DE-5000 LCR Meter Measurement Accuracy of Small Capacitors

[dermeister]

I just got my DE-5000 and have been measuring some SMD capacitors I had lying around with the tweezers. It works fine for capacitors above ~15pF, however, below that the values are quite a bit off. Here are some measurement results at 100kHz:

Specified: 1.0pF +/- 0.25pF, Read: 1.95pF
Specified: 1.5pF +/- 0.25pF, Read: 2.41pF
Specified: 6.8pF +/- 0.25pF, Read: 7.84pF
Specified: 12pF +/- 5%, Read: 13.01pF

The claimed accuracy of the meter is 2.0%+5 in the 200pF range using a 100kHz test signal. Tweezers were calibrated, all of these capacitors were fresh from Digi-Key and none of the measurements are within spec. I realize that measuring such small capacitance values can be tricky, but I was expecting better than a 50%+ error in the 1-2pF range. Can anyone verify these deficiencies in accuracy? Is a meter like the BK Precision BK880 better suited for such measurements since it has a 4pF range (albeit with no accuracy specification for it)?

[precaud]

Did you do an Open/Short cal with the tweezers in before measuring them? The Open cal will have a large impact on the accuracy for small-value caps.

[dermeister]

Yep, open/short calibrated the tweezers multiple times, always with the same results. Hooking the guard line up to ground reduces the reading by about 0.2pF, but there always is a ~0.6-0.9pF offset to the real value. Is there some kind of hidden calibration menu where the meter is calibrated at the factory?

[Mr. Scram]

I'd be interested to hear about this too.

[glarsson]

Add a capacitor around 20pF in parallel and subtract it's measured value.

[Carrington]

Try again, but now use something like this (see below) to perform an open an short/calibration.



The distance between the tips should be as close as possible to the component dimensions.

[indman]

DE-5000 - the excellent measuring instrument for small capacities, if is correct to use him.
Here my results of measurements of small capacities
1. Ceramic tubular 0.47 pF
2. KT-1 1.5 pF ± 0.4% 250B TKE P100
3. KT-1 2.2 pF ± 0.4% 250B TKE P100

[Fgrir]

Try again, but now use something like this (see below) to perform an open an short/calibration.

The distance between the tips should be as close as possible to the component dimensions.

I think this is your answer.  If I cal my DE-5000 with the tweezers fully open and then slowly squeeze them to almost closed I can watch the capacitance reading work it's way up to about 1pF, which would more or less explain the OP's offsets.

[Teuobk]

Try again, but now use something like this (see below) to perform an open an short/calibration.

The distance between the tips should be as close as possible to the component dimensions.

+1 to this. If you don't do the "open" part of calibration with the tweezer tips separated by the length of the SMD part, the resulting error will dominate in measurements of small values.

When I calibrate my DE-5000 correctly, it will easily meet its specs even on 0402 caps down in the neighborhood of 1.0 pF.

Here's a thread where I go through a measurement of a 1.2 pF 0402 cap with my DE-5000. The DE-5000 read 1.19 pF. The cap tolerance was +/- 0.1 pF, and the DE-5000 uncertainty under those measurement conditions was about +/-0.07 pF, so the DE-5000 performed well within its specs.

[cvanc]

I think this is your answer.  If I cal my DE-5000 with the tweezers fully open and then slowly squeeze them to almost closed I can watch the capacitance reading work it's way up to about 1pF, which would more or less explain the OP's offsets.

+1 to this. If you don't do the "open" part of calibration with the tweezer tips separated by the length of the SMD part, the resulting error will dominate in measurements of small values.

(mind=blown)

Wow, I never thought of that.  It makes perfect sense now that someone's explained it to me.  Cool stuff, guys; thanks.

[Mr. Scram]

(mind=blown)

Wow, I never thought of that.  It makes perfect sense now that someone's explained it to me.  Cool stuff, guys; thanks.

Same here. At first I didn't get what the cal bar is for and thought it was conductive, but now it makes perfect and intuitive sense. I just never considered the open circuit terminal distance to play a role. I now also understand why people don't want to move the cables of certain devices around.

[korlatos]

Does anybody know where those calibration bars are available for purchase?

[dermeister]

Try again, but now use something like this (see below) to perform an open an short/calibration.

The distance between the tips should be as close as possible to the component dimensions.

+1 to this. If you don't do the "open" part of calibration with the tweezer tips separated by the length of the SMD part, the resulting error will dominate in measurements of small values.

When I calibrate my DE-5000 correctly, it will easily meet its specs even on 0402 caps down in the neighborhood of 1.0 pF.

Here's a thread where I go through a measurement of a 1.2 pF 0402 cap with my DE-5000. The DE-5000 read 1.19 pF. The cap tolerance was +/- 0.1 pF, and the DE-5000 uncertainty under those measurement conditions was about +/-0.07 pF, so the DE-5000 performed well within its specs.

Yep, this works. Reading is spot on now! I'd also be interested in where to get one of these calibration bars.

[cvanc]

I am really impressed that this inexpensive meter can reliably read all the way down to 1pF.  I've got one and a while back it impressed me just sorting out some 20pF from 22pF, I thought THAT was living on the edge.  Live and learn.

Good thread, folks.

[TERRA Operative]

Does anybody know where those calibration bars are available for purchase?

I too am interested...

[Carrington]

I have no idea of where you can get just the cal. bar. But it's only a piece of plastic and gold plated metal.
http://www.ezurentals.com/Product/Model.Photo-Agilent_16334A.aspx

[ogden]

I have no idea of where you can get just the cal. bar. But it's only a piece of plastic and gold plated metal.
http://www.ezurentals.com/Product/Model.Photo-Agilent_16334A.aspx

IMHO could be fabricated from two or three 1.6mm FR4 triangle boards superglued together (obviously with help from atlignment holes). Maybe ENIG or even tin plating (instead of hard gold) shall be considered because out of 10pcb * 10x10cm order you will have many, many spares which most likely will outlast your RLC tweezers. Sounds like group buy project?

[cvanc]

Sounds like group buy project?

It does.  I'd buy one or two.  But it should be stepped like a staircase, not tapered like a triangle.  More repeatable that way.

[ogden]

Sounds like group buy project?

It does.  I'd buy one or two.  But it should be stepped like a staircase, not tapered like a triangle.  More repeatable that way.

Staircase looks better to me as well.

It could be one or all options: 1) someone (not me) do all the job and prepare gerbers/instructions for one or few PCB manufacturers, "buyers" send him "thank you" money in PayPal and just download gerbers, use as please 2) seedstudio project which most likely is most expensive 3) pure "group-buy" where someone (not me) is doing all the shipping hassle, but it's pocket business for him as well