🚀 go-pugleaf

In article <47ocb4$16bq@serra.unipi.it> ferrovia@pimac2.iet.unipi.it (Massimo Macucci) writes:
>We are in the process of purchasing a new digital scope for our lab.

>Tektronix TDS784 (1 GHz, 4 ch, 4 GS/s, 50 ksample/ch)
>Tektronix TDS744 (0.5 GHz, 4ch, 2 GS/s, 50 ksample/ch)
>LeCroy 9370M (1 GHz, 2 ch, 2GS/s, 500 ksample/ch)
>LeCroy 9350AM (0.5 GHz, 2ch, 1GS/s, 500 ksample/ch)
>HP 54522A (0.5 GHz, 2ch, 2GS/s, 32 ksample/ch)

>We would like to know the opinions of people who have been using some of
>these scopes for a while, in order to help us making a decision. We did
>not have a chance of measuring the effective bit number vs. frequency on the
>demo units we had available, and there are rumours that the LeCroy's could
>perform worse than the Tektronix's from this point of view.
> . . . .
>We would like to have some feedback from users of the scopes I listed at
>the beginning, in particular about the accuracy, and the actual number
>of effective bits.

I have measured effective bits on both Tek 'scopes, and on 4-channel
versions of the LeCroy and HP 'scopes.  (It's part of my job to make such
measurements, since I design 'scope front-ends at Tek.)  But assuming
you don't want to take my word for it, since I am not an independent
third-party reviewer, let me suggest a way to observe the relative
acquisition accuracy of the DSOs yourself.  This can be done in a few
minutes time on demo 'scopes if you have the signal source available.

Connect to the 'scope under test a pure sine-wave.  The input frequency
should be near the top end of the frequency range for which you plan on
using the 'scope, but not a small-integer harmonic relation to the sample
rate (so that harmonics generated within the 'scope will not alias onto
the fundamental).  Adjust the input amplitude to cover most of the screen
(e.g. 90% of full-scale).  A good source for a pure sine-wave is a
frequency-synthesized signal generator followed by a low-pass or band-pass
filter to remove harmonics.  (In fact, a major reason for using a sine-wave
for the test is because it is relatively easy to generate known-pure sine-
waves in this manner.)  A high-frequency, high-amplitude sine provides the
best opportunity to look for dynamic errors within the 'scope's front-end.

Run an FFT on the acuired data (FFTs are available in all of the 'scope
models mentioned above), and observe the relative height of the fundamental
component (the true input) and all the other spurs.  Harmonic distortion
will show up at frequencies that are an integer multiple of the input, or
aliases thereof.  For 'scopes that interleave several channels to obtain
higher sample rate, interleaving errors will show up at frequencies that
are separated from DC or the fundamental by integer multiples of Fs/N, where
N channels are interleaved.  Yet other spur frequencies can result if there
is phase modulation of the sample clock (e.g. due to signal crosstalk) or
various other errors.

Keep in mind that 6 dB represents a factor of two in accuracy.  In other
words, if the spurs in one 'scope's FFT are 6 dBc higher than in another's,
the errors are twice as big.

I would be glad to provide more information on how to interpret FFTs of
sine-wave inputs in evaluating 'scope acquisition error sources, if people
are interested.  I'm not going to provide the measurements of specific
models (though I've about done them all), since this is not an appropriate
place to advertise.

- Dan Knierim
These opinions aren't worth the standard disclaimer form they're printed on!