AnsweredAssumed Answered

Measuring Accurate Group Delay

Question asked by Dr_joel Employee on Jan 30, 2006
Latest reply on Mar 7, 2006 by LeonardHayden
A customer askes: how do I measure +-10 ps delay on an S-band device (in RF Globalnet).  My answer:

This is a really tough requirement.  One thing you left out of your question is what is the maximum delay aperture? (Delay aperture is the allowed delta frequency used to measure the delta phase).  Typically it's some small percentage of the channel bandwidth.  This is important as the delay accuarcy is inversely proportional to the aperture. 

The matching delay line in the R channel will help in two ways:

1) if you are sweeping (as contrasted with stepping), it will normalize the signal in the R channel and the test channel.  But for good result, you should not sweep, but use the step sweep mode.

2) it will normalize the phase noise in stepping, a little bit.  This is theoretical, and in my experience, has only marginal improvement.

The most inportant thing is trace noise.  The 8722 trace noise in the middle of S Band is about 4 or 5 times worse than the PNA.  On my 8722 I see abouit 85 millidegree of trace noise.  This corresponds to 200 psecs of delay noise, in a 1 MHz aperture (using 1 kHz IF BW).  If I try to lower the IF BW on the 8722, I run out of phase resolution (minimum resolution looks like about 4 millidegrees, below this the ADC doesn't gives phase that looks like discreet .004 mdeg steps).  On the PNA at 30 Hz IF BW I get about 15 ps of peak-to-peak trace noise (50 GHz PNA at 3.3 GHz).  I don't have an RF PNA right here, but I would guess it should be better.  Again, these are at 1 MHz aperture (smoothing set to 1 MHz).

You might also try the new Agilent VNA discussion forum.  I'll cross post your question and my answer, and you might find others with a good comment.

The Agilent forum is at www.agilent.com/find/agilent_forum

(note the underscore between agilent and forum).  

Outcomes