| ammitsboel wrote on Tue, 16 May 2006 13:55 |
Hi there,
Why is it that manufactures often put caps in series with the I/O transformers? A part of the reason must be DC, but many circuits doesn't have dc on them. Is it to improve jitter performance?
Best Regards
Henrik
|
First, most of the transmitters and receivers are single ended drive and the DO have a DC offset. The DC offset is half way between the logic high to logic low, so a 0-3V driver has 1.5V DC average. You do not want to have a transformer winding shorting that driver 1.5V DC to ground. Worse, is when the signal "stops", and the driver is at high state - now you have 3V or "real DC" across the grounded winding. Clearly you want to block DC.
The AES and SPDIF signals, as well as many others, have the property of "signal independent DC". The average DC is exactly half way between the low and high voltage level, regardless of the audio in the data. That makes a push pull drive operate very near no DC. But say the signal "stops", then you are shorting an "high output" to a "low output" through a winding... So you want to block the DC.
So it is more then just a saturating transformer issue, it is also protecting the IC, or at least, when short circuit protected) preventing it from being in a "less then desired mode"
There are similar considerations on the other side (the receive end, leading you to some DC blocking).
Fortunately it is easy and cheap. Keep in mind that the signals are very high frequencies (compared to audio) so both the magnetizing inductance and the DC blocking caps are for the MHz range. As a rule, I do not see using electrolitics here.
regards
Dan Lavry
