Capacitance Measurement

 

I wanted a quick and easy test to check for small capacitance in the pF range, and inductors in the nH range for that matter.  My current equipment consists of a NanoVNA H4 and an off brand LC200A LC meter.  The NanoVNA is a great tool but don’t like bringing out, setting up, and calibrating just to make a quick measurement.  The LC200A was $30 (at the time) and I had my doubts on accuracy in the lower ranges.  Searching around I found an LCR tester around $70 (Amazon) from FNIRSI called the LC1020E.  This seemed like it could be a nice LCR meter and it even included Kelvin clips for in circuit measurements.  Unlike the LC200A the FNIRSI looks and feels well built.

The NanoVNA, in my opinion, is by far the most accurate.  The draw back is the setup involved. For the NanoVNA I referenced a YouTube video from Electronics for the Inquisitive Experimenter (this and his other videos are a must watch!) – https://www.youtube.com/watch?v=Pti8Erw_Kkg . You can read capacitance directly on the VNA as follows:

  • In Stimulus set the Start to 40MHz and Stop to 50MHz
  • In Display turn off traces 2,3, & 4
  • In Display set the first Format to SERIES C (press MORE 2x to get to this part of the menu).
  • Rotate the marker to 45MHz
  • (Calibrate – if not done already)
  • The capacitance will show up in the upper left corner of the screen.

A quick note about NanoVNA-Saver:  I typically use the NanoVNA-Saver program which putting the screen on a PC is easier on the eyes.  Occasionally I see spurious/erratic data and complete crashes, especially after a program update.  Clearing the cache fixes these problems.  Typically in Linux I will delete the NanoVNASaver folder in the /home/.cache area.

A final mention on the NanoVNA.   In the video above 40-50MHz was used as a test frequency range.  But what about a lower frequency around 100KHz-1MHz like the other meters use?  I tried the same procedures on the Nano from 100KHz-200KHz and arrived with erratic results like negative capacitance and unstable readings!  Since the Nano is only good to around 50KHz this is probably a lower limit issue.  Other attempts at 1-2MHz and 10-11MHz resulted in very similar results as the 40-50MHz tests but did noticed at 1-2MHz the value seemed to jump around a bit more.

The LC200A has worked great for me in the past and claims a resolution down to .01pF.  The accuracy for a small capacitance, like 18pF, reads OK but over time it will drift and periodically have to re-zero the unit.  For example in our 18pF experiments below leaving the capacitor in the fixture will result in a read over 22pF after several minutes.  Zeroing out the meter and leaving run will result in a positive drift of around 0.15pF/minute.  Below is a shot of drift after about 15 minutes.

Enter the LC1020E….. Will it do the job?  I guess we will see in the results below.  After we get set up and calibrated we’ll check out an 18pF capacitor then go down to 4.7pF, 1.0pF, .82pF. and finally .68pF.

Trying to compare apples to oranges……..

The LC200A uses alligator leads for measurement, which can vary small capacitance quite a bit down in the pF levels as you move around.  The NanoVNA will need a fixture of some sort, and I’m skeptical how good a 24″ long Kelvin leads will be on a couple pF or less capacitance on that LC1020E (Hint: they do indeed change value as you move them around).  In the past I made up an SMA to alligator leads for the NanoVNA which was OK but I could see capacitance changes as the leads were moved around and wanted something as little more stable.  So let’s build some fixtures!

The fixtures consists of banana jacks, terminal blocks, and a piece of prototyping board.  Stack-able banana jacks were used and easily removed from the plastic housing.  M3 screws, washers, and locking nuts were used to secure the jack to the perf board as well as a point to attach a small piece of wire going to the terminal blocks.  The jacks came from Tayda (A-1824 & A-1825) but can also be found on Amazon, Ebay, etc.  The terminal blocks were 5mm DG300 style (A-666) which happen to be spares in the junk box.  If I had the part on hand I would rather use something spring loaded like speaker wire terminals (such as a Tayda A-6895).  For the NanoVNA a male SMA PCB connector was in the junk box.

 

Set up and Calibration

Now that all 3 units have their fixtures in place the units need to be calibrated:

  • The LC200A was easy by pressing the zero button with nothing in the fixture.
  • The LC1020E was open/short calibrated per the manual with the fixture in place.  One thing I noticed was the meter was finicky reading small capacitance in auto mode.  The solution was to place it in capacitance mode, setting  range to 100K or 10K ohms, speed to Slow, and the second measurement as X (impedance).  The Frequency depended on capacitance measured (see tests).
  • The NanoVNA calibrated with the open/short/load calibration procedure (at the text fixture – do not use the SMA ones!) and set up to read series capacitance.  The short test was a very small piece of wire, enough to make good contact with the terminal block.  a 49.9 ohm 1% resistor was used for the load test.

