Version 0.4 Results

Here are some test plots from the v0.4 VNA of 2018. The first is a plot of S ₁₁ for a Z₀ load from 1 to 30MHz. As expected from the calibration, it is exactly correct. The open and short standards produced the same kind of expected perfect results. The next is of a 20 Ohm resistor over the same frequency range This was just a quarter watt resistor held in the end of the BNC connector and of course shows a little inductance. The next one is a 100 Ohm resistor with a bit shorter leads. All of the S₁₁ plots were after SOL calibration, using the also homemade reflection bridge.

The next plots are testing the isolation. There is no connection between the TX and RX ports in these plots, and they don't use any kind of calibration, so as to show the isolation. Magnitude and phase plots of S₁₂ showing the rather poor of between 50 and 60 dB up to 100 MHz. Even below, it would be nice to see more isolation. The phase response is rather linear, which might indicate that there is some constant delay getting from the TX port to the RX port, which might mean there is one principle place the signal is getting through.

The insertion loss was measured by putting a short coaxial cable between the input and output. Here is the result between 1 and 100 MHz. The corresponding phase response shows the electrical length of the cable.

A more rigorous test is connecting a length of coaxial cable to measure S₁₁ for an open and short circuit. This test shows problems. See for example, the open circuit test below.

Note that the reflection coefficient appears larger than one at frequencies of 70 MHz and 83 MHz (count the dots; 1 MHz per dot on S₁₁ and 500 KHz per dot for S₁₂) and this corresponds with the most rapid positive changes in the magnitude of the S₁₂ plot. These problems could also be with the standards used to calibrate the analyzer. Another possibility is that we are not measuring the source voltage which doesn't give us a perfect reflection bridge. This analyzer should measure as a reference the open circuit voltage (not the loaded source voltage). See my analysis of the reflection bridge below. The WxMaxima analysis of this circuit is: This shows that we really want to measure the open circuit source voltage. We have a spare output on the Si5351a, so we could use that to feed the TX output, and send this one directly to the sampler. That way we will me measuring the open circuit source voltage as the reference. I hope that makes the SOL calibration less sensitive to the errors in our home made standards.