Rob Watts said:

- of course time is not accurate,

Click to expand...

After so much talk of timing accuracy in your DACs, I'm surprised to hear you say that, Rob ...tssk tssk

@Rob Watts
I have discovered and fully embraced parametric equalization within JRiver and using simple spectrum analysis tools - have an absolutely flat response on my speakers and more low end. Its quite spectacular ...provided all my content goes through JRiver. Its a shame that my SPDIF feeds don't benefit.

Rob, any chance of including a parametric equalizer at the tail end of your DAVE(-2) digital section? Even supporting a low/high shelf or two would be very helpful to then allow all inputs to the DAC properly equalized. While you are at it, just incorporate a microphone input (ala Davina) and build in an automatic room/speaker optimization tool: Perfect analog reconstruction + perfect playback tuning.

...when you have time
Dan

It's a shame they have stopped making more time!

dmance said:

@Rob Watts
I have discovered and fully embraced parametric equalization within JRiver and using simple spectrum analysis tools - have an absolutely flat response on my speakers and more low end. Its quite spectacular ...provided all my content goes through JRiver. Its a shame that my SPDIF feeds don't benefit.

Rob, any chance of including a parametric equalizer at the tail end of your DAVE(-2) digital section? Even supporting a low/high shelf or two would be very helpful to then allow all inputs to the DAC properly equalized. While you are at it, just incorporate a microphone input (ala Davina) and build in an automatic room/speaker optimization tool: Perfect analog reconstruction + perfect playback tuning.

...when you have time
Dan

Click to expand...

Wouldn't that make the DAC more complex? Those tools belong in the source, I'd think. Or better yet, in the new digital amps that Chord are working on.

Actually, EQ is something I plan to investigate post digital power amps - if I can get some subjective benefit from it. I have designed bi-quad DSP's before.

I think it would be more interesting to have active speakers where each drive unit has a dedicated power pulse array driving it. Each, of course, would have band-limiting filtering for the driver it's powering. Then, "room correction" is a secondary function of these digital amplifiers and there's no stupid passive crossover sucking up ridiculous amounts of power.

Since you're expecting to be able to get 10s of watts of power from a power pulse array in the first iteration and since there's no passive crossover, the resulting speaker should be quite something.

Now, to find a speaker that's amenable? ATC?

Now playing: Stars of the Lid - Dust Breeding (1.316) +

dmance said:

Its a shame that my SPDIF feeds don't benefit.

Click to expand...

My Blu-Ray/cable source is S/PDIF or Toslink. I too require DSP to tune out some subtle room bass peaks. Unfortunately, the best I can do is to get miniDSP nanoDigi and program it for the DSP but it inevitably converts 44/48kHz to 96kHz first which obviously means I'm losing timing accuracy and the wonders of all the extra taps in DAVE. Also to achieve maximum galvanic isolation between the Blu-Ray player and DAVE, I use toslink to feed the minidsp and use Uptone LPS-1 to power the minidsp. Definitely a suboptimal solution until Mr Watts decide to tackle EQ.

Has anyone experienced the benefits of external desktop servers playing bit perfect core. It has been extremely beneficial to my listening experience, those bits seem to matter going into the dac. Rob what is going on here, your TT never sounded so transparent.

Rob Watts said:

Actually, EQ is something I plan to investigate post digital power amps - if I can get some subjective benefit from it. I have designed bi-quad DSP's before.

Click to expand...

The real prize, imo, would be to implement digital filtering to support fully active loudspeakers - one channel of DAC and power amp for each drive unit. You place such stress on the importance of timing, yet a conventional passive or active crossover is a disaster when it comes to preserving the timing relationships between the drive units in a speaker, and drive units themselves vary in phase. A complete digital active solution would allow a completely linear phase loudspeaker. There would surely be a subjective benefit!

Yes I do place a huge emphasis on timing. But it's very important to be clear that I am not worried about non-linear phase in the traditional fashion. This is defined when the phase response changes with frequency, and this happens naturally anyway; and I have no evidence that this type of behavior is important subjectively (unless it's gross in which case FR would be affected too).

