How should I use a trimmer potentiometer to tweak a ratio of resistors

[TheUnnamedNewbie]

(click images for high resolution versions)

So I have a amplifier of which I want to set the gain to very precisely 2 times. I have a feedback network that controls this.

Because temperature coefficient reasons, I want to use two fixed resistors (0.1 %, 15 ppm/K) to set the gain and then use a small trimmer to just nudge it the right way.

My reasoning was that by making the trimmer very small, it's temperature coefficient will have less impact on the set value. If I have two 0.1 % resistors, then I know that my trimmer needs to only trim over a fraction (0.1 %) of the value of the two other resistors.
By my problem is the way I should wire the resistors up. The simple example would just be as follows:


However, I was told it was bad practice to use this technique, as during the movement of the wiper the trimmer might go open, and your circuit becomes open loop. This causes the op amp to potentially rail and damage the circuit that follows (which isn't really an issue in this case though).

Alternatively you can do the following:

However, in this case you can only tweak in a certain direction, so you would need to make the two resistors have different sizes so you know your ratio will start out too low and then use the trimmer to trim it into the right ratio. This isn't possible in my case because the size of the trimmer is then dictated also by the difference between the two fixed resistors, which will be a few percent because of the nature of E12 and E24 series. And since most of these low tempco resistors are only available in E24 series resistors, I can't get the size of the trimmer as low as I would like it. In addition, it's annoying on the BOM because you now need two types of resistors instead of one.

So what is the "right" way to do this?

[Alex Nikitin]

So what is the "right" way to do this?

The second circuit is better, just add another resistor with a half value of the pot in series with the top resistor.

Cheers

Alex

[Gyro]

Surely the first circuit must be better - assuming a high impedance load. The wiper is only used in voltage tap mode rather than having to pass divider chain current (where variable contact resistance, oxidation etc. could become a factor) . Assuming that it's value is small compared to the divider resistors, it ought to have minimal effect on TC - especially if its adjustment point happens to be close to the ration of the divider?

EDIT: I'm thinking of calibration stability here -  If the wiper simply becomes higher impedance then it will still be referenced to that voltage point on the track as seem by the opamp. The possibility of a completely open-circuit wiper could addressed by a much higher value less critical divider attached to the wiper. I suspect that the best configuration depends on how close the divider is to a 2:1 ratio.

[TheUnnamedNewbie]

Gyro, if I understand you correctly you are suggesting something like this, where the Ropen resistors are very, very large, perhaps 2-3 orders of magnitude larger than the trimmer. This way the amplifier will never go open loop, but when the trimmer is operating normally, all the noise and drift of the big resistors is shunted by the in comparison tiny trimmer resistance.


Ofcourse, this would add two more components but the fact that I can use a single component for the 'expensive' main resistors in the network is nice. Perhaps an off-the-shelf network can provide me with a smaller footprint as well. Will have to see!

[f5r5e5d]

one too many R - a classic is a fixed R divider, then the pot ends to the rails and a single R from the pot's wiper to the fixed R divider center

seting pot and wiper series R to give adj range needed at the fixed R divider center

if the divider ratio isn't close to 1/2 then an added R in series with the appropriate leg of the pot to center the adj range

[TheUnnamedNewbie]

one to many R - the classic is a fixed R divider, then the pot ends to the rails and a single R from the pot's wiper to the fixed R divider center

seting pot and wiper series R to give adj range needed at the fixed R divider center

That makes sense, I can see how that works. But would drift in wiper contact resistance and temperature coefficient of the resistor going from the center of the divider to the wiper not cause an issue?

[Alex Nikitin]

one too many R - a classic is a fixed R divider, then the pot ends to the rails and a single R from the pot's wiper to the fixed R divider center

seting pot and wiper series R to give adj range needed at the fixed R divider center

if the divider ratio isn't close to 1/2 then an added R in series with the appropriate leg of the pot to center the adj range

Yes, that is another sensible way to do it and for a good reason, however nothing comes for free, and in that configuration the power dissipation on the pot could be a problem, so the pot value needs to be quite large.

Cheers

Alex

[PartialDischarge]

This post is of relevance here,

https://www.eevblog.com/forum/metrology/fine-resistance-adjustment-for-standard-resistors/

[Conrad Hoffman]

Certainly covered in the above, but for a simple divider one resistor, by design or by chance, will be too high. Determine what parallel resistor will bring it to spec, then use a slightly smaller resistor and a trimpot. Any changes in those parts will be a tiny percentage of the main values.