I disagree that measures required for EMI compliance are generally bad for sound quality. As an example: in early builds of SB2, it was discovered that certain headphone cables made good 2.4GHz antennas, and would pick up the intentional emissions from the Wifi radio and feed them back into the device. The headphone amplifier chip, as is often characteristic of bipolar amps, acted as a decent radio receiver, demodulating the RF energy into a small audible click that was then sent back out to the headphones. A click would be heard as each packet was transmitted - those being the outgoing ACKs as each packet of the TCP stream is received.
Even though our ears don't go to 2.4GHz, this was an easily audible phenomenon because the "envelope" of the signal was at just a few Hz. I'm not even talking about stuff you need lab equipment to observe. The problem was easily addressed by adding some 10pF caps across the headphone output, but I still find it a compelling reminder of how important EMI control is, even in the audio band.
As for s/pdif: I really enjoyed those TDR plots you posted on the other forum. While the ferrites do make it look terrible on the TDR, if you do jitter measurements at the receiving end, you will find their impact is insignificant even though they yield a meaningful decrease in EMI. The reason is that even though the corners of the waveform are rounded off, and there is clearly a HF impedance mismatch as shown on the TDR, the timing of the zero crossings, and the dV/dT at those crossings, is barely compromised. I'll trade a couple ps of jitter (out of 100+ by the time the s/pdif receiver is done with it) for a few dB of reduced EMI, regardless of regulatory requirements. Certainly such filtering would be unacceptable for, say, analog video, but for s/pdif we care about compatibility, data integrity, and timing, not necessarily the entire waveshape (right?).
Now, of course if I were to put some cap or ferrite on an output and find that it sounded or measured poorly, I would look for other solutions. And of course, ferrites are just one of many tools and techniques used to control EMI. But I wouldn't discount them as being something that is only needed for high volume products - in some situations they can be critical for sound quality too.
Results 21 to 23 of 23
Thread: SB3 vs Duet output noise
2008-10-19, 22:14 #21
2008-10-24, 12:48 #22
I guess I need to elaborate more on this. We are probably in agreement more than not. However, having worked with RF for 40 years or so, and having used tons of ferrites, the very last place I would stick one is directly in the audio chain.
But, as I have said numerous times, your products are priced such that the ridiculous measures than many of my colleagues (and I) go to are impractical. And, to be honest, one may not hear all that much of a difference without the bead. (I know guys who swear that they can hear one on the AC cord, and refuse to use them there. Since they make more $ than either of us selling gear, they may be on to something. But then, the reason they make more is their stuff sells for 10s of kilobucks. It better sound good.)
Still.......one could use a stagger-tuned 2-pole RC filter instead. I know that thought has crossed your mind!
More on that later........
2008-11-09, 23:31 #23
Here is one for all of you DIYers to ponder.
Take a look at the data sheet for the WM8501. Take note of the bypass arrangement on the Vref pin.
Now do the same for a typical Burr-Brown DAC. Especially the older R-2R ladder types. Think about what that pin does, and how it can be improved upon.
While I am at it.........he says gingerly.......you will notice that they recommend using COG caps for the post filter. I'll shut up now, before I invoke someone's wrath.