My worries with timing is something very different, and the non-linearity that really has huge subjective impacts are when the timing changes with signal amplitude, or due to sampling. In this case I am worried when the timing of transients is different in a random or any systematic fashion; the key is that the timing is constantly changing. This the ear/brain is highly sensitive too this, and can perceive minute changes in timing. This is completely different to the conventional definition of non-linear phase against frequency.

That said, passive crossovers are both phase non-linear, and amplitude phase non-linear. For sure, amplitude phase non-linearity is highly audible - this occurs for example with inductors, whose inductance value changes with current, hence the time delay is dependent upon signal amplitude. The ear/brain is highly sensitive to this, but I don't believe that phase shift changing with frequency is actually that important. So for sure replacing passive components with digital direct drive is on my road-map, and I think it will give huge benefits - if done correctly (sound-stage depth probably being the biggest).

Rob,
Regarding EQ performed at the source prior to your WTA vs an implementation in your DAC's digital pipeline ...does EQ alter the behavior and/or performance of your WTA filters? Meaning that the changes in amplitude/steepnes of an equalized signal somehow affects your timing reconstruction?
If that's the case then it certainly makes sense to EQ within the DAC!
Dan

I don't think timing reconstruction is affected by EQ; so long as the sample rate is unchanged. But I could be wrong about that, as this is based upon theory not listening tests. You do get better accuracy if you process at higher sample rates for EQ at around 20 kHz for SR of 44.1 kHz.

But there is another issue, and that's re-quantization. When you use DSP, the output has a bigger bit depth; but to get to the next stage, you need to truncate the bit depth. It's the truncation where the difficulties lie. When you truncate a 16FS signal using aggressive noise shaping, you can maintain the original bit depth performance in terms of SQ and measurements - let's say it is 64 bits being converted back to 24 bits. I can do this with -400 dB THD and noise in the audio bandwidth, hence perfectly maintaining the 64 bit accuracy. And for sure, this sounds much better (better depth perception) than doing the truncation with conventional triangular dither. Pseudo Gaussian dither gets you closer to the sound of noise shaped dither (it's about half way there) and actually this is a problem, as you can't employ aggressive noise shaping for anything lower than 16FS (705 kHz). It's a problem for Davina, when I need to be able to have good depth with 16 bit 44.1 kHz. I am hoping that a better psuedo Gaussian dither will get much closer to noise shaped. My current Gaussian uses 14 sources; I plan to increase this to 56, getting you much closer to ideal Gaussian.

So for these two reasons (HF accuracy and re-quantization) I always EQ and volume at at least 16FS, and any future EQ would probably be done at these rates.

Rob,
Thanks for all your insightful comments. So, if I was to ask for a single new product (as an amalgam of input from forum posts) it would be:

- a digital source companion to Dave
- mscaler at # of taps required for full analog waveform restoration
- 100% isolation of the digital coax link to Dave ...to avoid the current cabling debate
- USB input plus all the wireless/streaming interfaces present in Poly/2Go
- Equalizer (manual) or with feature to auto tune amp/speakers/room with microphone input. Perfect playback of perfectly reconstructed analog.
- Related firmware update to Dave that repurposes unused WTA fpga gates to further refine the SQ as you deem possible.

Pls relay my wish to Mr Franks.
...there, Thanks!
Dan

Jawed said:

I think it would be more interesting to have active speakers where each drive unit has a dedicated power pulse array driving it. Each, of course, would have band-limiting filtering for the driver it's powering. Then, "room correction" is a secondary function of these digital amplifiers and there's no stupid passive crossover sucking up ridiculous amounts of power.

Since you're expecting to be able to get 10s of watts of power from a power pulse array in the first iteration and since there's no passive crossover, the resulting speaker should be quite something.

Now, to find a speaker that's amenable? ATC?

Now playing: Stars of the Lid - Dust Breeding (1.316) +

Click to expand...

Now that would be a fantastic product, a pair of ATC SCM50T with pulse array and digital crossover, or a pair of SCM20 + some "standard-amp" subwoofers.

Where do I preorder?

Rob,quick question....as far as i understand the pairing of the Dave with the Blu 2 will decrease by about 3DB the output to HP's....as it is already a chore for the Dave to drive the LCD-4 or Abyss should i assume an external amp will now be warranted...i am waiting on my blu 2 and want to be properly prepared!Thanks