[sdiy] Moogey jitter

[Magnus Danielson]

From: "elmacaco" <elmacaco at nyc.rr.com>
Subject: Re: Re: [sdiy] Moogey jitter
Date: Tue, 18 Apr 2006 17:17:49 -0500
Message-ID: <011f01c66335$eda9e9f0$3a39af44 at santaelena>

Ed,

> > I agree, of course, that digital recording has imperfections and
> > limitations.  But measuring the zero crossings of a strong steady signal
> is
> > a perfectly valid method of looking for timing variations on the order of
> > milliseconds, which is what Kevin claims to see.  Note that in the SMPTE
> > collection of suggested measurement techniques for jitter there are
> methods
> > involving digital scopes, but there is nothing similar to Kevin's
> measurement.
> >
> <http://www.smpte.org/smpte_store/standards/pdf/rp192.pdf#search='jitter%20m
> easurement'>

One should recall that SMPTE RP192 is mainly to address the measurement of SDI (a
270 Mbps uncompressed 10-bit digital NTSC/PAL video signal - "studio quality")
signal. There are other means to measure jitter and wander than those mentioned here,
altought these can be usefull. The ITU-R specs should be looked at, or for that matter
the ITU-T & ETSI Telco-specs.

The downside of using an oscilloscope to measure jitter is that first of all, jitter
frequencies with cycle-time above the trigger-to-display time will experience a 6 dB/Oct
slope. Also, for frequencies above that, there are zeros in the response-plot. It has
become legio to measure the recovered 27 MHz clock, but that has the downside that a
number of zeroes is introduced. Demodulated phase deviations is a much better method.

For lower frequencies, frequency counters can be used directly, but even for the pure
270 MHz can direct and indirect measurments be used. Your milage may vary depending on
what exact equipment you use.

In the end, for the SDI measurement all you want is a WM-700 with the right options.
Regardless of what people say, it is the golden standard. The drift-rate display has one
particularly nice feature that I keep recommending other vendors to look at. Unfortunatly
Tektronix failed to include it in their HD-SDI instrument WFM-700, which otherwise is a
very nice HD-SDI instrument, which at the back has the demodulate output for further
jitter analysis, but that fails for wander/drift-rate work, but that is another story.
I've talked to Tek and they just plain realized that they had not understood the issue of
drift-rate requirements in HD-SDI production, especially since it ripples over to HD-2-SD
cross-production as well as MPEG2 specs. Ah well.

I find a curious joy in using the same old trimpots that I used in audio for trimming up
the HD-SDI PLL parameters (to match jitter & wander requirements). I must be getting old
and grumpy.

> Right, this is great stuff to guide us in further inquiry.  Right now we
> have established that there is Jitter in the moog oscillators, and we have a
> questionable scope measurement and someone counting 40 cycles of a wav file.
> I think we can do better than that.  What we are left with is the question
> of is the jitter significant to the sound?

See my other post. There's some hints at least.

> and perhaps jittery ones too.  If it isn't jitter then it might be something
> else.  Still, I think the whole Power supply voltages question that Kevin
> brings up is interesting enough to warrent further investigation.

The PSU of a MiniMoog isn't exactly the precission work we crunch out today.

If you want "interesting" PSU, check out the EMS VCS-3 or Synthi-A Mk I. There is a
reason for scraping the old design for the Mk II. The envelope-lamp clearly modulates the
pitch of the VCOs. Lovely!

> I see your point, but how much jitter is audible?  It seems that any
> measurement would provide fairly irrelevant numbers until we know how we
> perceive it.  if the number is small we can think it is undetactable but we
> have not proven that yet.  One might notice it but not be able to say it is
> instability, but if it is removed we might immediately hear the difference.
> We all have our suspicions, but more data would certainly help.

Consider that we have a "lively" VCO and a "quite" VCO which we in A/B testing is able
to separate. If we with the knowledge of their jitter/wander properties makes the "quite"
VCO match up with the "lively" VCO (assuming all other aspects such as overtone-spectra,
strength etc. can be considered the same) on a strict measure-wise and now in a new A/B
testing is unable to make the distinction between the two, then and only then we can say
that we have "nailed" it. But then we have only "nailed" that character, which is not
necessarilly what we are after in the first place. Doing such exercises take alot of
effort. Also, in the A/B testing one has to present them in a random order etc. so you
don't know when A or B is what you hear. When sufficient many "golden ears" have reduced
their bets down to the level of coin-tossing you have eliminated the difference.

Then again, which part of the full set of characteristics is the usefull one requires
even more tests.

> > >How much and if it is
> > >audible needs to be determined further since it is so controvercial now,
> but
> > >I don;t think moog warmth is magical, it sounds good, and no one can
> explain
> > >it satisfactorily.
> >
> > OK, it sounds good to you.  Does it sound good on CDs or just live?  :-)
> 
> Good on both counts, but live it certainly sounds different, as there is
> more perceived, and more to perceive.

Oooo... I have once orderd the gig to go on in mono, since 21:30 in a tent at a festival
nobody really cares about these little finesses where as running music rules. :o)
Live certainly leaves alot of things to desire, but it _feels_ better most of the time,
which is a different thing.

Cheers,
Magnus