C measurement tool

[jh]

When I bought a (rather cheap) new multimeter some years ago,
I made sure that it had capacitance measurement (as well as
frequency, inductance and hfe - all very limited).
Is it accurate ? No. The capacitance is always a few pF too high,
and there are probably a few per cent of error anyway.
Does it hurt ? No. In fact, it's a blessing to select matched
capacitors. Not important if both are 1nF or 1.02nF or even
1.1nF as long as they are equal. And the Multimeter provides this 
information.

BTW, my first method to "measure" capaitors was using a transistor
tester (high impedance AC source). I would connect the capacitor in 
question and see how bright or how dim the LEDs would shine.
Precicion? Not at all. But it helped me to decide if a cap marked "100"
was 10pF, 100pF or 100nF. Hmm, for deciding on 10pF or 100pF
I had to put the whole thing under the cover of my bed, to se the dim
light of the LED. (A frequency range switch would have made it easier,
but it worked,so why change it.)
Next method was building such a C measurment kit (either based on
Elektor or Elrad - don't remember). Good circuit, but separate box to be
connected with the mA input of a multimeter. Guess what ? I *never*
used it.
Now I have the multimeter where I just plug a cap into a slot, and there
it is, the (approximate) capacitance value. 

JH.

-----Original Message-----
From:	Martin Czech [SMTP:martin.czech at intermetall.de]
Sent:	Monday, April 19, 1999 9:58 AM
To:	synth-diy at mailhost.bpa.nl
Subject:	C measurement tool

Some design need low tolerance caps, so one has to measure...

I don't want to build a bridge, it is expensive and not very easy to
use. Bridges have means to adjust for cap loss, so many controlls have
to be adjusted at the same time.

Since such demanding applications need good caps polystyrene or the like)
we could assume them to be ideal.  This is a hypothesis.  This would
allow to measure the RC time constant.

I think this is the way that cheap multimeters do it.
But these meters are very limited (this starts with the
mechanical parts, clamps).

There was a cicuit in Elektor:

A regulated voltage, a known R and the C (DUT), a window comparator
(counter window), and a discharge logic.  The window comparator gives
an estimation of charge time, so a quartz counter is needed to measure
this time.  The discharge time is very long, so the whole circuit is
some kind of oscillator, but the frequency does not matter.

The charge time should be in the 1ms range, to avoid comparator swithching
time influences etc. The cicuit has some stray capacitance,
maybe it is possible to add a one shot that shortens the window
in order to trim out these stray effects.


Are there any experiences ? What possible errors could limit the accuracy?
Series R and L? Stray capacitance?

m.c.