18pF test

We’ll start with something easy with an 18pF +/-2% 100V NP0 ceramic capacitor.  We took 3 capacitors from the box and ran them through each meter.  Each capacitor should read between 17.64 and 18.36pF.  The lead length matters! Cutting the leads down enough to fit in the fixture brought the capacitance on the NanoVNA down from 18.60pF to 18.35pF, that was only a mere 0.3″ (7.6mm) of lead removed.  Another capacitor trimmer went from 18.54pF to 18.26pF.  As you can see what was out of tolerance due to the lead length was now within tolerance.  All capacitors was trimmed to length before measuring.  The LC1020E performed best at 100KHz.  A bit disappointing is the FNIRSI only has a display down to .1pF.  However, using the second measurement (X) and calculating with the formula C = 1/(2*pi*F*C) gives us a more precise reading.  Maybe FNIRSI will add a couple more digits in a firmware upgrade?  All three were pretty close with the NanoVNA closest.

LC200A LC1020E (and X calculated) NanoVNA
1 18.50pF 18.7pF   -85.32Ω -> 18.65pF 18.34pF
2 18.68pF 18.8pF   -84.67Ω -> 18.80pF 18.51pF
3 18.65pF 18.9pF   -84.25Ω -> 18.89pF 18.43pF

 

4.7pF test

Next up a 4.7pF +/- 0.25pF 60V NP0.  Each capacitor should read between 4.45pF and 4.95pF.  The LC1020E performed best at 10KHz.

LC200A LC1020E (and X calculated) NanoVNA
1 5.29pF 4.2pF   -3.76MΩ -> 4.23pF 5.06pF
2 5.35pF 4.3pF   -3.72MΩ -> 4.28pF 5.09pF
3 5.45pF 4.3pFpF  -3.661MΩ -> 4.35pF 5.18pF

1.0pF test

Next up a 1pF +/- 0.25pF 63V NP0.  Each capacitor should read between 0.75pF and 1.25pF.  The LC1020E performed best at 10KHz.

LC200A LC1020E (and X calculated) NanoVNA
1 1.64pF 1.8pF   -8.71MΩ -> 1.83pF 1.37pF
2 1.63pF 1.8pF   -8.78MΩ -> 1.81pF 1.35pF
3 1.63pF 1.8pF   -8.83MΩ -> 1.80pF 1.33pF

0.82pF test

Next up a 0.82pF 60V NP0 from Stetco.  We don’t have a tolerance but most of my Steco marked boxes are +/- 0.25pF so will assume this tolerance.  Each capacitor should read between 0.57pF and 1.07pF.  The LC1020E performed best at 10KHz.

LC200A LC1020E (and X calculated) NanoVNA
1 1.42pF 1.7pF   -9.11MΩ -> 1.75pF 1.15pF
2 1.46pF 1.7pF   -9.21MΩ -> 1.73pF 1.20pF
3 1.47pF 1.7pF   -9.09M -> 1.75pF 1.20pF

0.68pF test

Next up a 0.68pF 60V NP0 from Stetco.  We don’t have a tolerance but most of my Steco marked boxes are +/- 0.25pF so will assume this tolerance. Each capacitor should read between 0.43pF and 0.93pF.  The LC1020E performed best at 10KHz.

LC200A LC1020E (and X calculated) NanoVNA
1 1.32pF 1.7pF   -9.56MΩ -> 1.66pF 1.05pF
2 1.34pF 1.7pF   -9.41MΩ -> 1.69pF 1.11pF
3 1.31pF 1.7pF   -9.57MΩ -> 1.66pF 1.07pF

Parallel Testing  18pF + 0.68pF

In the under 4.7pF capacitors all the values on the testers were fairly high, not unexpected due to stray capacitance.   The last test was to use the LC1020E and pair up the 18pF and 0.68pF together by first measuring the 18pF (@100KHz) then putting the 0.68pF in parallel and taking the difference to read the actual 0.68pF value.  This seemed to work fairly well.  This old school technique still holds true today!  A similar test was also done on the LC200A and the NanoVNA with similar results in the 0.70pF to 0.75pF range (not shown).

18pF Cparallel Difference
1 18.3pF   -87.21KΩ -> 18.25pF 19.0pF   -83.99KΩ -> 18.95pF 0.7pF (0.70pF)
2 18.4pF   -86.56KKΩ -> 18.39pF 19.1pF   -83.23KΩ -> 19.12pF 0.7pF (0.73pF)
3 18.4pF   -86.66KΩ -> 18.36pF 19.1pF   -83.2KΩ -> 19.13pF 0.7pF (0.77pF)

And the results are………….

  • Small capacitance is hard to measure.
  • Keep leads as short as possible.
  • Use a fixture to keep measurements stable (no wiggly wires!).
  • When measuring small values use a larger known capacitor in parallel around 10x-25x the smaller value.
  • The NanoVNA is still champ!
  • The LC200A did much better than expected.  Just remember to zero occasionally to counter drifting.
  • I little disappointed in the LC1020E with reading the lower values and only 1 decimal display but came through during the parallel test.  For the most part the LC1020E will still replace my LC200A even with this minor shortcoming.