Jim Lesurf wrote in
:

In article ,
Lostgallifreyan

Jitter has a surprising effect in using audio outputs with DC blocking
caps renoved for DC coupling for laser projector galvo scan amp drive.

FWIW so far as I know correctly implimented spdif includes a transformer or
equivalent. Mainly to block loops and common mode, but it would also block
dc. So act similarly to a dc blocking cap IIRC.

[big snip]

Afraid I don't know what some of the terms you use like 'ILDA' and 'WDM'
mean in terms of the waveform details. However...

Anyway, what I found was that this interface has two main driver
variants, one of them a WDM version. If you run this pattern without
blanking the beam so you see the draws as the beam is quickly
positioned for the next drawn element in the pattern, you can see a
partial loop, like a bight of rope, as it leaps from the bottom of the
circle to a small vertical line drawn above it. Scanner galvos have a
lot of mass, they're like heavy duty fast moving- magnet meters, and I
don't know how much this relates to audio systems, but it appears as if
the resonance in a scanner being fixed, reacts with the small timing
errors in the non-WDM driver to form a dramatic and erratic shift in
the 'bight' formed by the scanners. It affects both axes, and the
vertical axis saw it shift in a range as great as 10 to 15 percent of
full scale.

How anyone interprets all this is up to them.

My main interpetation is that the waveform shape 'as scanned' is altered.
That is the equivalent of saying 'jitter exists and can be caused by the
transmission system having a frequency response that isn't flat'. So not a
surprise.

I only mention it because it happened, and I found it interesting. There
is no doubt that removing jitter makes the resonance play nice, but
whether this would be important in sound playback I don't know.

The significant point is that 'playback' requires a DAC which is clocked in
some way. If:

A) The DAC uses a clock that does *not* in the medium and short term depend
on the embedded pilot tone in the spdif.

B) The spdif 'jitter' is reasonably small compared with the bit-length
cycle

Then the 'jitter' on the signal is essentially suppressed by the 'replay'
system.

So you'd need to measure 'jitter' on the *output* from the *replay system*
- not just observe it on the spdif waveform - to assess the level so far
as any listener is concerned.

IIUC The systems you described drive the scanning directly, or by some
subsystem 'receiver' whose details are not clear to me from what you wrote.
In essence what you described would then be like looking at the spdif
waveform on a scope - not looking at the DAC output.

Hence like yourself I have no idea if what you say actually has any
relevance to the audio case.

I'd imagine that our hearing might react to it as sight reacts to
dither. It might even sound 'better' in many cases, if we notice it at
all.

Alas the problem these is as I previously commented. Many things "might be
so", but only some of them (or none) *are so*. Indeed, imagining something
doesn't even show it is 'possible'... :-)

I don't doubt that at a suitably high level 'jitter' would be audible. Wow
and flutter if bad enough are easily audible. But that doesn't tell us if
what people claim is 'true'. :-)

Slainte,

Jim


Yes, there's a DAC in the Layla 24/96. (Is its main task). I noticed that the
WDM driver (Windows Driver Model) under Windows XP would make the 'bight' in
the scan even and stable. The point here is that the audio interface driver
itself (and perhaps the operating system's context for it) had small timing
errors. While you're talking of digital signals, the same might be true here
because the software driver has no clue whether the output would later be
digital or analog, yet the driver was clearly a source of jitter (because the
stable output of the WDM driver in WXP proved that the hardware wasn't), so I
assume a software driver could be so for a digital OR analog output in other
systems too.

No worries about the ILDA thing, the crucial point in that context (given for
completeness only) is that this occured during a very short moment, where a
mechanical system has to travel some largish chunk of full-scale deflection
in the duration of one or two samples. Note that the resonance is in no way
causal, not on its own anyway. I think it has to do with the timing over a
single sample duration varying in moderately large proportion to the nominal
sample rate. it might also be specific to the way a scanner amplifier uses
closed loop positional feedback, so maybe could only show up in an audio
speaker if it also used high output drive and a servo-type system to attempt
to get high tracking of an input waveform. Such systems exist but I don't
think many people use them. (Must cost a lot..)

Anyway, my smoke is in danger of masking any fire, I just added this stuff
because I thought it might help give a usefully different perspective on an
actual case of jitter. As the hardware is a professional high-end recording
system used in many studios during its time, I doubt that jitter had any
adverse effect on its sound. I never had any clue it was there till a
scanning galvo pair showed it to me.