Thursday, September 20, 2018

DVD-9000 and HDCD

Since I now have an Oppo BDP-205 in my living room system, I only use the Denon DVD-9000 to play HDCD.  The DVD-9000 is a special player for HDCD, I have always thought.  Among other things, it seems to uniquely implement the HDCD peak expansion, so that HDCD's can put out 4dB more than regular CD's.  I can say that authoritatively because I have measured it now on my Lavry.  Plus or minus < 2dB.

Further, I have measured it using digital input.  This means I can pipe in SPDIF from my server, as I now do through the "Living Room 2" Sonos Connect, and have the DVD-9000 output HDCD in analog, then resample back to 24/96 digital on my Lavry for playback through my DSP's.

I was worried because there is no HDCD indicator which lights up when you use the DVD-9000 as a DAC.  And the DVD-9000 brochure I have on my computer says only that you can play HDCD discs.  When you use the DVD-9000 as a DAC, it puts the sample rate in big letters on top of where the HDCD indicator would otherwise have been.

I have long thought about doing a similar thing with MQA, though another friend sent me Archimago's takedown of MQA and it looks pretty convincing.  But I think I might add inexpensive MQA decoding capability to my system anyway, though I am intrigued about the "unfolding" thing which might be an option.

Interestingly, the DVD-9000 does not seem to contain a PMI chip, like the PMI 100, to decode HDCD as the Denon DVD-5000 did.  Instead, the Denon has an analog devices chip as "audio decoder" which apparently implements a number of codecs including HDCD and possibly Denon's AL24 now in 32 bit.  Because this processor is a 32 bit chip instead of 16, it may well implement HDCD to the same standard as the unobtanium PMI 200.  But the same thing may be true of virtually all later HDCD players, including the Oppos, that switched to using licensed HDCD implementations on their digital filter chips, rather than PMI chips.

Famously HDCD has no digital unfolding because the core of HDCD is actually changing the digital filter.  What this does is not fully expressable even within a 24 bit stream.  It can exchange bit accuracy for timing accuracy when the program material or producer deems it useful; it's giving them the power to switch the digital filter mode on-the-fly.  Meanwhile, the newer MQA only has one fixed digital filter, and an extremely leaky one to high frequency images which I have usually called aliases.

But when resampling the analog at 24/96, much of that added timing accuracy is captured by the faster sampling rate.

I have also decided to change my "standard" levels so the Lavry is made 2dB less sensitive.  This means I do not have to attenuate my Oppo through the Emotiva XSP-1 before sampling on the Lavry.  It seems to me that keeping things as close to 0dB as possible on the Emotiva is best (though I had previously worried about distortion or even clipping on the XSP-1, I have been told it can produce up to 12V output, though that is not in the specs).

So the new "standard" levels are:

Level setting on Lavry: +8dB (2dB aove 3.88V balanced input)
XSP-1 gain for Oppo BDP-205: +0.5dB
XSP-1 gain for Sony 9000ES: 0dB
XSP-1 gain for HDCD  on Denon DVD-9000: -4dB
XSP-1 gain for phono: 2-6dB
Stealth DC level for DAC for Hafler 9300: -3.75 dB (this matches Krell at -7dB)

Update: this is trickier than you might presuppose, because of sample-overs.
The DVD-Audio of Two Against Nature has shown it necessary to lower the Oppo input level on the Emotiva to -2.5dB if I have this correctly, showing sample-overs of about 3dB therefore.  That may be a decent number for the 9000ES also.  I doubt it would change the Denon DVD-9000 which reserves the highest levels, up to 3.5V, for HDCD extra dynamic range, and I've tested a range of discs for that.




No comments:

Post a Comment