EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: warpigs330 on November 12, 2022, 05:57:27 pm
-
I have been working on a Voltage Controlled Amplifier for use in various synthesizer circuits and have noticed that I have significant DC offset that increases as the control voltage turns up the VCA volume. I believe the main contributor to this effect is the slight mismatch of resistor values used in the differential amp. Here is a simulation of the situation.
https://tinyurl.com/27cnhnz2
So my question is what is the best way to solve this? I have thought of three solutions. The first I thought of was to add a trim potentiometer that could offset the input to match the offset at the output. I believe this would work, but it seems like a brute force solution that adds cost and complexity. The second thought I had was high tolerance resistors, which I think is a better solution, but still somewhat expensive. The resistors don't need to be absolutely accurate, just well matched to each other. So I thought of resistor arrays. So my question is, how well matched are resistor arrays generally? The datasheets don't seem to specify the resistor matching tolerance, just absolute tolerance. Would a 1% resistor array be significantly better matched than two 1% resistors?
-
Regarding the matching of 1% individual resistors vs matching of 1% resistors inside an array, my experience tells they are comparable. This experience includes liaising with various FAEs from resistor manufacturers, who were specifically asked this question. They were clear about the arrays having no particular advantage, unless specified so explicitly in the datasheet.
They had technical arguments that made sense (too extensive to reproduce here), so they convinced me.
Now, statistically speaking, arrays might hold a advantage, but no guarantees.
Cristian
-
The transistors also need to be matched.
-
The resistors are easy to match, just place a small value pot with the wiper connected to the +5V and its two legs each connected to the top of one of the 20k resistors and adjust the pot until you have zero offset. The transistors however are the hard part and require matching over the temperature range of the circuit. They should be a well matched pair in preferably a single package like a MAT02. Otherwise if you can afford some low frequency roll off you can simply AC couple to remove offsets.
-
I should have mentioned that I am using a BCM847 matched transistor IC for the transistors, so those should be matched already (whether that matching is good enough, I'm not sure).
-
They are not bad, max of 2mv offset, but any offset in the transistors gets multiplied on the collector load resistors. It is audio, so AC coupling is probably the best and simplest solution.
-
Yeah, I didn't really get into it, but I am using this for the input into a Thru Zero Frequency Modulation input of an oscillator, and AC coupling definitely helps, but when the VCA opens quickly you still get the offset coming through the AC coupling before it settles back to the correct value. This results in a pitch shift whose envelope timing can be tuned with the AC coupling filter, but with quick changes of the VCA those will get through the AC coupling. For this reason absolutely minimal offset isn't really needed, just offset that doesn't change much as the VCA opens.
-
Mismatch with the transistors acts similar to mismatch with the reistors. So very good matching of the resistors only makes sense if the transistors are also good. One could really consider having a trimmer as this can correct for both the resistors and the transistors.
For the resistors, just 1% grade resisistors from the same role are often quite close, and it gets better with 2 or 3 resistors combined each.
-
Unfortunately with the single differential pair you are going to have substantial DC shifts on the collectors of the diff pair depending on gain/current, that then has to be rejected by the CMRR of your high gain differential amplifier, now that is supercritical on the matching of the resistors around it. What you really need is a doubly balanced differential pair, like the LM1496, to get rid of those DC shifts.
-
You be better off using a VCCS in the tail. This will reduce DC feedthough from the control amp. You can can add a cascode transistor in the tail to set the DC operating point independently of the tail current.