[SI-LIST] : D/A interfacing

About this list Date view Thread view Subject view Author view

From: Larry Miller (ldmiller@rhapsodynetworks.com)
Date: Fri Apr 20 2001 - 07:06:40 PDT


In the recent exchange with Dr Eric Daniel the topic of high bit count A/D
converters came up.

The interface between analog circuits and digital circuits seems to have
always been a troublesome area, even back to the days of converters made out
of discrete parts (resistors, caps, etc.).

In most systems through the 80's the "standard" parallel-interface DAC and
ADC range was on the order of 10 to 14 bits. This was governed as much as
anything by the limits of the converters' internal noise ("good" converters
being hybrid thick film devices and rather poor by comparison integrated
mixed signal chips). By and large, with 14 bits of converter resolution most
people were lucky to realize 10 to 12 bits of actual usable range, which was
usually enough for control systems.

***Of course, the SI reflector did not exist and even the notions of SI that
we discuss and use were in a very rudimentary stage of evolution!***

The first 16-bit applications I heard of were for digital audio, and a whole
new class of devices and noise control techniques evolved for this
application. Later on, much larger ranges (up to 24 bits) were found
necessary for CT (Computer Tomography). Companies like Analog Devices and
Linear Technology came up with very inexpensive sigma-delta converters
(which also won out in the audio CD arena).

None of these high bit-count converters operated at very high speeds (i.e.,
not much above audio signal processing). That made use of opto-isolators
between the analog and digital areas of the system feasible, and I believe
that this is the standard method used in CD players. (It also works great in
motor control systems!)

Video "flash" converters typically had much lower bit counts. 8 bits is
better than most CRTs (black & white OR color) can display. Indeed, many PC
video DACs have much fewer than 256 output levels; this is easily shown by
playing with the color toy in Windows or whatever system you are running. 40
dB down on analog noise is not that hard to come by (though, truth to tell,
many PCs don't make it) and is OK for consumer video, but it is not good
enough for your sound card, as your ears tell you. You may set the sound
card for 16 bits but you do not get 16 bits of dynamic range by a long shot.

Now, with digital video (and whatever Eric is working on) raising the bar on
performance AND speed, the SI problems continue to increase.

Most DAC/ADC manuals give some rules for the interface between the analog
and digital systems ("connect grounds only at one place", yada yada yada
yada), but these often do not work. Also, if you have multiple systems
distributed over space in a system, what do you do? "Connect together at
only one place" is not on the menu.

Some years ago I ran across a novel approach described in a paper
attractively titled, "Method for Noiseless Digital/Analog Interfacing in
Complex Systems Using An Intermediate Ground Plane" and a Google search
showed that it is on the Web.

http://www.hybricon.com/techfocus/HTF102.PDF

I never tried this scheme. Has anyone used this or something similar?
Cellular phones use something like this to achieve the 120+ dB isolation
required between the transmitter and adjacent receiver.

I would be curious to hear what has worked for people. I have seen questions
about this from time to time on this reflector, but very little follow-on.
(Maybe people who find the answer also find a new career as a high-priced
consultant!)

Inquiring minds want to know,

Larry Miller

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Jun 21 2001 - 10:11:41 PDT