Every now and then we need to check a signal waveform. Simple ones can be detected with a DMM or a LED probe.
But more complex waveforms can only be detected and shown by means of an oscilloscope. From 1982 to 2006 I
used my (t)rusty old Hameg HM 203-4 20 MHz analog scope. In most cases it was enough to get the job done.
A few weeks ago I saw an affordable used DSO (Digital Storage Oscilloscope) at Ebay. And I won it, thanks to a
sniperbid. The scope is here and I'm checking the operation and possible improvements or repairs.
Roughly, there are two kinds of oscilloscopes:
The Analog scope comes closest to a television receiver. A signal catcher and conditioner collects the
presented signal and sends the information to the vertical deflection of an electron beam inside a CRT tube.
- analog scopes
- digital scopes
The horizontal movement is fixed and goes from left to right (from my point of view). The combination of the
vertical deflection due to 'the input signal' and the steady left/right movement of the beam, produces a
waveform on the television screen. You can do a lot with an analog scope. An analog scope is honest. What it
shows, is present. Period.
The digital scope is more close to a computer. The input signal is conditioned and then fed into an Analog to
Digital Converter (ADC). So a voltage is translated into a number between (in many cases) 0 and 255. All
operations are now carried out on these numbers. The original waveform gets lost. It has disappeared.
After the processor has done its trick with the numbers, these are retranslated into a voltage (in a DAC:
Digital to Analog Converter) and fed into the circuit for the electron beam vertical deflection. Or, with
modern equipment: the data are plotted on a small VGA screen.
The danger of digital scopes is, that YOU, the user, must know what you are doing. The digital scope has no
conscience. It will lie to you, if you ask it to. And you won't see the difference...
The DSO is ideal for proving a hypothesis. If you're just taking measurements to see what's going on with the
signals: be very careful, since the DSO will not show you the original waveform anymore. Ever. So after a
series of measurements you need to go back to the desk and come up with a theory and a suitable model. Based
on that, a series of follow-up tests must be conducted.
The DSO takes samples from a signal. If something happens between two samples, the DSO will never find out. It
just didn't look. So it wasn't there.
I will explain the above with a few examples:
Examples 2 and 3 are called 'aliasing'. If you take your samplerate wrong, aliasing is SURE to happen and you
get the most fantastic waveforms on screen. The analog scope will show a screen full of garbage due to the
mice running in eachothers tracks. But the DSO will gladly fool you. It will show a rasor sharp image on
screen, showing what happened when it took a sample.
You're driving in your car, 30 mph (50 km/h), in downtown Tilburg. You have your eyes shut tight. Once every
second, you take a short peek (open up one eye) to see what and where you are. In most cases this will work
out quite well.
The other day, you drive the same route at the same speed and method. But suddenly, there's a pedestrian on
the road right behind your car. It wasn't there when you looked last. It is now. But you only look ahead so
you will never know.
Opening your eyes to see the road and traffic is similar to taking a sample of the input stream. You have a
short look. If something happens between samples, that particular event is lost.
You're sitting in your chair in the house while watching the open staircase to the second floor, eyes
closed. Five identical mice are walking up the stairs, one by one. You open up your eyes when mouse 1 is on
the first step. You open your eyes again when mouse 2 is on the second step. And again: mouse n is on the
n-th step when you look.
How many mice did you see go up the stairs? One. It walked very slowly, you mention to the other people in
the room, who had their eyes wide open. In fact, of course, there were five mice. But due to 'bad sampling'
you only looked when the previous mouse was already gone and the next mouse was already one step higher than
the previous one in the previous sample.
Same experiment as above. But now your open-eyes-interval is slightly slower than one mouse takes to go up
the flight of stairs. What happens? How many mice did you see?
Still one mouse. But now you are sure you saw ONE mouse, going DOWN the stairs, walking BACKWARDS.
It's wintertime. You live on 52 degrees lattitude. You get up at 7:30 AM and look out of the window: the
automatic light in the garden is on. You leave home for work and at 7 PM you look outside the window once
more. The lamp in the garden is on. As far as you know, the state of the lamp has not changed. Hence, the
lamp was on all day long between 7:30 AM and 7 PM.
Of course, this is silly, but the sample speed is too low so you only see what you want to see. Oversampling
is a necessity.
Examples 1 and 4 are about the samplingrate. I myself hold as a rule that the samplerate you set must be at
least 16 times the speed of the fastest signal you want to catch. If you know the signal and have a
mathematical model, you can start proving your theory by dedicated experiments, possibly with lower sample
This topic page
shows how my Meguro MSO 1270 looks like from inside and outside.
Page created March 2006,