From: Mike Monett
X-Mailer: Mozilla 2.02 (Win16; I)
Subject: Re: Binary Sampler
References: <3E2A07F4.email@example.com> <3E2A5081.4C97@sneakemail.com> <3E2C7D00.C7D@sneakemail.com> <3E2D7074.645A@sneakemail.com> <firstname.lastname@example.org> <3E2D9D5B.768A@sneakemail.com>
Date: Tue, 21 Jan 2003 15:51:18 -0500
NNTP-Posting-Date: Tue, 21 Jan 2003 15:50:46 EST
Organization: Bell Sympatico
Mike Monett wrote:
> John Larkin wrote:
> > On Tue, 21 Jan 2003 11:08:20 -0500, Mike Monett
> > wrote:
> > >The amazing thing is how it acquires the signal after slewing through a
> > >large range. The acquisition takes one sample, just as the simulations
> > >show. There is no overshoot, ringing, or long dribble time as it settles
> > >on the new amplitude. The capture is instantaneous.
> > >
> > What's amazing is how slow it is. It's like crawling across the
> > desert, and being surprised at how fast you can stop.
> > John
A conventional sampling scope samples at a maximum of 100,000
samples/sec. 1,000 samples of a waveform takes 1e3/1e5 = 10mS.
To get 1,000 samples of a waveform with the binary sampler, the
offset frequency is 1/1,000 of the signal frequency.
For a 1 GHz signal, the offset is 1MHz. For 1,000 samples, it takes
1e3/1e6 = 1mS to acquire the waveform. The binary sampler is ten
times faster than a conventional sampler.
Since we have 1,000 sample points, the maximum slope between samples
is 0.1% before slew rate limiting occurs. If the slope of the
waveform exceeds this, we have to go to a lower offset frequency
which slows down the binary sampler.
The binary sampler tells you when the slew rate is exceeded. Look
for long strings of ones or zeros. The probability of 16 bits in a
row is 1/2^n, or 1/65,535.
Going to 100,000 KHz offset improves the response to fast signal
edges by a factor of ten, but the accuracy of each sample is also
ten times better, or 0.01%. It now takes 10mS to capture the
waveform, the same as a conventional sampler.
The slope effect also occurs in conventional samplers. To get
accurate waveform capture with fast edges, the conventional sampler
With a conventional sampling scope, ringing and overshoot account
for 3% to 5% error. Dribble-up and dribble-down are not spec'd.
(I may have made errors in the above calculations. Please check my