ard at p850ug1.demon.co.uk
Fri Apr 8 17:20:51 CDT 2005
> I'm rapidly approaching the point where I will
> need to learn how to use an oscilloscope as part
> of my quest to learn more about debugging the
> old DEC hardware. What kind of advice can folks
> offer to me about a decent scope and some info on
> how to use one? I think I used one 30 years ago
> in school, but it's been so long that I remember
> little-to-nothing about it.
OK, a little introduction...
The basic idea of a 'scope is to have a CRT where the spot is deflected
horizontally by the X-plates and vertically by the Y plates. Therefore
the picture on the screen is a graph of the voltage applied to X against
the volage applied to Y.
Now, normally the X-voltage is a ramp signal produced by a circuit called
the 'timebase'. The spot on the screen starts at the left side, moves
across at a constant speed until it gets to the right hand side, then
quickly 'flies back' to the left side and starts again. This means the
screen image can be considered to be a graph of the signal on the Y
plates against time.
There is also normallly an amplifier that takes the signal under test and
amplifies it to apply to the Y plates, for obvious reasons.
Finally, there's a trigger circuit. It's obviously essential (at least
for repetitive waveforms) that the trace starts on the same point of the
waveform each time. Otherwise you'll not get a steady display. Same point
really means same voltage and sloping in the same direction (either up or
down). This then starts the timebase ramp.
Notice I said 'normally' when I said the X-axis was the timebase signal. On
most 'scopes yuo can apply external signals to both X and Y. This is how
you get the lissajou's figure beloved of science fiction film producers,
and which do actually have some real use for comparing the phase of 2
sinusoidal signals. I also said 'normally' when I mentioned the Y
amplifier. There have been a few 'scopes where the input is applied
directly to the CRT plates, normally because no normal amplifier would
have a high enough bandwidth. The Tektronix 519 is the clasic example of
this. You DO NOT want one of these for general-purpose work.
Some 'scopes can display multiple traces -- that is serveral signals at
the same time, normally on one common timebase. You can then compare the
relative phase or timing of those 2 signals. There are 2 main ways to do
this -- double beam scopes have 2 electron guns and display 2 spots
simultaneously. The CRT has separate Y plates for the 2 beams and
normally a common set of X plates but there have been 'scopes with
separate X plates for each beam too. The other way is to electronically
switch one set of plates between the 2 signals -- either 'chopping' which
involbes swithcing between the 2 signals in much less than the timebase
sweep time (so effectively you are displaying them simultaneously) or
'alternate' where you select one input, sweep the timebase (thus
displaying it), then select the other input, sweep the timebase again,
and so on. Note that some double-beam 'scopes can take multiple trace
plug-ins on each input. So you might have one beam switched between 2
signals, the other switched between 4, and thus have 6 traces o nthe
screen at once...
Storage 'scopes are always useful. This means the image on the screen is
stored, either electrostatically in the CRT, or by digitising the input
signal, then 'playing it back' to the CRT. This means you just need one
sweep of the timebase to capture the signal, you can then look at it for
as long as you (sensibly) like. This is obviously useful for
non-repetitive signals, which can't be usefully displayed on a repetitive
timebase and a non-storage CRT.
Delayed timebase is also very useful. What this does is to detect the
trigger, then wait a certain (settable) time, then sweep the timebase.
You might, for example, waits 10ms, then sween the timebase in 100us.
This lets you 'enlarge' a small section of a complex waveform.
Differential Inputs. Many 'socpes can display the difference in voltage
between 2 input sockets, ignoring the voltage between them and ground
(there will be a limit on this voltage, so don't try applying 1000V or
anything stupid). You might want to look at the outputs of the
differential amplifiers in a disk drive read chain or something
Plug-ins. Many of the better older 'socpes had the Y input stages (at
least) as plug-in modules that could be replaced with other modules. You
might get simple amplifiers, multiple-trace amplifiers, differential
amplifiers, curve tracers, etc. Personally, I like this idea, but
collecting the plug-ins becomes as addictive as collecting computers...
Now, as to user controls :
1) CRT images controls. Birghtness and Focus, basically. Keep the
brightness as low as you can, both to protect the CRT phosphor and to
give a sharper trace
2) Shift controls (X and Y) move the trace around the screen. Pretty
obvious. Multiple-trace 'scopes have a Y shift control for each trace.
Multiple timebase 'scopes have an X shift control for each timebase.
3) Y amplifier gain. There'll be a switch calibrated in 'V/cm' or 'V per
division' or something like that. The meaning is pretty obvious. If you
set it to 2V/cm and the trace is 1.5cm high on the screen, that's 3V.
There will also be an adjustable control with a click-stop at one end.
This lets you very the gain, at the click-stop position you get the gain
marked on the switch.
4) Timebase speed. Similar in concept to the gain controls, but
calibrated in s/cm (or ms/cm. us/cm). If a trace on the screen taks 5cm
for a complete cycle, and you have the timebase set to 10ms/cm. then
that's a period of 50ms, or a frequency of 20Hz. Again there will be a
variable control with a click-stop to get you the calibrated times
5) Trigger. There will be a trigger level control that sets the voltage
level that starts the ramp, and a slope switch that selects between
rising and falling. THere will also normally be a trigger selector, which
lets you take the trigger sigal either from the Y amplifier (normal for
simple work), another input socket (you might, for example, want to
trigger from the index signal of a disk drive, but display the output of
the read amplifier -- in fact that's exactly what you do for head
alignment), 'line' (trigers from a transformed-down mains signal, useful
for seeing if a ripple comes from the mains or not), and maybe some 'TV'
modes, useful for locking to a video signal and useless for almost
As to good 'socpes. I would recomend a Tektronix, there are very few bad
designs -- but there are some very specialised ones that are not suitable
for what you want. Depending on the sort of instrument you want, you
might consider :
545, 547, 555 -- Old, heavy, full of valves, and built beautifully.
People collect these like we collect PDP11s, but some people also like
having a more modern 'scope alongside them, rather as most people who
have classic computers have a modern PC too (I don't!. I have my 555 and
love it). Plug-in amplifiers, very versatile. The manuals are nothing
short of excellent!
561, 564. Smaller, valved, not as high bandwidth IIRC. The 564 is a CRT
465, 466, 468, etc. Smaller, portable. I think the 468 is a storage
'scope, one of that series is, anyway. No plug-ins. Transistorised.
7000 series. Plug-ins for X ad Y. Full of custom ICs that are essentially
unobtainium now. I used a 7904 (2 Y plug-ins, 2 X plug-ins, 500MHz
bandwidth) at university, and loved it. But due to the custom chips I'd
not want one as my only 'scope
Avoid the 5000 series. They're low-bandwidth devices, not what you want
The manuals have really gone down for more recent 'scopes. You don't even
get a schematic now...
More information about the cctalk