# Oscilloscope question

Peter N. Glaskowsky png at ideaphile.com
Fri Apr 8 12:55:19 CDT 2005

```Just a quick comment on this one--

You want an scope with several times as much "bandwidth" as the frequency of
the fastest digital signal you will examine-- 4X minimum. The analog
frequency content of a digital signal is much faster than its clock rate.
Without a substantial margin, the signal will be badly distorted.

Analog scopes are cheap and widely available, but they have limited value
for troubleshooting digital electronics. They can only show you repetitive
signals, or the envelope of a randomly changing signal. They can't show you
something that happens only once, or very rarely, but that's the nature of
most digital problems. Expensive analog "storage" scopes can hold a signal
on the screen, but only if you can figure out how to capture it in the first
place. You might as well spend the money on a digital scope.

If you're getting a digital scope, you need a "sampling rate" that is also
much faster than the frequency of your fastest signal. The Nyquist theory
says you need 2X the sampling rate to reconstruct a signal, but that theory
applies ONLY to reconstructing a known waveform given a large number of
samples. Again, in practice, you want a sampling rate at least 4X your
digital clock rate, and preferably more like 10X. This ratio determines how
accurately you can measure the position of your signal edges.

You will find old digital scopes that quote two sampling rates. The higher
figure will be a "repetitive sampling rate" specification that applies ONLY
to reconstructing periodic signals such as radio carrier waves or computer
clock signals. For non-periodic sources such as logic signals, only the
lower "raw" or "real-time" or "one-shot" or "instantaneous" sampling rate
matters.

Today, even for under \$1,000, you can get a digital scope with a true
sampling rate much faster than you'll probably ever need. These are usually
priced according to analog bandwidth. For troubleshooting old computers,
even 100MHz is probably more than sufficient.

A good probe reduces the load on the signal being observed, as David Holland
said. An o-scope input has a 50-ohm impedance; that kind of load will stop
many circuits from working. So-called "active" probes present an almost
undetectable load, but you will rarely need anything that sophisticated.

If you're going to use a scope for ANYTHING, get this book:

Troubleshooting Analog Circuits, Bob Pease, ISBN 0750694998

http://www.amazon.com/exec/obidos/tg/detail/-/0750694998/
Or buy an autographed copy from the author by emailing rap at galaxy.nsc.com.

Though written about analog circuits, all of these electrical principles and
troubleshooting techniques apply to digital circuits. Let's face it, the
circuits themselves are analog anyway.

.                  png

```