Found this AES paper from Wolfson's website - http://www.wolfsonmicro.com/uploads/documents/en/A high performance SPDIF receiver_Oct 2006.pdf
But I couldn't find any of their spdif receiver products that have this architecture ? Did wolfson ever start production of this kind of receiver ?
Does anyone else make an spdif receiver that works on the same principal ?
But I couldn't find any of their spdif receiver products that have this architecture ? Did wolfson ever start production of this kind of receiver ?
Does anyone else make an spdif receiver that works on the same principal ?
Call me a naysayer but I think all these schemes are flawed. This one appears to have a small buffer and then they continually speed up or slow down the data out clock to stop the buffer over/under-flowing. Not particularly original.
By keeping the output clock fixed for short periods of time and occasionally making incremental changes, the instantaneous jitter is reduced by an order of magnitude - Bravo! But there are still going to be small temporal distortions over the time frame of a song/album/movie.
I'm still not convinced that the ASRC method which is currently in vogue is any better. Here, the incoming data, which we assume has been recorded with a jitter free clock is arriving at a non-constant data rate. The idea of upsampling to a massively oversampled clock and then phase interpolating the samples to the output clock points in time is equally disruptive.
Ideally, you'd just output those precisely recorded samples at a constant rate.
I still think the best way is to have a HUGE buffer, I'm talking like a GB or so and just play everything back at 0.1% slow to a rock solid clock. OK by the end of the movie you'd have accumulated a couple hundred meg of buffered data, the movie would go on for a few minutes after the DVD stopped spinning. No biggie, small price for perfect playback.
By keeping the output clock fixed for short periods of time and occasionally making incremental changes, the instantaneous jitter is reduced by an order of magnitude - Bravo! But there are still going to be small temporal distortions over the time frame of a song/album/movie.
I'm still not convinced that the ASRC method which is currently in vogue is any better. Here, the incoming data, which we assume has been recorded with a jitter free clock is arriving at a non-constant data rate. The idea of upsampling to a massively oversampled clock and then phase interpolating the samples to the output clock points in time is equally disruptive.
Ideally, you'd just output those precisely recorded samples at a constant rate.
I still think the best way is to have a HUGE buffer, I'm talking like a GB or so and just play everything back at 0.1% slow to a rock solid clock. OK by the end of the movie you'd have accumulated a couple hundred meg of buffered data, the movie would go on for a few minutes after the DVD stopped spinning. No biggie, small price for perfect playback.
For what it's worth, the eye diagrams in the AES paper you link to are reproduced in this white paper analyzing the performance of the WM8804/5:
http://www.wolfsonmicro.com/uploads/documents/en/SPDIF_Paper_v1.2 May 2007.pdf
So it's probable that the WM8804/5 do include "elastic buffers" even though this isn't specifically mentioned in their datasheet.
http://www.wolfsonmicro.com/uploads/documents/en/SPDIF_Paper_v1.2 May 2007.pdf
So it's probable that the WM8804/5 do include "elastic buffers" even though this isn't specifically mentioned in their datasheet.
The only other unofficial reference I found to this is - http://www.electronicsweekly.com/Ar...wolfson-has-jitter-scrubbing-architecture.htm
It does say that "The firm’s WM8804 and WM8805 chips include the new scheme"
It does say that "The firm’s WM8804 and WM8805 chips include the new scheme"
this becomes interesting - contacted Wolfson directly, I was told their current spdif receivers (wm8804/5) do NOT have the elastic buffer technology. I wonder how in the world then are they able to acheive that feat ?? As shown in a chart (pg 5, fig 2) jitter rejection in the wm8805 starts as early as 100hz and increases dramatically by 1-2Khz.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.