Capacitance measurement with limited gear

[dannyayers]

For a project I'm working on I'd like to use reasonably accurate (say 1%) capacitors, which I hope to achieve by hand-picking (and if necessary combining) regular capacitors - I believe the film ones I've got are 10%. But I'm a little limited in the equipment for measuring them.
Ballpark range, say 100pF - 1uF (to be decided).

I do have a 'scope (a Bitscope - nice little bit of Aussie kit!) and a cheapo multimeter with a capacitance range, but I'm not sure how much I can rely on them for accuracy.

Looking around, most designs I see fall into one of two categories: a traditional bridge or an (often Arduino-based) timing-based setup.

But I have a problem with each. The bridges intended for capacitance measurement all seem to feature a capacitor in two arms, and I don't have an accurate one for reference. The timing-based setups all seem to assume the (say 5v) charging voltage is accurate - and I don't have an accurate voltage reference.
But it should be possible to get pretty accurate timing from something like an Arduino, and I have ordered some 0.1% 10k resistors, so I've at least got something to work against.

I've only just started looking at this, and I could be way off the mark, but it feels like it should be possible to use a bridge with 3 arms resistance plus the cap under test (driven by AC). Alternately, going the timing approach, rather than assuming an accurate voltage and taking a single point in the charging curve, take 2 or more readings, work the sums from there. (Hmm, noticed a flaw there as I type - the voltage measurements would need to be accurate as well as the timing).

So I'm wondering how y'all might go about it.

This isn't at all a critical thing, but it does make a nice little exercise in figuring out how to bootstrap test gear.

The project is an analog computer (https://github.com/danja/analog-computer), mostly a pile of op amps, the capacitors needed for the integrators. Even though I've been thinking about it for a long time, only recently gathered the components, got a load of breadboarding coming up next.
I'm thinking of using standard 1% resistors, maybe hand-pick, but gives me a baseline for accuracy.

So, suggestions?
 

[Conrad Hoffman]

You can buy close trimmed voltage references for not too much money, so you can get voltage. Crystals and some radio stations will give you an accurate time reference. Resistors are fairly easy to get at 0.1% tolerance, so in theory you have everything to measure capacitance. In practice, it's probably a pain, but worth trying. Bridges only need one capacitor. You can probably get silver-mica caps at 1% and maybe 1/2%, so that route is open.

[bob91343]

The easy way is to pick up an old bridge.  I have several that were cheap.  In fact the Chinese 'magic box' I have appears to be within about 1% and of course can match capacitors much more closely.

https://www.ebay.com/itm/Multifunction-Transistor-Tester-Diodes-LCR-TC1-Full-Color-Graphics-Display-New/114248283517?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2648

[Henrik_V]

Soundcard?  AC Generator and AC Input ...  All it needs for an AC bridge
Even better with an high impedance  (low cap) sense input (but any OP is fine for testing) 

and some known Rs (means value and uncertainty )

[David Hess]

Given the limitations, I would place the capacitor in an RC circuit with a precision resistor and then measure the frequency where the phase shift reaches 45 degrees, and then calculate back to get the capacitance.  Phase and frequency are much easier to measure accurately than amplitude.

[JohnPi]

You don't need a precise voltage reference to measure capacitors -- at almost a relative voltage reading is needed. Fundamentally, you need either another capacitor (to compare against and scale), or a resistor and time (frequency) reference.

Given that you are making integrators (and consider that measuring values as low as 100 pF directly will introduce all sorts of difficulties with parasitics), why not use the integrator directly ? Apply a square wave input (V) to the input resistor (R), and measure the slew rate of the output (dV/dt). Then C = V/R/(dV/dt). If you measure with your oscilloscope, its voltage calibration doesn't matter. The integrator configuration will generally be quite immune to parasitic capacitances.

John

[SilverSolder]

Another idea is to just use one of the cheap Chinese component testers.  The absolute accuracy isn't stellar, but they can definitely be used to compare two capacitors to see if they are within 1% of each other, as long as you test both capacitors within a few minutes.

[CaptDon]

I had an old 'bridge' style unit with a magic eye tube. The ranges were in decades such as .1uf to 1uf across a 270 degree dial face. You could certainly tell within 1 or 2 % what you had. Every range had a range adjust internally for accuracy. We used it in an R&D lab.

[bob91343]

I have one of those old bridges with the eye tube.  I'd be happy to get rid of it, as it never gets used because I have better instruments for any function it can provide.

They aren't terribly accurate and the leakage test (especially) and power factor test are not well calibrated.  For casual measurements and comparisons they are fine.  Mine even has a comparison range but again, not too precise.

[David Hess]

I have a pair of ESI 250DA impedance bridges and they work great but they are not very quick to use.

[Henrik_V]

A low cost capacitive bridge based on voltage drop balance
https://www.sciencedirect.com/science/article/abs/pii/S0263224110001089


In short an AC half-bridge build by a known resistor and unknown cap,
vary the frequency until the AC-voltage on both elements is equal
measure the frequency , apply math

the paper states 0.1% accuracy , limitetd by the AC- measurement (need high impedance, low input capacity)

[Conrad Hoffman]

Nice idea and very simple. Even simpler if you use your bench sig gen and scope. Not sure how much you lose if you don't add a dissipation factor adjustment of some sort (not sure how).

[SilverSolder]

That setup would work for the 45 degree angle trick as well (varying the frequency to do it)