Bandwidth importance on a scope for digital electronics

[trmatthe]

Hi. My hobby is working on old arcade boards - generally fixing dead
ones. At the moment I have a Hameg scope with 20 MHz bandwidth. I've
noticed a number of times that when looking at data lines on higher
speed boards, I get a confusing picture. For example, watching a data
line on a 12MHz 68000 gives me apparently two pictures super-imposed.
I will have a clear and bright wave showing the changing logic states,
but also a fainter set of logic transitions super-imposed.

I have a feeling that this is related to the bandwidth of my scope -
I've searched the archive here and have learnt some more information,
but I'm still not knowledgeable enough to shell out cash yet. I have a
feeling I need a higher bandwidth scope. To further confuse myself, I
have somehow convinced myself that I need a faster timebase. I can go
down to 0.5uS but was thinking that even faster would allow me a more
accurate display. As a final thought, I did wonder if the "echo" I'm
seeing on the screen is actually a badly terminated bus and I'm just
seeing reflections.

My understanding is that an faster timebase allows me to see less on
the screen, but with more accuracy - e.g. seeing the slope on a square
wave. Having been brought up in a digital world, I equate bandwidth
with resolution and accuracy.

Have I made enough sense to allow you to give me advice? As I learn
more I'll be progressing to CPUs with faster clock speeds, so I can
only assume the problem will get worse in that I could miss logic
transitions.

If none of this makes sense, I'd very much appreciate somebody
explaining the relationship between bandwidth/timebase/resolution (or
a pointer to a website?)

thanks,
Tim

 

[Marra]

Could be a triggering problem ?

If the scope is not triggering exactly on teh same point you can get
multiple displays.

I use a cheap scope similar to yours and dont have any problems.

 

[David L. Jones]

On Dec 7, 6:10 am, trmatthe <trmat...@gmail.com> wrote:

Hi. My hobby is working on old arcade boards - generally fixing dead
ones. At the moment I have a Hameg scope with 20 MHz bandwidth. I've
noticed a number of times that when looking at data lines on higher
speed boards, I get a confusing picture. For example, watching a data
line on a 12MHz 68000 gives me apparently two pictures super-imposed.
I will have a clear and bright wave showing the changing logic states,
but also a fainter set of logic transitions super-imposed.

I have a feeling that this is related to the bandwidth of my scope -
I've searched the archive here and have learnt some more information,
but I'm still not knowledgeable enough to shell out cash yet. I have a
feeling I need a higher bandwidth scope. To further confuse myself, I
have somehow convinced myself that I need a faster timebase. I can go
down to 0.5uS but was thinking that even faster would allow me a more
accurate display. As a final thought, I did wonder if the "echo" I'm
seeing on the screen is actually a badly terminated bus and I'm just
seeing reflections.

My understanding is that an faster timebase allows me to see less on
the screen, but with more accuracy - e.g. seeing the slope on a square
wave. Having been brought up in a digital world, I equate bandwidth
with resolution and accuracy.

Have I made enough sense to allow you to give me advice? As I learn
more I'll be progressing to CPUs with faster clock speeds, so I can
only assume the problem will get worse in that I could miss logic
transitions.

If none of this makes sense, I'd very much appreciate somebody
explaining the relationship between bandwidth/timebase/resolution (or
a pointer to a website?)

thanks,
Tim

What you are probably seeing there is a typical data bus signal.
Because data busses are a non-repetitive signal, and things get
switched ff and onto the data bus you will typically see a "jumble" of
signals on an analog oscilloscope display. A true repetitive signal
like the clock line will not have this problem, and you should be able
to trigger off off and display a nice clean signal.

With a 20MHz bandwidth scope, you won't be able to accurately see the
true characteristics of say a 20MHz digital (square) wave, because a
20MHz square wave is made up of much higher harmonic frequencies which
will not make it through. So your square wave will display more like a
sine wave. In fact, the base signal frequency does not matter, it's
the rise and fall time of the signal that is the key, but that's
probably more than what you need to know at this point.

See:
http://www.tek.com/Measurement/cgi-bin/framed.pl?Document=/Measurement/scopes/selection/bandwidth.html&FrameSet=oscilloscopes

A 20MHz scope will be fine to "see if there is something there", but
100MHz oscilloscopes are cheap and readily available on the used
market, even 4 channel ones, well worth having for this kind of work.
So you are not off the mark at all in thinking you need a higher
bandwidth scope, you do.

What you may find very handy though is a digital storage oscilloscope,
for capturing and analysing non-repetitive signals like serial data
lines. But these do not come cheap, and they have a whole new set of
performance criteria. But a basic 1GS/s 60MHz+ 10KB memory scope will
go a long way. All three of those numbers matter a lot in a digital
store oscilloscope. Get one of these *in addition* to the analog
oscilloscope if you can afford it.

Dave.

 

[Rich Grise]

On Thu, 06 Dec 2007 11:10:30 -0800, trmatthe wrote:

Hi. My hobby is working on old arcade boards - generally fixing dead
ones. At the moment I have a Hameg scope with 20 MHz bandwidth. I've
noticed a number of times that when looking at data lines on higher
speed boards, I get a confusing picture. For example, watching a data
line on a 12MHz 68000 gives me apparently two pictures super-imposed. I
will have a clear and bright wave showing the changing logic states, but
also a fainter set of logic transitions super-imposed.

Sounds to me like jitter either in whatever you're clocking the scope
with, the data, or the scope trigger itself.

It has nothing to do with the scope's bandwidth - that would only affect
the rise/fall time of the trace itself, and could mask overshoot and
stuff, which probably isn't a problem here.

But, to know about jitter in a 68000, you should probably ping John
Larkin - he's the local 68000 expert - when I was fixing video games,
it was all 6502s, 6800/6809s, and Z80s. ;-)

Hope This Helps!
Rich

 

[trmatthe]

Thanks for all your input - very much appreciated. I'm re-reading my
scope's manual on triggering but what David says seems to match what I
see.

Thanks again guys,
Tim