impedance of capacitor using network analyzer

[robismyname]

I measured "Z" of a 1pF capacaitor: passive plus,inc. (0402N1R0BW500) over a frequency range of 2390MHz - 2500MHz

The "Real" part of Z is different by about 10 ohms across the two test parts. Two seperate parts were used from the same reel

Can someone offer a logical explanation to how or why this exist? Perhaps it due to temperature? Perhaps I added more solder on the test fixture to mount the cap than I did in the previous test? Is this a big deal?

W=ohms

Test set 1

Freq
(MHz) Real Imaginary capacitance
2390 20.4 W - 38.4 j 1.7 pF
2395 20.2 W - 37.3 j 1.8 pF
2400 19.8 W - 36.6 j 1.8 pF
2405 18.9 W - 35.6 j 1.9 pF
2410 18.7 W - 34.4 j 1.9 pF
2415 18.6 W - 33.5 j 2 pF
2420 18.4 W - 32.3 j 2 pF
2425 18.2 W - 31.2 j 2.1 pF
2430 17.9 W - 30.2 j 2.2 pF
2435 17.8 W - 29.2 j 2.2 pF
2440 17.7 W - 28.1 j 2.3 pF
2445 17.6 W - 26.9 j 2.4 pF
2450 17.6 W - 26 j 2.5 pF
2455 17.7 W - 25 j 2.6 pF
2460 17.7 W - 24.3 j 2.7 pF
2465 17.8 W - 23.4 j 2.8 pF
2470 17.8 W - 23 j 2.8 pF
2475 17.5 W - 21.9 j 2.9 pF
2480 17.7 W - 21.1 j 3 pF
2485 17.9 W - 21.2 j 3.2 pF
2490 17.9 W - 19.3 j 3.3 pF
2495 18 W - 18.8 j 3.4 pF
2500 18.1 W - 18.1 j 3.5 pF


Test Set 2

W = ohms

Freq
(MHz) Real Imaginary capacitance
2390 30.2 W - 37.4 j 1.7 pF
2395 29.9 W - 36.5 j 1.8 pF
2400 29.5 W - 35.4 j 1.87 pF
2405 29.1 W - 34.5 j 1.92 pF
2410 28.7 W - 33.5 j 1.98 pF
2415 28.2 W - 32.51 j 2.02 pF
2420 27.9 W - 31.5 j 2.09 pF
2425 28 W - 30.57 j 2.15 pF
2430 27.6 W - 29.5 j 2.2 pF
2435 27.7 W - 28.4 j 2.3 pF
2440 27.6 W - 27.4 j 2.3 pF
2445 27.51 W - 26.4 j 2.45 pF
2450 27.5 W - 25.7 j 2.51 pF
2455 27.3 W - 24.9 j 2.6 pF
2460 27 W - 24.15 j 2.6 pF
2465 27.1 W - 23.4 j 2.8 pF
2470 27 W - 22.7 j 2.8 pF
2475 27.1 W - 22.2 j 2.89 pF
2480 26.8 W - 21.9 j 2.9 pF
2485 26.9 W - 20.9 j 3.04 pF
2490 26.7 W - 20.3 j 3.14 pF
2495 26.8 W - 19.64 j 3.24 pF
2500 27.2 W - 19.1 j 3.3 pF

 

[VSMVDD]

as far as i know if you add more wood to a new guitar
than another of the same guitar design

than needed,
it sounds a little different
no two 'instruments' sound the same

i expect this applies to capacitors resistors inductors
or identical twins

so.. do you really ?? expect two chip capcitors to be identical ???
or condicativly the same???

the truth is no two things are the same or they would try to occupy the same space in time

 

[vfone]

The answer to the question can be find looking to the price of a high quality test fixture for SMD capacitors.
These fixtures try to minimize as more as possible any influence of external parasitics to the measured component.
But finally any measurement error could be compensated somehow in the circuit, the same as Page or Iommi who can play probably any guitar in the world.

 

[FvM]

Without fúrther details about the test fixture and the calibration performed for this measurement, I tend to doubt the results completely. They don't look reliable at all.

 

[G0HZU]

The ESR of a good quality 1pF cap at 2GHz should be less than 0.5 Ohm. Some will be less than 0.2 Ohm. But you have measured over 10 Ohm and this is probably because it is very difficult to get a test jig that can measure a 1pF cap accurately at >2GHz.
So I would guess that both measurements are flawed as your cap measures to have a Q of about 2. (instead of maybe a more realistic Q of 200 for a good quality 1pF cap at 2.5GHz)

Also, you are measuring a high pF value. Due to package inductance the 1pF may appear more like 1.05pF to 1.1pF at 2.5GHz but I doubt the inductance will be as high as to make the cap appear much higher than this.

