Verifying the performance after calibration

Verifying a VNA has been calibrated properly is not a trivial task. Just measuring the calibration standards gives you an idea of repeatability, but does not prove the calibration has been performed correctly. Commerical VNA verification kits are available for this purposed, but these are very expensive, as they contain a precision airlines like in the photograph below, which shows the two airlines from an HP 85053B verificaton kit. At the time of writing, (March 2018), the Keysight 85053B verification kit costs $8249 (USD).

50 ohm and 25/50 ohm stepped impedance airline (Beatty Standard)

In addition to the two airlines, the HP/Agilent/Keysight 85053B verification kit contain two attenuators, with data measured by the manufacturer. The data is in a format designed to be read by a VNA, not a human.

It is impossible to produce airlines and put them in a low cost kit, but Kirkby Microwave do provide some ways to verify performance in their calibration and verification kits. It is strongly suggested that you perform these verification steps after the VNA has been set up to use the kit, in order to verify that everthing is set correctly.

Measure the calibration standards

Having the correct coefficients loaded into the VNA is critical, yet sometimes people enter them incorrectly. The following procedure will detect some errors. If possible, to ensure maximum stability when doing these tests, do not use any cables, but use adapters to connect to the VNA.

  1. Using only adapters, and no cables, put some adapters on the VNA so that there is a female SMA or female N socket.
  2. Let the VNA warm up properly. Typically this is one hour for most laboratory VNAs, but check the manual.
  3. Perform a 1-port calibration using the male open, short and load. Do the load last, and leave the load connected.
  4. Measure S11 of the male load. The phase data will be meaningless, but the amplitude should show a very high return loss. Exactly how high depends on your VNA, but 50 dB or better (|S11| <= -50 dB) should be observed on any laboratory VNA. This does not mean the return loss of the load is 50 dB! Irrespective of how good or bad the load is, if you measure the device that has just been used to calibrate the VNA, it should show almost perfect results - even if you use a wire-wound resistor!
  5. Replace the load with the short. Measure S11 of the short. This should show an amplitude of very close to 0 dB, and a phase that looks almost a stright line with freqeuency. Compare the phase measured to that in the file male-short-xxxx.s1p on on the USB stick, where xxxx is the serial number of the kit. It is hard to say how close agreement should be, as this depends on the VNA, but one might reasonably expect to see differences of less than 1.5 degrees from the measurements supplied.
  6. Replace the short with the open, and again check the amplitude of S11 is close to 0 dB, and the phase is close to that in the file male-open-xxxx.s1p.

After using one of our kits to calibrate your VNA, it is sensible to check the VNA is giving good results. A failure to give good results indicates either a fault in the VNA or an error in uploading the coefficients properly. Unlike calibration kits from other manufacturers, all our coaxial kits include a fixed attenuator with measured data. The data will be found on a USB stick included with your kit in the directory "Measured_data_of_attenuator". If you get similar results to what's shown on the USB stick, there is unlikely to be any serious problems. You should get good agreement on S21 and S12, but do not expect good agreement on S11 or S22, since the return loss of the attenuator is high, so the measurement uncertainty is high.

Note the aim of the verification attenuator is not to prove your VNA is within specification - if you need to know that, send it back to the manufacturer, and not a third party lab. But the attenuator will give you a good idea if there are any serious problems.

Procedure to use the attenuator to verify there are no serious problems on transmission (S21 and S12) measurements.

This procedure is applicable to our N and SMA kit. Some parts can only be performed on the N kit, and some only with the SMA kit but most can be done with either kit.

The data measured by Kirkby Microwave will be placed in a .s2p Touchstone file. This will have the name attenuator-xxxx.s2p, where xxxx is the serial number of the calibration and verification kit. It will have the measured data on all 4 S-parameters. PDF plots of the data are also provided for the convenience of the user, although it is easy to plot them yourself, once you understand the .s2p Touchstone format. (An understanding of Touchstone files is essential if you intend making any serious use of a VNA.) You can plot the data using OpenOffice, Microsoft's Excel, or any other software capable of producing graphs. Kirkby Microwave use Gnuplot since it allows one to create files without user interaction.

4 graphs (one for each S-parameter). Each graph has two y-axes, since two sets of data (magnitude and phase) are shown.

As stated earlier, measuremments of S11 and S22 are subject to very high uncertainty, so whilst these are provided, they are of limited use.

