Check out our page on how
to build a pulsed RF source!
The first measurement we will
add to this page is recovery time.
Let us say first, that there
is no better oscilloscope than a Tektronix oscilloscope. It's not
just us, there are people out there that collect old Tek scopes
as a hobby!
We don't have the resources to do justice to the history of oscilloscopes,
but here is a web site has made an attempt... check
Oscilloscope measurements are
to the time domain what spectrum analysis measurements are to the
frequency domain. Most microwave oscilloscope measurements require
you to use some type of envelop detector circuit, since microwave
frequencies are high compared to the bandwidth of typical oscilloscopes.
However, modern digitizing scopes can record and display waveforms
as high as millimeterwave frequencies. This is a very recent development,
if you asked the Unknown Editor twenty years ago if he would ever
be able to go into the lab and display an X-band sine wave on an
o-scope, he'd have asked you what you were smoking!
We will assume that you know
a little bit about oscilloscopes, or you wouldn't have made it through
your last job interview. Here are some terms you will need to understand
for o-scope microwave measurements...
Bandwidth determines an oscilloscope’s ability to measure an AC
signal. Bandwidth is specified where the frequency response
As signal frequency increases,
the capability of the oscilloscope to accurately display the signal
Samples per second
Yes this page needs work, help
a brother out and send us some material!
Rise/fall time measurement
In pulsed systems it is often
important to determine how sharp the transition from "RF on"
to "RF off" is. This is measured using a detector and
an oscilloscope, and a pulsed RF source. We have a new chapter on
building a pulsed RF source, check it out!
An RF detector is merely some
type of diode that rectifies RF incident on one port and outputs
a DC signal on the opposite port, Typical detectors use Schottky
diodes or tunnel diodes. Check out our separate page on detectors!
This measurement is typically
done on a receiver in a radar system, or bits of a receiver such
as the front-end limiter and/or low noise amplifier. By the way,
we've searched the world-wide web for a description of recovery-time
measurements, and trust us, this is the only description out there
that explains how to do this critical measurement. Try it and impress
your friends and boss!
A radar system typically duplexes
a transmit and receive signal through one antenna, as shown in the
figure below. When the transmitter is on, some amount of power will
reflect off of the antenna and leak into the receiver, putting it
into large signal operation (perhaps even saturation). This does
some funny things to the receiver, most notably, it can kill the
small-signal gain for competing signals. When the transmit signal
is turned off it takes some time for the receiver to recover to
it's cheerful, high-gain self. Hence, the term "recovery time".
Transmit and receive
paths duplexed through a single antenna
In order to measure recovery
time, you will need two signal sources, one providing a pulsed high-power
signal (with rise/fall time better than the recovery time you want
to measure), and one providing a CW lower-power signal. They must
be separated substantially in frequency (a few GHz should do it),
but both be within the bandwidth of the device under test. You will
need a bandpass filter that passes the small CW signal, and rejects
the large signal by 50 dB or more (hence why they need to be separated
in frequency). We recommend a waveguide filter for this. The filter
is the "trick" that allows you to examine the detected
waveform of the small signal at the output of the DUT, without blowing
up your lab equipment: typically 23 dBm will blow an expensive detector
Suppose you want to measure the
recovery time of an X-band limiter (bandwidth 8-12 GHz perhaps),
after it is hit with a ten watt pulse. The recovery time specification
you need to beat is 50 nanoseconds. The two signals you want to
apply to your DUT are a ten-watt pulsed signal with less than 10
ns rise time, and a 10 dBm CW signal (close to the threshold power
where the limiter starts to limit).
In the figure below, sweeper
1 supplies a 11 GHz CW signal to the medium-power amplifier. Here
we have chosen to drain-pulse this amplifier and use it as the modulator
for the high-power signal, to create a nice sharp square-wave RF
pulse. Why did we do this? Because the sweepers you will find in
your lab usually are not capable of fast rise times. The drain-modulated
amplifier supplies a one-mW square-wave pulsed signal with less
than 10 Ns rise/fall time to the TWT, which creates the 10 Watt
pulsed signal to DUT. Note that charge storage is important for
the drain-modulated amplifier (the 10 uF and 5000 pF caps), as well
as having short leads (low inductance) to the power supply.
(Need to fix the figure so it
shows a quadrature combiner instead of a Wilkinson for better power
Recovery time measurement
Sweeper 2 and the second medium
power amplifier supply the weak CW signal to the DUT (~10 dBm CW).
Don't be tempted to combine the two signals earlier and amplify
them through the same high-power amplifier (the TWT) because you
may well end up measuring the recovery time of the amplifier instead
of the DUT.
The bandpass filter after the
DUT essentially removes the strong pulsed signal from the output
of the DUT. It is rejected by ~50 dB (if your filter is any good!),
and is reflected back toward the source, and dissipated in the two
isolators J3 and J4. This allows the best possible view of the recovery
time on the scope because no attenuation is needed in front of the
detector to protect it from the strong signal. Just be sure some
fool doesn't mess with the frequency of the strong signal and put
it inside the filter passband, which will cost you a detector for
Barrier diode detectors from
Krytar have 1-2 nanoseconds rise time when terminated with 50 ohms.
The 20 dB coupler and power meter in the set-up are used to calibrate
the detector output voltage. Remember that since you are using a
square-wave, the peak power is three dB higher than the average
Coherent versus non-coherent
RF pulses--what does this mean?