Newbie pulse question

[jack]

Hi,
I have a 60 Mhz tech digital oscope and was trying to find out what
is the shortest pulse that would be faithfully reproduced on this
scope. Would a pulse with a rise time of 1 ns and decay of 8 to 10 ns
be faithfully reproduced?. As a second question,if my scope can't do
this does anyone know of a circuit that could scale every thing up so
I could view a pulse like this on this scope and (by knowing the scale
factor of the circuit) be able to deduce the real timing information
of the pulse? Thanks for any help. jack

 

[John Larkin]

On Fri, 25 Jul 2003 19:34:06 GMT, jack <road12fg@netscape.net> wrote:

Hi,
I have a 60 Mhz tech digital oscope and was trying to find out what
is the shortest pulse that would be faithfully reproduced on this
scope. Would a pulse with a rise time of 1 ns and decay of 8 to 10 ns
be faithfully reproduced?. As a second question,if my scope can't do
this does anyone know of a circuit that could scale every thing up so
I could view a pulse like this on this scope and (by knowing the scale
factor of the circuit) be able to deduce the real timing information
of the pulse? Thanks for any help. jack

Your scope's risetime will be about 6 ns, so that pulse will be pretty
severely rounded off. There's no easy way to scale a waveform in time.

John

 

[A E]

jack wrote:

Hi,
I have a 60 Mhz tech digital oscope and was trying to find out what
is the shortest pulse that would be faithfully reproduced on this
scope. Would a pulse with a rise time of 1 ns and decay of 8 to 10 ns
be faithfully reproduced?. As a second question,if my scope can't do

No.

this does anyone know of a circuit that could scale every thing up so
I could view a pulse like this on this scope and (by knowing the scale
factor of the circuit) be able to deduce the real timing information
of the pulse? Thanks for any help. jack

Risetime and pulse width are two different things, but for sure if your scope is
too slow for the rise time, you'll never see a narrow pulse.

It depends on how your scope rolls off after 60MHz but in general, the rule
relating frequency and rise time is:

F ~= 0.35 * (1/Tr).

Tr is the rise time of the scope, F is the bandwidth. For example, a 1GHz scope
has a rise time of 350ps. In general, to get accurate measurements, you want
your instrument to be 10 times faster than the signal you are viewing. So the
1GHz scope can directly accurately view 3.5ns rise times.
The fun starts when trying to define rise time. The standard is 10-90% of the
amplitude, but sometimes you'll see it defined as 0-50%. Stick to 10-90%.
A 60MHz scope will have a rise time of around 0.35/F ~= 6ns. This is a bit too
slow to glean any useful info if you know your rise time is in the 1ns range.
You can still make width measurements if the pulse is a lot wider than 6ns
though. But the constant error that the scope introduces will increase in
proportion as the pulse narrows.
If you had a 1GHz scope, you'd be closer to the required bandwidth (1GHz = 350ps
rise time)
There is an equation relating the rise time of the scope to the rise time of the
signal:
Real rise = sqr( (measured**2) - (equipment**2) )
In other words, the real rise time is the square root of the differences of the
squares of your equipment and the measured time.
*But*, in order for this to work, you have to be able to measure something, so
you need your scope to be about a third of the rise time of your expected rise
time, to get an error of about 10% with that formula.
So, you need a minimum of 1GHz bandpass to measure 1ns rise times.
Your 60MHz bandwidth scope's rise time is 18 times too slow.
Sorry...

 

[jack]

Thanks for explaining this. It's as I feared,and I'll guess I'll have
to explore the possibility of a scaling circuit ,if some such animal
exits. Much appreciated to all. jk

On Fri, 25 Jul 2003 17:38:56 -0400, A E <aeisenhut@videotron.ca>
wrote:

jack wrote:

Hi,
I have a 60 Mhz tech digital oscope and was trying to find out what
is the shortest pulse that would be faithfully reproduced on this
scope. Would a pulse with a rise time of 1 ns and decay of 8 to 10 ns
be faithfully reproduced?. As a second question,if my scope can't do

No.


this does anyone know of a circuit that could scale every thing up so
I could view a pulse like this on this scope and (by knowing the scale
factor of the circuit) be able to deduce the real timing information
of the pulse? Thanks for any help. jack

Risetime and pulse width are two different things, but for sure if your scope is
too slow for the rise time, you'll never see a narrow pulse.

It depends on how your scope rolls off after 60MHz but in general, the rule
relating frequency and rise time is:

F ~= 0.35 * (1/Tr).

