Updated July 6,
here to go to our main page on group delay
here to go to our page on simulating group delay using Agilent
There are two primary means of
measuring group delay. The most widely used involves frequency domain
method converting small-signal S-parameters to group delay. The
time-domain method directly measures the time that a pulse takes
to move through a network.
Group delay frequency domain
Pay attention, in the following
discussion you might learn how to use one of our free
downloads to calculate group delay from manufacturers' S-parameters...
At microwave frequencies, we
usually measure stuff in the frequency domain. If you have vector
S-parameter data over frequency on your DUT (device under test),
you already have the means to evaluate group delay as a function
of frequency. Quite often your network analyzer will support direct
read-out of group delay measurements. But below we will describe
how to make the conversion from S-parameters to group delay, you
will find this useful when comparing devices from manufacturers
data sheets which almost always include S-parameters but seldom
include group delay data.
Check out the following simple
equation. Group delay is the negative-slope of the transmission
phase angle with respect to frequency.
Remember that units must be consistent,
if radians/second are used for frequency, you must use radians for
phase angles, then the calculated result will be in seconds. We
recommend that you use degrees and degrees/second in the formulas,
because most data is handled in degrees, so that's one less thing
you need to convert. Why bother converting degrees to radians?
To get frequency in degrees per
second, multiply Hertz(cycles/second) by 360 degrees/cycle. One
other trick to remember: if you start with frequency in GHz,
you will end up with group delay in nanoseconds, without
having to mess around with moving the decimal place around.
You don't need to do the calculation
yourself ever again, because Microwaves101 offers a free download
of an Excel file that has equation built in, and plots the data
for you! It's called S-Parameter
Utilities 101, check it out in our
The plots below are calculated
group delay for a TriQuint amplifier based on S-parameter data that
we took from their web site while they were sleeping. The plots
were all generated using our free download.
One interesting plot is the S21
phase angle, after it is "unwound." The slope of this
plot is the group delay. You can generally see were the group delay
will not be flat, by any regions of non-linearity on the plot. From
the plot you can expect funny things to happen to the group delay
of this circuit starting just below 8 GHz...
We included an S-parameter plot
in the download. Might as well plot it, the data is there!
Finally we come to the group
delay plot. It does suck at eight GHz by golly!
Group delay time domain measurements
Coming soon! Maybe!
ADS to simulate group delay
Tis topic has moved to a separate