1

I'm designing this circuit to measure temperature with PT100.

enter image description here

R10 must be calibrated to obtain the desired gain.

I have thought of using a resistor trimmer to adjust the gain similar to the following:

enter image description here

Considering that the humidity and temperature conditions are within the ranges described by the manufacturer, is it a good idea to use this potentiometer?

I mean if the internal resistance can change over time, or the screw can move over time.

I once thought I saw it being used in thermocouple conditioning board, but the screw had a glue on it, which I imagine is to avoid the problem I mention.

Fabián Romo
  • 391
  • 2
  • 11
  • 1
    If you put 0.1 or 1% resistors, you don't need to adjust. Why do you think you need this? – Marko Buršič May 10 '21 at 16:35
  • The gain is a value not close to a whole number, for example for -50 degrees the voltage is 80.31 mV and for 200 degrees Celsius it is 175.856mV. – Fabián Romo May 10 '21 at 16:39
  • 4
    Those trimpots are made for that sort of application. As you have seen, the adjustment is often "locked" to prevent accidental mis-adjustment. – Peter Bennett May 10 '21 at 16:48
  • 1
    Change of the potentiometer setting with vibration isn't a huge concern with trimmers; most of the time when people "lock" the setting it's so that they can tell if someone's been messing with it if they have to go back in and fix it. You can use nail polish or they make tamper-proof sealants specifically for this type of thing. Try searching "torque stripe". – vir May 10 '21 at 16:57
  • 1
    So what? The engineering world is not made by whole numbers. – Marko Buršič May 10 '21 at 16:57
  • 2
    If you can use high-tolerance resistors, they are the most reliable. Otherwise these are used frequently in industry and are very reliable. Remember to consider temperature coefficients if the temperature ever could vary more than a few degrees. – rdtsc May 10 '21 at 17:03
  • @MarkoBuršič Of course, one option could use fixed resistors and the gain and offset could be by 'software' in the microcontroller that is going to perform the processing of the measured signal. What I want is to know opinions since it is the first time that we have a project of this type. Sometimes someone else's experience is better – Fabián Romo May 10 '21 at 17:06
  • 2
    Before you consider **any** "solution" you should first **specify what you need**. Suppose your amplifier is 1% accurate and you're improving that to 0.1%. However, your sensor is only 2 % accurate and you're not compensating for that (using calibration). Then the improvement of your amplifier is useless as the 2% sensor error still determines the overall accuracy. – Bimpelrekkie May 10 '21 at 17:44

2 Answers2

6

If you don't want to do calibration in software (which has its own potential issues), use the trimpots but limit their range by adding other resistors so that they don't have excessive adjustability.

For example, if you have 10K + 10K in that position you could do something like this:

schematic

simulate this circuit – Schematic created using CircuitLab

This is just a rough example, you should do a worst-case error budget including all reasonable factors and give yourself a bit of headroom so the pot is never right at either end. This will also prevent the technician doing adjustments from "calibrating out" a serious error such as the wrong part being populated elsewhere, which can cause subtle issues.

In this case you can easily set the ratio to a few ppm and it should stay within 50ppm or so. If you don't limit the range, the stability may not even be as good as ordinary 1% resistors, and perhaps much worse if you're not using it as a voltage divider.

The "paint" you have seen is typically used to indicate that the unit has been calibrated and to combat "screwdriver drift", so warranty can be voided or at least the issue detected. Since those trimpots are sealed, pretty much any kind of paint can be used, such as nail polish or lacquer.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • Hi, could you clarify why it would be "perhaps much worse if you're not using it as a voltage divider."? I'm also trying to solve a similar issue and I was not aware that using the potentiometer as a variable resistor by shorting 2 of it's 3 pins could cause stability issues. – Joaquin Brandan Oct 08 '21 at 04:38
  • 1
    @JoaquinBrandan When it's not used as a voltage divider (i.e. used as a rheostat) the contact resistance enters into the total resistance directly so (particularly on low value pots) the effect can be significant. The temperature coefficient of the contact resistance is not guaranteed either, generally. It's preferable therefore to use a pot as a voltage divider where possible, and where not to use a relatively high value (such as 10K) and to limit the adjustment range as much as practical. – Spehro Pefhany Oct 08 '21 at 05:01
4

If you use a pot, it should replace both R10 and R11. With both resistors on the same potentiometer track, they will have the same temperature coefficient and there will be very little drift in gain vs temperature. If R10 is a pot, but R11 is a fixed resistor, they'll have different tempcos, so you'll have to do the math to check if it's a problem.

bobflux
  • 70,433
  • 3
  • 83
  • 203