Yes, at 300 MHz, 0.8 pf has a capacitive reactance of 660 ohms. That value will induce a return loss of about -11 dB (VSWR ~1.1:1). Part of the requirement is that the frequency be chosen such that the reflections don't add up badly. One idea that I plan to try is to choose the lengths such that each cable is an odd number of quarter wavelengths. This should allow a cancellation of the reflections.
- The tap for getting the signal out may add capacitance of about the
same order of magnitude.
The signal extraction tap in the test fixture is 10k, so its reflection is negligible.
The pads on the PCB for the 10K resistor will have some capacitance. Unless you are careful, they can be an appreciable fraction of a pF.
- You can probably reduce the reflections by putting attenuators in the
delay lines, but these would make the DC biasing of the varactors more difficult.
My model has assumed ~1 dB of loss intrinsic in each coax line, and I see you used lossless lines. It may also be necessary to put a resistor in series with the varactor to help with the reflections.
I can try simulating with the series resistor, but my gut says it will mess up the linearity of the phasing vs. tap. 1dB of loss per section of coax might help. I'll try simulating it. I don't know if we can count on that much loss from such a short piece of coax at these frequencies, though. High loss coax might not help, since loss would mostly come from leakage, not absorption, which would be a problem.
I have attached my simulation. It was done in qucs, which runs on linux. See qucs.sf.net or get it from your distribution.
Following your tip last weekend, I picked up QUCS from SourceForge. Nifty program. I have had trouble doing a sim at a constant frequency with a variable capacitor. Your example gave me some hints, but I still have problems doing a sim under such conditions. I'm learning!
I don't know how to make it NOT sweep frequency. However, you can make the start and stop frequency the same, and use only 2 points. Then you can make the capacitance value into the independent axis of the graphs.
If there is anything else you'd like me to simulate, let me know.
Matt