Measuring Group Delay -- HP Vector Network Analyzer

[[email protected]]

Doing some research to understand a bit more about a measurement I am
required to make, Group Delay. From internet reading, the definition
of group delay: - derivative of the change in phase over the change in
angular velocity. Tg = -d(theta)/d(omega)

It seems relevant to make as it will give some indication of the signal
distortion for a particular modulated signal. In my case OQPSK.

Now for praticality. Need a bit of help interpretting the results of
the measurement made using an HP8722D per the manual.

my case has a BW of 500MHz. The plot is a waveform with ripple over the
range. The spec I am comparing against has three components: Linear
(time/Bandwith), parablic (time/BandWidth^2), Ripple (time).

How do I get these results from a trace? The parabolic is the hardest
for me to conceptualize. Do I need to make some kind of interpolation?

Thanks.

 

[PeteS]

This is not easy without pictures, but let's look at the trace. The
trace you have is very probably NOT straight. The curvature of that
trace is the parabolic you are seeking. (Note that it's not always a
parabolic curve, but the below is close enough for most measurements -
not always, but usually)

To get the actual value, take a straight line from the low end of the
trace to the high end. Now measure the tangential 'distance' to the
edge of the curvature part. That's the parabolic.

Cheers

PeteS

 

[[email protected]]

PeteS,
Thanks for the response. I am having a bit of trouble visualizing what
you described. I have post a jpg that represents the traces of the
group delay measurement on an 8722. In my case its a 500Mhz BW and the
the time scale is in nS.

http:\\www.naturallywired.net\groupdelay.jpg


If possible, annotate the graph or use it as a reference to discuss.
Thanks for the help