (1+R_A/R_B) * (V1+V2)/2. If R_A = R_B then the output is just V1+V2.
But of course if V1 and V2 are both large then V1+V2 may be larger than the supply voltage and the output will be truncated.
You could omit R_A, R_B, and the ground connection, so just have the voltage divider connected to the + input and the output directly connected to the - input. That’s a unity gain voltage follower whose output is equal to the voltage at the + input, in this case (V1+V2)/2.
Hah of course thank you! I’ve been overcomplicating things again. Combining a passive summer with only 2 resistors and a simple buffer do the trick just fine
Then it’s not a summer, though. (V1+V2)/2 is the average, not the sum
(the reason people use inverting summers is that they keep the opamp inputs at zero volts, making sure that the only limitation is the opamp’s output drive capacity, not its input range, and a proper summer is also independent of the number of active inputs.)
To amplify (sorry not sorry) on the last bit of @fredrik’s comment, if no input is connected to V_2 then the output, with a unity gain follower, is V_1; but if V_2 is normalled to ground, or if it’s got an attenuator (or there’s an output attenuator on whatever you’ve plugged into it) and you’ve turned it all the way down (which connects it to ground) then the output is (V_1)/2. So that can get confusing. With an inverting mixer, the output is -V_1 regardless of whether V_2 is connected to ground or disconnected.
Hah you are absolutely right @fredrik . I’ve not been paying attention here! I continued this project without the summer since that would require extra opamps which I did not have any space for on my board.
I did realize that for a VCA it is preferable to give it a voltage between 0V and 5V to control the gain, so I shifted my efforts to that rather than to create a summer for multiple input CV’s.
In my search for solutions, I did run into a fantastic document by Texas instruments on how to design scale/offset circuits using opamps. Basically, it allows you to specify an input range and output range and guides you through the calculation. For my VCA, I wanted a 0-5V input to be translated to a -12 / +12 output which is used to control the LM13700. Here is the document:
)