# S21 VNA Connector & Cable Loss?

I've tuned plenty of RF devices using my VNA for impedance matching. But I've never done an S21 measurement.

I have a shielded tent with some connectors and SMA cables that are going to introduce some kind of loss.

If I run an S21, will the results be across a frequency span like an S11?

Is there a way to calculate the average signal loss?

• Bypass D.U.T. and measure s21. This is your test system loss. If you calibrated with this to null out this loss, then the result is a flat line 0dB then reconnect to the device under test to measure D.U.T. insertion loss/gain = s21 . The other parameter s11 is input Return Loss due to mismatch Z where a perfect match is -max and short or open is 0dB. With a feedthru SMA barrel to connect cables you might get 0.1~0.5dB insertion loss , which can also be calibrated. – Sunnyskyguy EE75 Mar 2 at 22:30
• What does "like an S1" mean? What is the general nature of the device you're testing (i.e., an amplifier, a passive transmission line, a free-space link between two antennas)? – The Photon Mar 2 at 22:37
• @SunnyskyguyEE75 -- In an S21, is that swept across a frequency band with a VNA? My antenna I have correction factors for (I think every 5MHz). I'm thinking to measure the same my final cables and adapters with the VNA at 5MHz increments. That sounds like the right approach? – Leroy105 Mar 2 at 22:47
• @ThePhoton - I'm just trying to measure cable loss from a few SMA barrel adapters, my EMI filter on my shielded, tent and the few cables I have connecting everything. I'm going to remove the antenna, hook up the VNA inside the tent, and measure the loss. I'm just wondering if the S21 is swept across the frequency bands so I"ll get a table of correction factors or if you can just generalize the number. – Leroy105 Mar 2 at 22:49
• Yes s21 is swept across and barrel loss can be calibrated , follow the user mode calibration in manual for DUT testing to null out cable and test jig losses The VNA stores in memory the result for every frequency – Sunnyskyguy EE75 Mar 2 at 23:40

Yes, for most devices the insertion loss ($$\S_{21}\$$) varies with frequency, so you should measure it across the whole frequency band where you would be using the device.
You'd usually use the same menus on your VNA to set the frequency range and number of steps for an $$\S_{21}\$$ measurement as for an $$\S_{11}\$$ measurement. In fact, for the best accuracy, the VNA will probably measure all 4 S-parameters at each frequency, because the $$\S_{11}\$$, $$\S_{22}\$$, and $$\S_{12}\$$ measurements will contribute to correcting the $$\S_{21}\$$ measurement for test set and cabling effects.