# calibrating vector network analyzer with $75\Omega$ coax

I am new to electronics and I just began to use a vector notwork analyzer for some measurements.

I calibrated my vector network analyzer with both ports connected to 50$\Omega$ coax cables with the other ends connected to each other via a coaxial cable coupler.

Every component I've used is 50$\Omega$ impedance matched but I started wondering if I could use 75$\Omega$ rated components instead and be fine conducting the same experiments if I just calibrate S parameters correctly.

There are several ways to use a 50ohm network analyser to make measurements in a 75ohm system. Not surprisingly, they are all inferior in some way or other to using a 75ohm analyser, but they'll work within their limitations.

1) Measure everything in the 50 ohm system, and convert mathematically.

Pros - no extra hardware
Cons - you might want best accuracy around the matched condition, and this doesn't give that. Small errors in calibration will degrade your near-matched results proportionately more than large reflection results. The need for results conversion means an extra step offline, so you don't get meaningful final measurements in real time on the analyser.

2) Convert your 50ohm ports to 75ohm ports using 50/75ohm RF transformers.

Pros - power is not lost so noise level stays OK.
Cons - transformers will degrade match so reducing measurement accuracy, and only work over restricted frequencies anyway

3) Convert your analyser's ports to 75 ohms using min-loss 50/75ohm pads

Pros - much wider bandwidth and more accurate than transformers
Cons - they lose significant power, so your dynamic range drops, and S11 measurements are made through twice the pad loss, so you're losing sensitivity

4) Calibrate it with 75ohm calibration pieces

Pros - work at full power, mathematically turns it into a 75ohm analyser
Cons - you need a computer corrected network analyser (CCNA) and a set of 75 ohm cal pieces (expensive, depending on accuracy). As you're working a long way from the intended impedance, second order effects will mean it doesn't calibrate as accurately as would a dedicated 75 ohm analyser.

For measurements where there are dedicated generator/analyzer ports, you could get away with impedance converters and adding the power loss in the conversion.

That's how I ran tests on a 75Ω spectrum analyzer when I only had a 50Ω generator available: configure the generator to +4.5dB, and use a converter with 4.5dB attenuation.

Network analyzers are usually more complex than that, as they also measure reflection, so ports may be used bidirectionally. The impedance converter might hide the impedance change at this point by dumping some energy into a resistor, but it will do the same for a return signal, so the measurement is no longer useful.

• Thank you for the answer. In case you have dedicated ports, like the case you mentioned, what was the math/logic behind configuring the generator to +4.5dB and using a converter with 4.5dB attenuation? Is it from the fact that $\Gamma = \frac{Z_L - Z_0}{Z_L + Z_0}$ somehow? – Blackwidow Jul 31 '18 at 22:27
• This converter was what I had readily available at that point. Impedance matching with a passive network will always lose a bit of power, which I compensated for by turning up the generator. – Simon Richter Jul 31 '18 at 23:36