0

If I have some resistors and a fixed-gain current shunt amp, that all claim "1%" accuracy can I make a system that's better than 1% accurate, if I put a calibration tuning pot into it, and adjust it manually while measuring it? Will it be better than 1% and repeatable if I do that?

What I'm after is something that's both accurate and repeatable - that once calibrated, it remains in that higher-accuracy state for a long time. What parameters am I looking for in datasheets that will explain how long-term repeatable such a setup would be?

LeoNerd
  • 317
  • 4
  • 12
  • See http://electronics.stackexchange.com/questions/8633/how-viable-is-it-to-just-use-1-resistors-and-calibrate-out-the-error?rq=1 : you can improve accuracy, but not precision; and remember to compensate for temperature and recalibrate from time to time. – pjc50 Oct 08 '15 at 12:32
  • It all depends on the components all of them and you have posted a circuit or links. – Andy aka Oct 08 '15 at 13:17

3 Answers3

5

Keep in mind that anything adjustable will eventually require (re)adjusting.

The actual accuracy of a complete circuit has to take into account tolerance buildup; all the part tolerances have to be assessed. Certainly, an adjustable part (potentiometer in your case) can increase the accuracy of a circuit, at the expense of needing to go through the same procedure regularly.

Your 1% resistors will have an initial resistance of 1%, but over time and temperature, this is more likely to be of the order of 3% (I work in an industry where we support equipment for decades so such an analysis is important for us).

Sometimes it is worth adding circuitry to add an Automated Calibration feature that could perhaps invoke at power-up or on some regular basis, although that would be bit over the top for a simple resistive network. This is useful for high end test equipment, though.

The simple answer - yes, you can get better than 1% (provided your measurement equipment is better than 1%). The general rule of thumb is that to achieve a particular accuracy, the measurement equipment should be 10 times as accurate(but this varies depending on who you ask).

Peter Smith
  • 21,923
  • 1
  • 29
  • 64
  • 1
    +1 for the mention of tolerance buildup. 10 times as accurate is hard to get in some places, low currents are horrible to get accurate as it seems, only the Keysight 3458A has an acceptable 1 year stability for our products and even that doesn't get to 10 times. – Arsenal Oct 08 '15 at 13:53
  • 1
    @Arsenal: I must agree on the comment on measurement equipment that falls into the 'precision, tiny' values. Some of the truly precision kit (and less than 10x as accurate) can have a truly eye-watering price tag as we well know. – Peter Smith Oct 08 '15 at 13:58
  • To see a degradation from 1% to 3% seems like a lot given that metal-film resistors have tempcos of just some dozens of ppm. (E.g. looking at http://www.rapidonline.com/pdf/551117_v1.pdf, the worst ones have 200ppm) Is it mostly aging that you've included in that? – Fizz Oct 08 '15 at 14:01
  • 2
    @Respawned Fluff: We have to account for tolerance over life and temperature in harsh environments (avionics in an unpressurised bay for some kit) for a life of perhaps 30 ( or more in some cases) years. I agree that over an ordinary equipment life span and environment it would not necessarily be that high, but I use it as a starting point on the grounds it can only get better. – Peter Smith Oct 08 '15 at 14:05
3

1% resistors are called that because their actual resistance is within ±1% of the label value, not that their resistance will vary (over time) by ±1%.

So you could use lower spec resistors with your calibration tuning pot and tune for increased accuracy.

All resistances will change value with temperature. There's a small table of temperature coefficients here; I don't know if it's relevant: (http://hyperphysics.phy-astr.gsu.edu/hbase/Tables/rstiv.html#c1)

nickandrew
  • 41
  • 2
2

Long term stability is given under the term of drift, long time drift, long-term stability or sometimes just stability. Problem with these values are, that most of the time each manufacturer gives a value at different operating conditions, making it hard to compare products.

Vishay and others published quite good articles on different resistor kinds and their drift performance:

Long term stability is mostly influenced by temperature and the load of the resistor (self heating, electromigration might be a thing). So a resistor which is used always used under 70°C and high humidity will age quicker than a component at 25°C and moderate humidity. This is a different effect than just your normal temperature coefficient, because it won't go away if the temperature returns to normal conditions.

The kind of precautions to take surely depend on where the device is going to be used. Is it expected to run in a laboratory under controlled conditions? You can probably get quite good results without too much trouble. Is it going to be used in some waste water plant? You better take every possible way to stabilize the temperature and prevent moisture from getting in.

I'm not sure how long term stable the potentiometer will be, but I can imagine that a cheap pot will make things worse over time.

So use a trimmer potentiometer as those are meant to be used the way you want to use it (one time setting, seldom recalibration).

There are of course good reads on trimmer pots as well:

Glorfindel
  • 1,245
  • 3
  • 15
  • 20
Arsenal
  • 17,464
  • 1
  • 32
  • 59