VNA Test port extensions

[dmills]

Just a quick one, but I have a bit of a 'funny'.

I have an 8753C with the usual 3GHz test set and a set of known reasonable SOL terminations all on APC-7 (I have S parameters for the load and short/open done with reference to a calibrated setup and they are reasonable).
I also have a couple of HP Test port extensions, also APC-7.

Thing is, when I do a cal with the reference plane at the instrument connector then measure the same load at the other end of a test port extension I am only seeing S11 around -25dB or so even at very low frequency, to what extent is this expected? A foot of coax should not be adding ~3 ohms round trip surely?

I was expecting the 1 Port measurements at the far end of the cable to be basically the same as the ones at the cal plane with just a few ns electrical delay added in, particularly at very low frequency (Think 1MHz), but that is not what I am seeing.

Any ideas appreciated.

Regards, Dan.

[TheUnnamedNewbie]

What happens if you just remove the load from the port on the VNA, wait a few minutes, and then connect it back? (perhaps some kind of repeatability issue?)

[dmills]

Could be but moving the cal load back to the ref plane made it come good again so it feels a little unlikely.

I will give that a try next time I have the thing powered on, it is the sort of thing you want to do after an hours warm up in any case.

Regards, Dan.

[KJDS]

Try looking at the response on a Smith chart, it may give more information on what's happening.

[hendorog]

Since you have two VNA ports, two port extension cables and multiple terminations, does mix and matching in different combinations make any difference?

There is a bit of information on this page about how to test a cable, there might be a fuller version somewhere if you search, this seems to be a quick summary:

https://community.keysight.com/message/64620?commentID=64620#comment-64620

"They are easy to test.  To test match uncertainty, connect a short, select data->mem and data-mem (data MINUS memory).  Set the format to logmag.  The result should be around -50 to -60 dB.  bend the cable back and forth and observe the highest peaks.  If the highest peaks are less than -50 dB, it is a metrology grade cable, if the highest peaks are less than -35 dB, it is a production quality cable.  If the highest peaks are worse than -20 dB, it is not worth using. "

[WPXS472]

You have a test port extension that is a cable? All the ones I have ever seen are only about 2 inches long and machined, not cable. If you do a 1 port at the instrument connector, then connect a load by a length of cable, you will see the usual peaks and valleys caused by the imperfections of the cable. The better the cable, the smaller they will be. If you are seeing 25 db of return loss at low frequencies from a load that you used to do the one port with, it sounds like the test port extension is bad. 25 dB at 3 GHz, maybe. But at low frequencies, I am like you, you shouldn't be seeing that. Interesting problem I will be waiting to see what you find out.

[Bud]

Test the cable for S21 on the VNA and see what attenuation it has in the frequency range of interest.