Tr is the rise time of the scope, F is the bandwidth. For example, a 1GHz scope
has a rise time of 350ps. In general, to get accurate measurements, you want
your instrument to be 10 times faster than the signal you are viewing. So the
1GHz scope can directly accurately view 3.5ns rise times.
The fun starts when trying to define rise time. The standard is 10-90% of the
amplitude, but sometimes you'll see it defined as 0-50%. Stick to 10-90%.
A 60MHz scope will have a rise time of around 0.35/F ~= 6ns. This is a bit too
slow to glean any useful info if you know your rise time is in the 1ns range.
You can still make width measurements if the pulse is a lot wider than 6ns
though. But the constant error that the scope introduces will increase in
proportion as the pulse narrows.
If you had a 1GHz scope, you'd be closer to the required bandwidth (1GHz = 350ps
rise time)
There is an equation relating the rise time of the scope to the rise time of the
signal:
Real rise = sqr( (measured**2) - (equipment**2) )
In other words, the real rise time is the square root of the differences of the
squares of your equipment and the measured time.
*But*, in order for this to work, you have to be able to measure something, so
you need your scope to be about a third of the rise time of your expected rise
time, to get an error of about 10% with that formula.
So, you need a minimum of 1GHz bandpass to measure 1ns rise times.
Your 60MHz bandwidth scope's rise time is 18 times too slow.
Sorry...

 

[A E]

jack wrote:

Thanks for explaining this. It's as I feared,and I'll guess I'll have
to explore the possibility of a scaling circuit ,if some such animal
exits. Much appreciated to all. jk

Why yes it does, it is called equivalent-time sampling, and it has a very interesting
history to it as well.
It only works if the signal is repetitive though.
Basically you trigger on the same point of the signal, and on every trigger cycle you
sample a tiny point a little bit further along every time.
You can get a feeling for this by reading some papers like this:
http://www.tektronix.com/Measurement/App_Notes/sampling_primer/47W_7209_0.pdf
http://www.caip.rutgers.edu/~kahrs/electronics.html

Or to get a look at a sampler manual from the 60s:
http://www.dfpresource.org/pdf/tek1s1_manual.pdf (warning 24M) Especially section 3.

Some nice guy went ahead and scanned and OCRd this thick manual... Look at the
schematics, aren't they clear? Wow.

The 1S1 uses a bunch of exotic diodes that you can't get these days, but just look at
the concept. Stuff like this is why I love the 1960s!

Thing is, you can't just build one of these on a weekend on a piece of Radio Shack
breadboard. Even though the output of such a sampler is in the KHz range, the
front-end is working at GHz frequencies, and you end up being in the position of
needing a sampling scope to check if your sampler has the correct waveforms on it,
and you need to know how to build RF circuitry. Also, without some kind of active
probe, sampling scopes have low input impedance (50ohms). And of course, you need
some parts somewhere that have the speed to work at those frequencies. No 2N2222s
here.

If this is for work, suggest to the higher ups that you really need a better scope.
I'm sure renting one for the required time is a good idea.
If this is for hobby use, you can hit eBay and scrounge for sampling heads. All the
RF work is done for you, but you still need to design some very high speed circuitry
for driving the head. They are fragile though. And you don't have a trigger circuit.
And you still need to design a fair bit of other stuff around it.

 

[jack]

Alas,I think I'm cooked then as my signal is a non repetitive
pulse,actually Poisson distribution I hope, the output from a PMT with
coherent light on the cathode. Again,thanks. jk PS the idea of scope
rental for a day might do it though!
n Sat, 26 Jul 2003 12:28:42 -0400, A E <aeisenhut@videotron.ca> wrote:

jack wrote:

Thanks for explaining this. It's as I feared,and I'll guess I'll have
to explore the possibility of a scaling circuit ,if some such animal
exits. Much appreciated to all. jk

Why yes it does, it is called equivalent-time sampling, and it has a very interesting
history to it as well.
It only works if the signal is repetitive though.
Basically you trigger on the same point of the signal, and on every trigger cycle you
sample a tiny point a little bit further along every time.
You can get a feeling for this by reading some papers like this:
http://www.tektronix.com/Measurement/App_Notes/sampling_primer/47W_7209_0.pdf
http://www.caip.rutgers.edu/~kahrs/electronics.html

Or to get a look at a sampler manual from the 60s:
http://www.dfpresource.org/pdf/tek1s1_manual.pdf (warning 24M) Especially section 3.

Some nice guy went ahead and scanned and OCRd this thick manual... Look at the
schematics, aren't they clear? Wow.

The 1S1 uses a bunch of exotic diodes that you can't get these days, but just look at
the concept. Stuff like this is why I love the 1960s!

Thing is, you can't just build one of these on a weekend on a piece of Radio Shack
breadboard. Even though the output of such a sampler is in the KHz range, the
front-end is working at GHz frequencies, and you end up being in the position of
needing a sampling scope to check if your sampler has the correct waveforms on it,
and you need to know how to build RF circuitry. Also, without some kind of active
probe, sampling scopes have low input impedance (50ohms). And of course, you need
some parts somewhere that have the speed to work at those frequencies. No 2N2222s
here.

If this is for work, suggest to the higher ups that you really need a better scope.
I'm sure renting one for the required time is a good idea.
If this is for hobby use, you can hit eBay and scrounge for sampling heads. All the
RF work is done for you, but you still need to design some very high speed circuitry
for driving the head. They are fragile though. And you don't have a trigger circuit.
And you still need to design a fair bit of other stuff around it.