[time-nuts] Reverse isolation

Magnus Danielson magnus at rubidium.dyndns.org
Sun Mar 8 23:23:18 UTC 2009

Didier Juges skrev:
> As long as one is not trying to measure extremely low signals (or extremely
> high isolation), the conventional method works, and I have used it many
> times, I am not sure that the method would scale when dealing with very high
> levels of isolation.
> I certainly would not trust a conventional VNA for isolation of 100dB or
> more, simply because such isolation is difficult to achieve in an instrument
> that is supposed to be able to switch its source and receivers between the
> two ports.
> The HP 8722D which we use at work is only specified with 100dB dynamic
> range, and I am not sure how that applies to reverse isolation.

Notice that the VNA does not have to handle all the dynamics by itself. 
You could use an external output amp which can be made sufficiently 
isolated from the input side as well as a LNA to gain the receiver side. 
By measuring the amps gain in a calibration setup you can get correct 
measures in the total setup.

If you have a 60 dB of total amplifier gain the VNA would see a 
remaining 60 dB dynamics on its output-to-input. That is more along the 
lines of what we trust. What remains an issue is how we ensure more than 
120 dB isolation between input and output of the DUT and also the output 
of measurement amp and input.

I agree with you that I would not entierly trust the VNA for 120 dB... 
directly. Additional amp stages near the DUT can give the VNA dynamic 
range the necessary boost.

> Using a slightly offset test frequency while driving the device at its
> normal operating frequency would only work when using a spectrum analyzer as
> receiver. Most VNAs don't have the capability of rejecting large signals
> close in, and separating the frequencies too much would make the test
> invalid with a narrow band amplifier. I have used that method to measure the
> hot output VSWR of a TWT amplifier for instance (that was interesting). Of
> course, the spectrum analyzer does not give you the phase, but it's better
> than nothing.

Actually, it depends on the distance from the hot carrier, the amplitude 
of the hot carrier and the particular VNA in question. My VNA allows me 
to change filters as well as integration time.

So, I agree that not all VNAs is well suited for it and that is indeed a 
good warning.


More information about the time-nuts mailing list