One should perform the following steps to ensure the the transmission data you measure matches that measured by Kirkby Microwave:

  1. Perform a 1-port calibration of the VNA using the female standards at port 1.
  2. Measure S11 of the female short. You should expect to see the magnitude to be close to 0 dB, and the phase agrees within a couple of degrees of the file female-short-xxxx.s1p. If not, the most likely cause of this is that the coefficients for the kit have not been entered correctly.
  3. Measures S11 of the load. Depending on your VNA, you should expect to see a return loss of better than 50 dB if performed immedately after calibration. This does not indicate the load has such a good return loss, but if the VNA does not indicate a good return loss, then something is wrong.
  4. Perform a full 2-port calibration of the VNA, using the calibration and verification kit.
  5. Measure S21 and S12 of the attenuator, and save the data into a Touchstone .s2p file.
  6. Import the Touchstone file into Excel, or other spreadsheet. Use a space as the separator - not a tab or comma which the spreadsheet may expect.
  7. Remove the comments (lines starting with !) and the option line (starting with a #).
  8. Plot:
  9. Compare the plots of S12 and S21, to ensure that the S12 and S21 data are similar to those provided by Kirkby Microwave.

verification plot

If you measure S11 or S22 of the attenuator, and compare the data to that supplied, you will find poor agreement. This is because the attenuator is good quality and reflects very little power, especially at low frequencies. Therefore the measurement uncertainty of S11 and S22 will be high. Also, movement of cables is likely to cause changes in magnitude and phase far in excess of those of the attenuator itself. Because of the problems in using S11 and S22 data of the attenuator, a revised set of data was provided by Kirkby Microwave on some kits shipped after August 20th 2016.

Verification of reflection measurements on SMA kits shipped after 20th August 2016.

As stated above, the high return loss of the attenuator means measurements of S11 and S22 of the attenuator are subject to very high measurement uncertainty. In August 2016 we developed a new reflection standard, allowing measurements of reflection without excessive uncertainty. On SMA kits we measure the reflection coefficient at port 2 (male port) while the female port is left unterminated . We call this measurement Gamma_2, as it is not an S-parameter measuarement, which would require the female port to be properly terminated. The Greek letter Gamma is often used to denote reflection coefficient, and the measurement is made at port 2 (the male port).

An excellent method of verifying the VNAs ability to measure reflection data

This makes a very useful test in reflection mode. (It is futile trying to measure at the female port on SMA connectors while the male port is left open, as the movement of the nut will make it impossible to obtain stable measurements at microwave frequencies.)

The USB stick will contain 2 files for these measurements.

  1. attenuator-xxxx-reflection-coefficient-at-male-port-with-female-port-unterminated.s1p - magnitude and phase of the reflection coefficient in a .s1p Touchstones file
  2. attenuator-0416-reflection-coefficient-at-male-port-with-female-port-unterminated.pdf magnitude and phase of the reflection coefficient in a PDF file.

In each case, xxxx will be replaced by the serial number of the kit. The graphs will look similar to this.

reflection measurement, with low uncertainty

These measurements at one port, with the other port left open, are made without any cables in the system, so the attenuator connects to the VNA only via adapters, which are electrically more stable than any cables. For measurements of transmission (S21 and S12), it is impossible to avoid using cables, although for best accuracy, only one cable could be used.

One should perform the following steps to ensure the data you measure matches that measured by Kirkby Microwave:

  1. Perform a 1-port calibration of the VNA using a female SMA or 3.5 mm connector that's connected directly to the VNA, or via adapters. Do not use cables.
  2. Measure S22 at the male port of the attenuator with the female port (#1) unterminated, and save the data into a Touchstone .s1p file.
  3. Import the Touchstone file into Excel, or other spreadsheet. Use a space as the sepparator - not a tab or comma which the spreadsheet may expect.
  4. Remove the comments (lines starting with !) and the option line (starting with a #).
  5. Plot:
  6. Compare the plots of your S22 measurements, with the female port open, are similar to those provided by Kirkby Microwave on the USB stick.

Verification of reflection measurements on N kits shipped after 20th August 2016.

For N kits shipped after 20th August 2016 we provide measuremenet at both the male and female ports while the other port is open. This is possible on N, but not on SMA, due to the fact movement of the nut on an N connector causes very little change in reflection at the female port, whereas it makes a massive difference on the SMA connector.

Valid XHTML 1.0 Strict Valid CSS!