The Sentinel DAC - Design Update #1

Official - Updates
42

Hi Dustin,
Thanks for chiming in.

A couple of things I don’t understand, regarding your belief that the transport will perform the deemphasis. Which of the following might be untrue?

  1. The CD/SACD transport reads the digital bits from the disc and transmits the raw data stream (via coaxial, TOSLINK, AES/EBU or I2S outputs) in a ‘bit-perfect’ way, without altering the bits

  2. Thus, the transport reads the subcode channel containing the pre-emphasis flag and sends this indicator along with the ‘bit-perfect’ digital audio data to the Sentinel Digital Director

  3. CD/SACD transports lack the analog or digital filtering circuitry required to invert/normalize the boosted high frequencies

Finally, my understanding is that emphasis is ‘coded’ as a flag into the 16/44 bitstream of a 1980s CD. If the CD/SACD transport is delivering the data in a bit-perfect manner to the Sentinel Digital Director, why would the display show the signal as having “24-32 bits”? I’m quite sure that in the 80s there was no ability for the DSP, A/D or D/A circuitry to handle word lengths exceeding 16 bits.

Thanks.

43

The only digital interface with a subcode flag for de-emphasis is S/PDIF. No other digital transmission format does. It is also very uncommon for DACs in the last 15 years to have the ablility to do analog de-emphasis (our DACs have never supported analog de-emphasis). And it is increasingly common for DACs not to support de-emphasis at all. Instead we do digital de-emphasis, when flagged to do so on the S-PDIF stream, as the first step in the digital signal chain. This also changes the 16bit original to 24bits to keep the signal at high fidelity after de-emphasis digital filtering, but the data now can be treated as any other PCM signal. However, the de-emphasis algorithm is extremely easy in the digital domain and does not require significant processing. A very small dedicated bit of logic can do it easily, or even a very low power microcontroller. So it is customary, for a very long time, for the transport itself to do the de-emphasis and never use the de-emphasis flag in the S-PDIF at all. Generally, only very old transports don’t do de-emphasis before sending the digital data out. This is generally built right into the first stage ICs that decodes the data from the CD itself. This preprocessing always results in a “data-expansion” of 24-32bits from the original 16bits that was recorded onto the CD. It is true that the data at this point is technically not “bit-perfect”, but as stated before, very very few DACs support analog de-emphasis now. It is a completely dead format extension, only supported for compatibility as a relic. I would expect a transport with I2S to do de-emphasis digitally because there is no way to signal the DAC that de-emphasis needs to be done. If the transport does not do de-emphasis when using an I2S interface then the de-emphasis can/will never be applied by anything downstream, unless a manual de-emphasis button were available, and the user was able to manually toggle it every time they wanted to play a disc with emphasis(I have never seen this on any equipment, ever).

The emphasis flag on the CD itself is not encoded with the 16 bit audio data. It is encoded as a sideband flag.

1 Like
44

Hi Dustin,

Thanks for the detailed explanation.

Guess the challenge for me is to find which one of my very old 1980s Japanese CDs have pre-emphasis, and whether they sound weirdly thin (deemphasis not applied), or balanced (deemphasis successfully applied)!

1 Like