Quote:
Originally Posted by jsiau /img/forum/go_quote.gif
Steve this is a great question.
There is more to the DAC1 design than meets the eye. The input of the pulse transformer has a balanced controlled-impedance connection back to the input jack where it is grounded to the chassis. It is impossible to see this because these traces are fully surrounded by ground planes. All of the input currents are returned directly to the point at which the coax shield connects to the chassis. This chassis ground is not the analog ground for the internal electronics, and RF currents into this ground point cannot flow through the analog ground plane.
The wiring of the pulse transformer secondary is as important as the primary. The secondary has a balanced connection to the digital audio receiver IC. This provides a well-defined path for the return currents.
Floating the digital audio input will void the authority to operate the DAC1 under FCC and CE regulations (unless the modification has been subjected to full FCC and CE testing). It is also not legal to sell modified units unless they are tested. Do you have FCC and CE test results for modified units?
Floating the coax shield usually increases RF emissions as well as susceptibility to RF interference. The digital inputs on the DAC1 are RF connections and need to be treated as such.
|
Impedance-controlled connection? You mean the wires hanging out in the air soldered to the circuit-board?
I can understand the need to pass FCC emission testing. This does break the Faraday shield. I did some work with this and some very high-frequency CPU's at Intel Corp in the past. It can be difficult. I suppose if I ever had a complaint about the emissions from my gear interfering with some other equipment that someone might lodge a complaint with the FCC. Very unlikely though. Neve had a complaint yet. I take a lot of care to insure that my transmission-lines are properly terminated and that the impedances and current paths are correct, so that the emissions are minimized. This is even better than shielding, grounding and other band-aids that are typically added after the fact. Even though the Farady shield is not perfect, the emissions are still low.
I mod lots of different DAC's. Some of them do have the input winding floating, except for a .1uFd cap to ground on the low-side. Evidently some were able to pass CISPR, CE and FCC emissions testing or maybe they just dont care about emissions. This still does not maintain the Faraday shield, however it could. If the connector were isolated, but surrounded by a ring of small SMT capacitors that each connect to earth ground, then the FS would be intact, without actually DC-grounding the BNC connector. This would be the best of both worlds, galvanic isolation to minimize gorund-loop noise and shielding to minimize emissions.
The problem with emissions testing is that most companies typically use the worst, cheapest digital cable they can find and do what they have to do in order to pass with this cable. All cable connections must be connected and running for the test.
This is an unfair goal IMO. It causes companies to compromise their designs, usually adding jitter or potential ground-loops in order to accomodate a cheap, poor performing cable, or perhaps a bad impedance mismatch etc...
My strategy is to force the customer to use my cable and my terminations. I have control over both ends of the transmission-line, the source and destination. This way I can truly control the reflections on the transmission-line and the current flows. It's the reflections and overshoots that are usually the emissions culprits.
Here is another question for you, since you brought up impedance control and wrote "The digital inputs on the DAC1 are RF connections and need to be treated as such":
Why did you use a 50 ohm characteristic impedance BNC connector for the S/PDIF input, which is specified at 75 ohms?
Do you feel that the edge-rates are slow enough that the absolute impedance is not important?
Steve N.