Maker Pro
Maker Pro

Scope for UWB

O

oopere

Jan 1, 1970
0
We are in the process of acquiring a fast (12GHz) scope for testing
impulsive UWB systems. We are considering the agilent DSO 81204 and
the tektronix DSA 7124. Can anybody share his/her experience on any of
these?

I have been using tek scopes for a long time and think they are
excellent. Almost no experience with agilent scopes excepting a mixed
signal scope. However, in a demo I saw, a tektronix DSA 70804 seemed
not to work as nicelly as could be expected: visible distortion in a
1GHz sinusoidal signal, plus some non-intuitive zooming after a single
shot capture. Seems there are different A/D techniques in both scopes.
Comments are welcome!

Thanks

Pere
 
P

Phil Hobbs

Jan 1, 1970
0
oopere said:
We are in the process of acquiring a fast (12GHz) scope for testing
impulsive UWB systems. We are considering the agilent DSO 81204 and
the tektronix DSA 7124. Can anybody share his/her experience on any of
these?

I have been using tek scopes for a long time and think they are
excellent. Almost no experience with agilent scopes excepting a mixed
signal scope. However, in a demo I saw, a tektronix DSA 70804 seemed
not to work as nicelly as could be expected: visible distortion in a
1GHz sinusoidal signal, plus some non-intuitive zooming after a single
shot capture. Seems there are different A/D techniques in both scopes.
Comments are welcome!

Thanks

Pere

I recently bought a refurb 7 GHz, 20 GS/s Tek scope (I forget the model
number, they're all Wintel PCs with data acq hardware nowadays.) I
bought the old one because the new one had something like 8% overshoot
on the step response. (No, this is not Gibbs effect due to brick-wall
filtering. It's due to a crappy vertical amplifier.)

The new one was an early production sample, and I really leaned on the
rep about it...so maybe it got fixed, I don't know. A scope with edge
artifacts like that isn't too useful, even if you _can_ play Halo on it.
The Agilent folks seems to be going through one of those occasional
periods in which they discover how to make scopes. They will probably
forget again soon--the HP/Agilent track record in oscilloscopes is
mixed, to put it kindly. I'll never forget the joy I felt about 18
years ago in finding a _digital_ HP scope on the corporate surplus list
for free. When it arrived, I discovered that I had to drill down three
menu levels to set the vertical gain. It went right back on the surplus
list.

One thing that Agilent continually gets wrong is that they refuse to
show you the actual measurement data. The trace on the screen has been
massaged, compensated, filtered, and totally pimped by the time you see
it. For my money, that's as bad as the ugly Tek impulse response.

To the scope makers of the world: Just show me the measured data and let
me interpret it myself.

Cheers,

Phil Hobbs
 
O

oopere

Jan 1, 1970
0
I recently bought a refurb 7 GHz, 20 GS/s Tek scope (I forget the model
number, they're all Wintel PCs with data acq hardware nowadays.) I
bought the old one because the new one had something like 8% overshoot
on the step response. (No, this is not Gibbs effect due to brick-wall
filtering. It's due to a crappy vertical amplifier.)

The new one was an early production sample, and I really leaned on the
rep about it...so maybe it got fixed, I don't know. A scope with edge
artifacts like that isn't too useful, even if you _can_ play Halo on it.
The Agilent folks seems to be going through one of those occasional
periods in which they discover how to make scopes. They will probably
forget again soon--the HP/Agilent track record in oscilloscopes is
mixed, to put it kindly. I'll never forget the joy I felt about 18
years ago in finding a _digital_ HP scope on the corporate surplus list
for free. When it arrived, I discovered that I had to drill down three
menu levels to set the vertical gain. It went right back on the surplus
list.

One thing that Agilent continually gets wrong is that they refuse to
show you the actual measurement data. The trace on the screen has been
massaged, compensated, filtered, and totally pimped by the time you see
it. For my money, that's as bad as the ugly Tek impulse response.

To the scope makers of the world: Just show me the measured data and let
me interpret it myself.

Cheers,

Phil Hobbs

Thanks for your inputs. Now I got to see an Agilent scope working with
the same sinusoidal input signal. It really seems that there is less
distortion. Agilent folks says this is due to very good matched A/D
converters (at these speeds it seems the only choice is to interleave
several A/D converters: Agilent interleaves 2 and Tek 5).
However, your comment saying that Agilent refuses to show actual data
has triggered an alarm: it could be that agilent's scope is cheating
(i.e. filtering), because there was really a huge difference in the
trace appearence for a simple sinusoid. In the Tek scope, it really
seemed that some samples were either taken at the wrong time or there
was a huge uncertainty in the quantization.

Pere
 
I

Ian

Jan 1, 1970
0
oopere said:
Thanks for your inputs. Now I got to see an Agilent scope working with
the same sinusoidal input signal. It really seems that there is less
distortion. Agilent folks says this is due to very good matched A/D
converters (at these speeds it seems the only choice is to interleave
several A/D converters: Agilent interleaves 2 and Tek 5).
However, your comment saying that Agilent refuses to show actual data
has triggered an alarm: it could be that agilent's scope is cheating
(i.e. filtering), because there was really a huge difference in the
trace appearence for a simple sinusoid. In the Tek scope, it really
seemed that some samples were either taken at the wrong time or there
was a huge uncertainty in the quantization.

Pere
There's no way to cheat - the next signal you look at might be a
completely different waveform. It is just easier to see the effects
on a sine wave. You can see the effects on a square wave readily
in the frequency domain.

There's an article in the Agilent Measurement Journal about some
of what they did at:
http://www.home.agilent.com/agilent...&lc=eng&ckey=893083&nid=-11144.0.02&id=893083

Regards
Ian
(Agilent employee)
 
P

Phil Hobbs

Jan 1, 1970
0
Ian said:
There's no way to cheat - the next signal you look at might be a
completely different waveform. It is just easier to see the effects
on a sine wave. You can see the effects on a square wave readily
in the frequency domain.

There's an article in the Agilent Measurement Journal about some
of what they did at:
http://www.home.agilent.com/agilent...&lc=eng&ckey=893083&nid=-11144.0.02&id=893083

Regards
Ian
(Agilent employee)

It isn't strictly true that there's no way to cheat--except on a
single-shot measurement. Repetitive sine waves are easy.

Cheers,

Phil Hobbs
 
I

Ian

Jan 1, 1970
0
Phil Hobbs said:
It isn't strictly true that there's no way to cheat--except on a
single-shot measurement. Repetitive sine waves are easy.

Cheers,

Phil Hobbs

Averaging on repetitive waveforms acts to smooth out noise.
Systematic errors in the timing of the interleaved A/D's can't
be fudged like that, you have to find a way to fix the problem.
The article also says specifically that the sine wave plots were
taken as a single-shot measurement.

Regards
Ian
 
Top