Measuring low capacitance accurately

Jump to Latest
noviygera

noviygera

Member
Joined 2009
#1
Hey guys, I need advice on how to measure accurately, to match some 10nF caps. This is for a phono stage project.

My Fluke 87 (in relative mode) let me down, or so I think it has. Every day I measure the values go up and down a little -- inconsistent. Is it temperature related or what?

Should I use another method -- the freq. generator with resistance? If so, I need need to purchase one. Anyone try the ones on Amazon ~$90usd? Are they ok?
What to do? Should I better invest in a scope?

thanks!
 
#3
Even my cheapy Aneng 8009 measures capacitance with a resolution down to 1-10pF.

Are you requiring matched caps to each other, or to an absolute accuracy value (which seems to be your concern)?

Some spectrum software used with soundcards have quite accurate audio frequency generation and identification. If you were using 1% resistors, and had matched them to better than 1%, then you could make up a filter that determines the capacitance knowing R and f. You could also check filter response change for changed temperature of capacitors, and resistors. etc.
 
#5
PRR,
values change from day to day. Ex: 9.8nF to 10.2nF, using the same testing technique, same room, same caps on the same Fluke 87.

trobbins,
>>Are you requiring matched caps to each other
yes.
 
Last edited:
#6
Same room temperature and humidity? Have the parts been moved or being held differently? Or did you use clip wires on the probe ends and then not touch the setup other than turn the meter off and on. SO tomorrow, you can carefully turn the meter back on without budging anything else from the day before
 
#7
You can measure with computer audio out ---10k----+---C1---gnd
+---C2---open
At 2748Hz the voltage at the 10n cap should be 1/2 the sound-card output. (F-3 should be 1591.5Hz)
To compare them ground alternately.
 
Last edited:
#13
> If you want to "match" two caps,

This is "a phono stage", and probably stereo.

So two steps:
1) 'match' caps and resistors for proper EQ
2) match left and right

He's seeing +/-4% wobble day to day. That's maybe 0.3dB error in RIAA. I remember when that was fantastic but today people like to see 1/4dB.

Is it the meter, or the caps?

What type are the caps?? Some are more stable than others. High-value ceramic is awful. <100pFd ceramic is typically "perfect" (unless a tempco is given).
 
#16
I checked 3 capacitors yesterday and today with a 22 year old Fluke 87 and an inexpensive 2 year old LCR meter DM4070.

Friday Saturday
Fluke DM4070 Fluke DM4070
3.05uf 3.05uf 3.05uf 3.05uf actual cap 3.0 uf 250V film
.098uf .099uf .098uf .099 actual cap .082uf film
.00252uf .00222uf .00252uf .00222 actual cap .0022 mica

note the Fluke when reading the .0022uf cap fluctuated between .00245 to .00258 on both days. Short or long leads the Fluke read around 340-360pf with no capacitor and leads open. The DM4070 read 0.00uf with or with out leads.

I dug out the fluke manual and it says readings under .005uf may be noisy and to use short leads or to use the REL feature to subtract the stray reading from current reading.
 
#18
One of the problems in measuring capacitance is dielectric absorption - the capacitance varies with frequency, and the effect is voltage dependent too. So its usually measured at a standard agreed frequency, depending on the intended use. Unless two measurements are taken at the same frequency (and amplitude) they are not strictly comparable.


And of course self-inductance affects the measured capacitace at higher frequencies too.
 
#19
jackinnj said:
DIYAUDIO's own Conrad Hoffman wrote up an easy to use Capacitance Bridge: http://www.conradhoffman.com/cap_bridge.pdf
Click to expand...

Conrad's bridge will measure both C and D.

For measuring the caps for an RIAA network or XO, instead of a "standard" capacitor, use one of the caps and sort through the batch until you get a pretty good match.

His simple design, he makes clear, is just like the GenRad 1608.
 
#20
You must log in or register to reply here.
Top Bottom