Questions on drift and tolerance

[SmallWalrus]

Hi guys,

I've been parts shopping for a little project that I'm looking to build, and I got stumped by this. The build requires a very exact matched pair of capacitors giving as close a total capacitance as possible to 1020pf (within 0.5% tolerance).

I noticed that RS has some silver micas that are rated to 1% tolerance for 1000pf which I can then combine it with another small silver mica rated to 0.5% tolerance, and get my 1020pf.

Alternatively, I can grab a LC meter and head down to the nearby electronics shop and cherry pick out some silver micas that are rated at 5% tolerance from their shelf. The store silver micas are much much cheaper.

My big question stems from the concept of tolerance and drift. Is tolerance only relevant for the actual measurement of a component with respect to it's rated value? Or will a component that was originally built to tighter tolerance be less susceptible to drifting as well?

I guess my queries can be summed up as, "Will I score a better deal (and not compromise anything) if I cherry pick out some nice 1000pf caps from the store, as opposed to getting the more pricey ones from RS?"

[amspire]

Capacitors are not the most stable of components in general which is why 1% tolerance and less is fairly rare.  Also capacitors change in capacitance with frequency, so a precision reference capacitor of 0.01% or better will specify a frequency and temperature.

Silver mica can be one of the more stable capacitors, but there are many different grades of mica, and you don't know what grade is in the 1% caps. But it is probably going to be as good a capacitor as you can get at an affordable price. 1020pf is only 2% above 1000pf so you will probably want a range of low values to get the caps to exactly match. Using 5% NPO ceramic capacitors in the 10pf to 30pf range would be fine - paying for a 0.5% 20pF capacitor is a bit ridiculous as the final value will change with simple things like the lead length.

To get them matching, you could use a high resolution LCR meter, but otherwise just make a simple bridge with two matching resistors in one leg and the two capacitors in the other leg. There is a bit of technique getting a really accurate balance from a bridge circuit, but it is definitely possible to get a super match with fairly simple equipment.

Richard.

[BravoV]

To get them matching, you could use a high resolution LCR meter, but otherwise just make a simple bridge with two matching resistors in one leg and the two capacitors in the other leg. There is a bit of technique getting a really accurate balance from a bridge circuit, but it is definitely possible to get a super match with fairly simple equipment.

Richard, would you mind share & explain more bout this "fairly simple equipment" setup, maybe a simple sketch ? Really curious here.