This is a very interesting topic, unfortunately its also fairly
complicated. 

First I'll cover a few basics, then in subsequent posts cover some
details.

As has been mentioned before, as far as jitter goes whats really
important is the jitter at the DAC chip itself. There are MANY
different paths for jitter to get there, some internal and some
external. There is no such thing as a jitter free oscillator or a
jitter free reclocking circuit. All you can do is aim for jitter
reduction, its impossible to completely eliminate it. (no matter what
some marketing departments claim). DAC chips themselves add jitter
inside the chip. 

Besides intrinsic jitter of the parts themselves, jitter can be added
by noise on the power supply lines and noise on the ground lines. As
the intrinsic jitter of parts get lower these external influences
become a larger proportion of the jitter in a system. For example the
oscilator circuit in a Touch has about 40ps intrinsic jitter. A fairly
small amount of noise on the PS and ground can almost double that.
Compare that to a good low jitter Crystek oscillator with intrinsic
jitter about 2ps, that same amount of noise will give you about 15 to
20ps jitter. In this case the intrinsic jitter is completely swamped by
the PS noise induced jitter.

Thus even when using very good low jitter parts your system can be very
susceptible to PS noise in the system. And I'm not just talking about
throwing an extra chip regulator in to deal with the issue. The noise
I'm talking about is caused by the digital signals themselves running
through the system. Every time an edge happens current spikes flow
through the PS and ground planes. There is enough inductance even in
solid ground planes that this current causes voltage drops. There is
nothing you can do about this, it happens when dealing with digital
signals. The more stuff you have going on in your system the more of
this noise you have. Its not a lot of noise, but it is there, and once
you get your parts and circuits good enough the effects of this noise
are the major contributor to jitter in a system. 

All this is to shed some light on sentiments in this thread such as
"this DAC uses a low jitter clock and jitter reduction circuitry, so it
can't possibly be sensitive to anything coming over the input, so
claiming one source or another is better is ludicrous". This simply
isn't true. Once you get the basics taken care of (such as low jitter
clock, asynchronous interface, low jitter reclocker, very low noise PS,
etc.) this ground plane noise starts becomming the dominant factor in
the design. And part of that ground noise comes from the input. 

The USB subsystem is chugging away processing data at the rate its
coming over the bus with whatever timing is coming out of the host and
jitter on the bus wires etc. All this generates noise on the ground
plane directly related to whats happening on the bus itself. With
S/PDIF the receiver is generating noise on the ground plane directly
related to whats happening on the wire. The PLL is hunting around
following whats going on with the wire also generating ground noise. 

To start dealing with this you HAVE to isolate the grounds between the
input circuits (whatever they are) and the DACs. Unfortunately its
impossible to eliminate the noise on the input side from the signal
crossing the barrier. Just putting in something like an opto-isolator
helps some, but the optical signal generated is the difference between
the ground plane and the signal, thus the optical signal still
containes the ground noise. When the receiver converts that back to an
elcrical signal, current flows in the "DAC side ground" that matches
the noise on the input side ground. If you do it right you can reduce
it, but you can't get rid of it. 

There are also some special ways of laying out a PC board which
essentially "steers" the noise generated by signal induced ground
current away from the most sensitive parts. This can help a lot, but
very few designers really understand all this. Its rather esoteric
stuff. The few people that really DO understand it are getting large
salaries from big telecom companies rather than working for high end
high fi firms. 

The overall upshot of this is that for most systems as you start
getting the "normal suspects" taken care of and your system is running
at quite low jitter levels its actually MORE sensitive to input
changes. The effects caused by whats happening on the input have a
proportionally larger affect on the jitter delivered to the DAC chip. 

Don't confuse input sensitivity here with being "BAD". The jitter
levels at the DAC chip ARE lower, they can just vary over a wider
percentage with input differences. In the example I started out with
the jitter at the DAC might be changing from 40 - 80ps, but with the
low jitter circuits its varying from 2 - 20ps. At its worse its still
lower than the best of the other, but its varying by 10X with input
differences VS 2X for higher jitter system.

I gues I'm trying to say that input sensitivity does NOT mean something
is broken or that it has poor performance. 

If you throw everything at it with multple super low noise power
supplies, isolation from input to DAC, get someone that really knows
how to lay out ground planes for low noise you can get that 2-20 spread
down to say 2-7, but its going to cost a LOT of money. Doing so is not
cheap. 

Its interesting to note that the less optimized solution (ie cheaper)
can still reach 2ps IF the source is optimized for it. So exceptionally
good results CAN be achieved fairly inexpensively IF you are willing to
spend time tweaking and optimizing the system. If you want that level
of performance no matter what, you have to pay for it. 

OK, thats the basics! Next installment goes over some details of
S/PDIF. 

John S.


-- 
JohnSwenson
------------------------------------------------------------------------
JohnSwenson's Profile: http://forums.slimdevices.com/member.php?userid=5974
View this thread: http://forums.slimdevices.com/showthread.php?t=84903

_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/audiophiles

Reply via email to