Masterpiece

[Stefanoo]

Originally Posted by FdW

You need to connect the servo to the output of the buffer, otherwise vgs (when connected to the input (as shown ())) of the fet's is used to create an imbalance in the mirrors. And if you insist, the experiment (that fails / will fail) will be to switch it to the buffer input (and back (very fast)).

I see what you are saying.
Let me correct it then!

[Stefanoo]

I was thinking why it was working on my prototype with the servo before the buffer.

Here is the updated schematic, this should work just fine.

Thanks for the input.

[FdW]

Originally Posted by FdW

You need to connect the servo to the output of the buffer, otherwise vgs (when connected to the input (as shown ())) of the fet's is used to create an imbalance in the mirrors. And if you insist, the experiment (that fails / will fail) will be to switch it to the buffer input (and back (very fast)).

Edit: Wrong I was. I should have said that the vgs of the fet's would be translated in a output offset if the servo-input is connected to the buffer input.

See picture (telling a 1000 words) [And yes, I have seen nicer drawings]

[Stefanoo]

Originally Posted by FdW

See picture (telling a 1000 words) [And yes, I have seen nicer drawings]

Fine, I don't understand why the latter wouldn't work, since the output resistor will restore the output offset to 0 (simulation shows that it properly works).
Servo will push 0V at the gate, Source of the Jfet will go roughly down to -0.2V and then other side of the resuistor will cancel the offset restoring it to 0V.

Can you tell me where this reasoning fails??

EDIT: I have the prototype running that way and output offset is about 1mV!

[Joachim Gerhard]

I would include the Calvin buffer on the test PCB.
The InterFet type can then be build external.
However good the InterFet sounds i think that the Calvin buffer has better technical data.
The InterFets could also being used in a Calvin arrangement.

[Stefanoo]

Originally Posted by Joachim Gerhard

I would include the Calvin buffer on the test PCB.
The InterFet type can then be build external.
However good the InterFet sounds i think that the Calvin buffer has better technical data.
The InterFets could also being used in a Calvin arrangement.

yeah it actually makes more sense.
It is easier to build that buffer vs Calvin so having it on a PCB will help development.
Good point!
True instead of the K389 we can use the IF3602 since it proved to have a superior sound.

[Joachim Gerhard]

Yes, we can try that.

[FdW]

Originally Posted by Stefanoo

Can you tell me where this reasoning fails??

EDIT: I have the prototype running that way and output offset is about 1mV!

It (probably) fails nowhere, but connecting to the low impedance output will be better than to the high impedance input, also it will remove the 1mV, and there is no chance that component tolerances, input voltage differences (vgs), aging, temp, and more factors will ever result in (any form of) output offset.

[Stefanoo]

Originally Posted by FdW

It (probably) fails nowhere, but connecting to the low impedance output will be better than to the high impedance input, also it will remove the 1mV, and there is no chance that component tolerances, input voltage differences (vgs), aging, temp, and more factors will ever result in (any form of) output offset.

ok now I agree with this

[Stefanoo]

Originally Posted by Joachim Gerhard

Yes, we can try that.

ok sounds good. I will update the schematic with Calvin buffer using IF3602 on driver.