From: "Bevan Weiss"
References: <3DB1F8E0.1464@Spam.Bots> <4Vrs9.92$0J6.firstname.lastname@example.org> <3DB27E3A.F0C@Spam.Bots> <0Wts9.218$0J6.email@example.com> <3DB3411D.1B53@Spam.Bots> <7WMs9.18$JW5.firstname.lastname@example.org> <7ZDLAbAr77s9Ew6C@jmwa.demon.co.uk> <email@example.com> <firstname.lastname@example.org> <4Zsu9.378$lU1.email@example.com>
Subject: Re: Aylward, Engelhardt: Noise in a transient sim?
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
Organization: Orcon Internet Ltd
X-Cache: nntpcache 2.3.3 (see http://www.nntpcache.org/)
X-Original-Trace: 26 Oct 2002 22:20:14 +1300, news.orcon.net.nz
Date: Sat, 26 Oct 2002 22:20:17 +1300
NNTP-Posting-Date: Sat, 26 Oct 2002 22:20:17 NZDT
I agree with what you are saying Kevin. A signal of 7GHz does not
necessarily imply that the bandwidth of the intelligence is 7GHz. You may
very well sample at 1Hz, if that is greater than 2x the intelligence
bandwidth. You also require a mixer however to remove the frequency
Otherwise you'll get all that other aliased crap as well as your desired
And other peoples intelligences just get in the way of a problem like this
"Kevin Aylward" wrote in message
> "Mike Monett" wrote in message
> > "Kevin Aylward" wrote in message
> > [...]
> > > > Kevin, I don't know how you got so far off track. I'm talking
> > > > sampling a repetitive waveform, the same as a conventional Tek or
> > > > sampling scope.
> > >
> > > I know you are.
> > >
> > > >The signal is any waveform that can be viewed on a
> > > > conventional scope, including eye patterns. The signal can be
> > > > from dc to 7GHz, although dc is somewhat difficult due to thermal
> > > > drift. Even so, the binary sampler is far better than a diode
> > > > bridge.
> > > >
> > >
> > > No problem here. I am taking issue with the claim that the signal
> has a
> > > (noise) BW of 7GHz. It most certainly does not.
> > What's your problem. The instantaneous bandwidth and noise bandwidth
> > are the same.
> I was pointing out the fact that the true signal BW, *is* the signal
> noise BW. This should indicate quite clearly, that a claim of
> 0.01nv/sqrth over a BW of 7Ghz, is quite nonsense. Why you don't see
> this is pretty amazing really.
> > > > Conventional sampling scopes have so much noise it is difficult to
> > > > measure signals below 1mV. With the binary sampler, detecting a
> > > > signal is easy, 1uV is difficult, and below 1uV looks unreachable
> > > > the moment. This is far better performance than can be obtained
> > > > conventional sampling technology.
> > > >
> > > > Does this help?
> > >
> > > Ahmm.. I understand all of this very well indeed, this is all very
> > > elementary. However, it seems that the subtleties I am pointing out
> > > not really being appreciated.
> > That is a good possibility. I have the feeling you are inventing
> > definitions to suit your arguments. These definitions change depending
> > on your position. Give me some equations that express your viewpoint,
> > then we can discuss.
> I have explained quite clearly, *accepted* facts of signal processing. A
> pure repetitive signal has zero information BW. Its that simple.
> > > Yes, the signal can vary from DC to 7 GHz, but this is *not* the
> same as
> > > saying that the signal has a *BW* of 7Ghz. The signal is not
> varying in
> > > a continuous manner, such that while it is varying, you are
> > > its information content. You are sampling a fixed repetitive signal,
> > > therefore it has a very limited BW, therefore you can filter out the
> > > noise without filtering out the signal.
> > Not exactly. The signal risetime defines the bandwidth needed to
> > preserve the frequency components.
> Technically, its the shape of the pulse.
> >All I need to know is the
> > fundamental frequency, then I can give you the waveshape which means
> > the harmonic content and phase relationship. The same as any
> > conventional sampling system is defined.
> You just don't seem to *want* to understand, the distinction between a
> repetitive signal and a random signal. A repetitive signal has no BW.
> End of story.
> Look, if *any* signal is on continuously, I can just keep on sampling it
> whenever I like, say once every second even. Ultimately, I can build up
> what that signal looks like, no matter what ever what its base frequency
> is. This is what the sampling theorem *means*. I can sample at a rate of
> twice a signals *BW* to recover it. Furthermore, I have explained, such
> a method how you can do this, i.e. with multiple multipliers.
> > > In summary, you are not measuring a 1uv, 7 GHz BW signal, you are
> > > measuring a signal that varies from DC to 7Ghz, that has a very
> > > BW. You are confusing signal frequency, with signal BW, they are not
> > > same.
> > No, Kevin. Any sampling scope has a defined bandwidth. I am using the
> > same definition.
> Unfortunately, a novice understanding/definition of BW from scope
> vendors manuals targeted at technicians, is not sufficient to explain
> the more advanced concepts required in signal processing.
> I explained to you what *real* BW meant. Your making the classic mistake
> equivalent to "how come I can sample a 100Mhz centred signal at a 10khz.
> The reason is that the sampling theorem is with regards to BW, not
> signal frequency. You are still confusing the two. A 7Ghz signal, does
> *not* necessarily, in fact it rarely does, take up 7Ghz of BW. Its that
> > Thank you for realizing the binary sampler has performance advantages
> > that cannot be achieved with current sampling technology.
> > Since there is no new information or insight in this thread, there is
> > little need to continue this discussion.
> I agree, unfortunately, you are a bit out of your theoretical depth. You
> need to take a decent course on signals and information theory. I have
> tried to explain to you where you are mistaken, but rather then
> understand the *accepted* theory behind it, you deny it in ignorance.
> Its your loss.
> Kevin Aylward
> SuperSpice, a very affordable Mixed-Mode
> Windows Simulator with Schematic Capture,
> Waveform Display, FFT's and Filter Design.