Joel Koltner wrote:
"Dave M" wrote in message
...
I'm genuinely interested in learning if the technique that I used
for years in calibration labs was, in fact, correct or totally
wrong. The technique that we used was this:
Using a high bandwidth scope, measure its risetime without the probe
being connected (scope connected directly to a fast-rise pulse
generator). Connect the probe being calibrated to the scope input,
and connect the probe tip directly to the pulse generator output.
Measure the resulting pulse risetime.
Using the formula that I gave previously (rearranged to find the
probe's risetime), calculate the probe's risetime and bandwidth.
This method of measuring the performance of a probe worked quite
well for the lab and our customers for the years that I was a cal
technician (commercial and military).

If all you have available is a pulse generator, this is a decent
enough method -- just know that the formula to convert between rise
time and bandwidth assumes a single-pole frequency rolloff, which is
usually -- but not always -- a reasonable assumption.

A more insightful means of measuring scope bandwidth is to take an RF
generator that can produce a repetitive frequency sweep (with a
leveled output), have the start of it trigger the scope, and then
adjust the timebase settings so that you get, e.g., 10MHz or 100MHz
or some other convenient MHz sweep per division. In other words, the
scope's display effectively becomes a Bode plot of the system (scope
+ probe combined) response.
---Joel

Thanks for the elaboration... makes sense. And yes, after wideband leveled
generstors became available, we did use them to measure frequency response
on scopes, but still relied on the risetime method to characterize probes.
The reason for that was that the leveled generators that we had didn't cover
the full spectrum of the scopes and probes that we encountered; we had a 30
picosecond risetime pulser that handled everything we saw.

It has been several years since I worked in a cal lab, so the technology has
progressed a good deal since. Are there any texts that describe the
technique(s) by which the response of any given scope/probe combination can
be calculated and verified? Not that I need to do it any more, just
interested in knowing.

--
David
dgminala at mediacombb dot net