 

[robismyname]

Thanks for the response. I provided a snapshot of the test fixture to satisfy your cynical remarks. It's not pretty but practical. As far as calibration is concerned, not sure what you mean by this. Do you mean calibrate the test fixture? Please explain.

---------- Post added at 12:16 ---------- Previous post was at 12:15 ----------

Without fúrther details about the test fixture and the calibration performed for this measurement, I tend to doubt the results completely. They don't look reliable at all.

https://obrazki.elektroda.pl/58_1285762381.jpg
https://obrazki.elektroda.pl/16_1285762381.jpg
https://obrazki.elektroda.pl/41_1285762381.jpg
https://obrazki.elektroda.pl/91_1285762381.jpg
https://obrazki.elektroda.pl/60_1285762381.jpg

 

[biff44]

I have seen some really poor quality capacitors out there! It is possible you do have that much variation. Make a simple bandpass resonator with one of the caps that measures good, and measure its 3dB bandwidth. Then replace the capacitor with the lossier one, and remeasure the bandwidth without changing anything else. If the bandwidth has widened considerably, the real part measurement is confirmed.

 

[G0HZU]

Why don't you try measuring a known good capacitor. eg an ATC600S 1pF capacitor? This should show ESR around 0.2 Ohm at 2.5GHz on a good test fixture.
Also, I would be very surprised if any manufacturer made 0402 (NP0?) 1pF caps with a Q of 2 at 2.5GHz and also looked like 2.5pF at 2.5GHz.

 

[FvM]

I provided a snapshot of the test fixture to satisfy your cynical remarks.

If you possibly refer to my previous post, simply try to understand that's it's pure technical. Reviewing your photos, my comment reveals to be very cautious, I think.

To start with a simple point. Calibration of a one port measurement usually involves measuring a short, open circuit and 50 ohm reference in the exact place of the test object, in this case at the location of the capacitor, not at the SMA connector. Depending on the measurement frequency and the S11 value of the device under test, a deviation between the calibrated reference plane and the actual DUT plane, a measurement error is introduced. But placing a reactive component in a wrong reference plane can normally create a S11 phase error only, |S11| should be near to unity anyway. So the reference plane issue doesn't explain the high real impedance part in the measurement.

In your test setup, I can't see a direct connection between the SMA ground pins and the capacitor. The top side pins are apparently unconnected, the bottom side pins are soldered to a copper plane, but without direct connection to the DUT. As one explanation of unexpected real impedance value, you apparently created an antenna with considerable radiation resistance.

 

[vfone]

...first posted picture looks like a Halloween mask
Instead soldering the measured capacitor from the central pin of the SMA to that floating ground, I think better is to eliminate totally the PCB and just solder the cap between central pin and one of the ground pins of the SMA.

 

[FvM]

I think better is to eliminate totally the PCB and just solder the cap between central pin and one of the ground pins of the SMA.

Yes, as an important advantage, you can perform accurate short, open circuit and possibly 50 ohm calibrations by simply using identical SMA jacks.

 

[G0HZU]

Even with a revised homemade test fixture I don't think it will be possible to get an accurate measurement of a 1pF cap at 2.5GHz. Such a task would require a very good test fixture indeed. Why is this measurement necessary anyway? If the cap is suspect then replace it with one from a manuf with published ESR data at 2.5GHz. I doubt the 0402 package inductance will be an issue at 2.5GHz.

 

[FvM]

I doubt the 0402 package inductance will be an issue at 2.5GHz.

I completely agree. A standard 0402 1 pF capacitor has a SRF of about 7 GHz. So the series inductance will make about 1/8 of the total reactance at 2.5 GHz, assuming a simple frequency independant model. But you would be able to distinguish 1.0 from 1.5 pF. All in all, the problem seems more like a VNA measurement lab exercise than a reasonable way to check components.

 

[biff44]

Like I said, i recommend using a fixture "like" the one you want the actual application to be. If I were looking for capacitors for a filter, I would make a 1 pole filter and tune it up and compute the capacitor Q. If I were looking for a low loss series dc block, I would hook up 5 caps in series with 50 ohm line in between, and measure s11 and s21--once I had a baseline of good caps in the fixture I would then install the cheaper substitutes and see how they compared.

Attached is an old hp app note on measuring.