AES/EBU SPDIFF and Optical interface converter

[Martin Czech]

On the Crystal homepage I found 8401.pdf datasheet CS8401A and the
an134.pdf app note. Combining this with the links that were just
posted one can say (please tell, if I'm wrong):

The design of such converters is straight forward, no magic.
The parts can be hard to get, however.

Optical to SPDIFF:
You need an optical input connector and the infrared input diode or
transistor 660nm (this can be directly build into the connector). Some
kind of amplifier for the  small opto device signal, this can be
a CMOS inverter/schmitt trigger. Another inverter (or a couple of
inverters) serve as spdiff driver. The crystal data sheet shows, that
the inverters should get a 374 Ohm series resistor, and then a 90.9
Ohm parallel resistor to ground. This provides the required 75 Ohm
Impedance for SPDIFF. AS well as the correct level. And then comes a
special transformer, for ground and hf decoupling.

The cheap version without transformer would use only a 75 Ohm series
resistor.  (The output impedance of the inverter is assumed to be 0,
if this assumption is too erroneous, the impedance matching resitors
have to be changed. I think this could be done with a dc measurement.)
Not using a tranformer means ground loops and even worse high frequency
currents. No pleasant idea.

It is required, that the electrical output signal is 0.5V peak-to-peak
+- 20% when the output is directly loaded with a 75 Ohm load (no cable,
just a resistor).

SPDIFF uses unbalanced cable with 75 Ohm (I don't care, for short distances
I use some sort of normal audio cable) and RCA Phone connectors.




 
SPDIFF to Optical :
You need some logic inverters and an optical connector and an IR LED
(660nm).  Just a normal diode driver cicuit with some series resistance
to limit the diode current. The electrical input signal is decoupled
with a cap and a 75 Ohm resistor (ie. a highpass structure), then comes
an inverter/schmitt-trigger to boost the signal from 0.5V peak-to-peak
to normal logic levels.



SPDIFF to AES/EBU and vice versa:
The only difference between cheap SPDIFF and expensive AES/EBU is that
AES/EBU is balanced via transformers, and has 110 Ohm termination.  So,
the digital signal is buffered (some inverters of a hex pack in parallel),
then comes the 110 Ohm series resistor and then the special transformer.
The level should be 2-7 V peak-to-peak , if the output is loaded with
a 110 Ohm resistor (no cable). AES/EBU uses XLR connectors with pin 1
to (chassis ?) ground. Unfortuately I don't know the polarity of the
other pins, but there a 50% chance that I do it right ;->. Otherwise
swap. On the input side there's again an XLR, then the transformer directly
connected to the 110 Ohm resistive load. Since signal levels are so large,
it should be no problem to use a inverter/schmit-trigger to get normal
logic levels. This seems all you need! It's a shame: some companys
charge the user with 1000 DM for a SPDIFF to AES/EBU interface!

Now the crucial point, the transformers:

The 25kHz to 75 kHz audio rate means 1.5 Mhz to 7 Mhz signal rate on
SPDIFF/AES/EBU. The transformers provide isolation from ground loops/
50/60 Hz hum, and common mode noise. Another request is low speakthrough
capacitance of the transformer coils, because we don't want to see hf
energy as common mode signal on the reciever side ground.

Crystal recommends the following products:

NO GUARANTEE FOR TYPOS!

Pulse Engineering
Telecom Products Group
7250 Convoy Ct.
San Diego, CA 92111
619 268 2400
Part number: PE65612

Schott Corporation
1000 Parkers Lane Rd.
Wayzata, MN 55391
612 475 1173
Fax 612 475 1786
Part Numbers: 
67125450 compatible with pulse
67128990 lower cost
67129000 smd
67129000 single shield

Scientific Conversions Inc.
42 Truman Drive
Novato, CA 94947
415 8922323
Part numbers:
SC916-01 single shield
SC916-02 smd


And now, I can see how to get these parts here in Germany!

m.c.