[Rob Watts]

The code for DSD for the M scaler and TT2 is identical. Unfortunately, it is impossible to repair the damage that DSD does; damage to resolution of transients and small signals can't be restored. It's not like PCM, where the original bandwidth limited sampled signal theoretically can be completely restored by the use of a sinc function. Transients corrupted by DSD noise shaping can't be restored. What we can do, is remove the HF distortion and noise that DSD creates, and this is done very efficiently with the DSD filters built into TT2 or the M scaler. By removing this noise we get much better conversion to analogue - the analogue and the original digital DSD64 files measure the same.

 

[iDesign]

"...must have innately similar performance to Dave, and I wasn't happy with the first attempt."

For different reasons, you're probably not the first or last DAC designer to say this. The DAVE is mighty hard to beat and I have yet to find similar performance anywhere.

 

[flyte3333]

The code for DSD for the M scaler and TT2 is identical. Unfortunately, it is impossible to repair the damage that DSD does; damage to resolution of transients and small signals can't be restored. It's not like PCM, where the original bandwidth limited sampled signal theoretically can be completely restored by the use of a sinc function. Transients corrupted by DSD noise shaping can't be restored. What we can do, is remove the HF distortion and noise that DSD creates, and this is done very efficiently with the DSD filters built into TT2 or the M scaler. By removing this noise we get much better conversion to analogue - the analogue and the original digital DSD64 files measure the same.

Do all your comments about issues with DSD over the years only apply to DSD64?

Or also for DSD256? Both recordings available for purchase (like at nativeDSD.com) and PCM-to-DSD256 upsampling to a DSD DAC...

 

[Rob Watts]

In a short answer - no - all speeds have SQ problems. When you listen from DSD64 to DSD256, then soundstage depth and impact improves, but it's nothing like as good as PCM via WTA. The other issue, which is not speed related, is the modulator changing it's behaviour due to input signal amplitude, as the noise shaper is artificially constrained with only two levels. This gives noise floor modulation, and the only way of eliminating this issue is via n bit modulation - the minimum needed is 3 bits to guarantee zero noise floor modulation from the modulator.

 

[feddar]

What about DXD?

 

[Rob Watts]

DXD is fine as it's regular PCM at 352.8kHz. This is assuming that the original ADC is n bit and not 1 bit of course!

 

[Mediahound]

Speaking of DSD, I use Roon. Which is the best DSD setting with the M Scaler for DSD?

Convert to PCM
DSD over PCM v1.0 (DoP)
or
Initial dCS method

?

 

[hmartin]

Speaking of DSD, I use Roon. Which is the best DSD setting with the M Scaler for DSD?

Convert to PCM
DSD over PCM v1.0 (DoP)
or
Initial dCS method

?

You should use DoP with the mscaler. Did some tests comparing Roon coverting to PCM, not even close. Robs method sounds much better. But you can just try it and see what you think.

 

[antastik]

@Rob Watts
Before posting here, I went through this entire thread, learning a lot. I think this is the right place to ask Rob about something that always puzzles me. Regarding measurements that are usually performed on audio equipment, the frequency response plot is ubiquitous. It seems to me though that such a plot completely misses transient and timing. Why is it that nobody takes a measurement of amplitude and frequency OVER TIME? The waterfall 3D diagram would allow to compare for transient response of real audio signals. It would also allow to better understand the role played by cables, finally archiving the traditional opinional nonsense that envelopes this subject. Is there a reason?
Thank you Mr Watts for the various insights. Next time, if I may, I will ask for more details on noise floor modulation due to RF noise.

 

[thePhones]

Before posting here, I went through this entire thread, learning a lot. I think this is the right place to ask Rob about something that always puzzles me. Regarding measurements that are usually performed on audio equipment, the frequency response plot is ubiquitous. It seems to me though that such a plot completely misses transient and timing. Why is it that nobody takes a measurement of amplitude and frequency OVER TIME? The waterfall 3D diagram would allow to compare for transient response of real audio signals. It would also allow to better understand the role played by cables, finally archiving the traditional opinional nonsense that envelopes this subject. Is there a reason?
Thank you Mr Watts for the various insights. Next time, if I may, I will ask for more details on noise floor modulation due to RF noise.

This might also be the main reason why headphones which have almost identical frequency responses (or are eq’d accordingly) sound totally different. I would say that my Utopia has very similar responce (after eq) to the Hifiman Arya but the character is something else. The Utopias attack is way more aggressive (and fun if I’m in the mood) while the Arya is softer and calm sounding. My preference is more on the analytical/revealing side of things but I like to have the options.
But like you said, it also really bothers me that the focus is often just on the frequency responce side (which you can change by eq) but not on the dynamics and speed of attack and decay (which you have very little control over).

 

[Kentajalli]

Before posting here, I went through this entire thread, learning a lot. I think this is the right place to ask Rob about something that always puzzles me. Regarding measurements that are usually performed on audio equipment, the frequency response plot is ubiquitous. It seems to me though that such a plot completely misses transient and timing. Why is it that nobody takes a measurement of amplitude and frequency OVER TIME? The waterfall 3D diagram would allow to compare for transient response of real audio signals. It would also allow to better understand the role played by cables, finally archiving the traditional opinional nonsense that envelopes this subject. Is there a reason?
Thank you Mr Watts for the various insights. Next time, if I may, I will ask for more details on noise floor modulation due to RF noise.

I maybe wrong, but:
- the pulse response and square wave response , (both illegal signals) give indication of phase, timing and ringing within an electronic audio device.
- the 3D waterfall gives indication of decay, usually on a sound transducer such headphones or speaker systems.
Lab measurements are just design and testing tools.
judging an audio device purely on lab results is not wise!
this is not to say that equipments need not measure well to sound good, yes they do, but not everything that measures well automatically will sound good.
but those that measure poorly , will sound bad.

 

[flyte3333]

This might also be the main reason why headphones which have almost identical frequency responses (or are eq’d accordingly) sound totally different.

Yes that's because they will likely vary in distortion and group delay plots (and probably others too).

Lesson here is one plot rarely gives the full picture...

 

[JohnA1]

@Rob Watts
Dave-related question: Got a pair of Neutrik '3 pole XLR female to RCA' adapters
Says wired according to IEC 268-12
Pin2: signal, Pins 1 and 3: connected to ground

Is it OK to connect them to another power amp while the normal RCAs are connected to the normal power amp?
(I'm experimenting with a second pair of speakers running a variation of the main signal simultaneously)

Wouldn't want to short out anything!

Cheers

 

[Rob Watts]

Ouch! Can't connect 1 and 3 to ground, as that will short the negative output. You need need a cable that just uses pins 1 (gnd) and 2 (pos OP) for connecting to a phono.

 

[Kentajalli]

@Rob Watts
Dave-related question: Got a pair of Neutrik '3 pole XLR female to RCA' adapters
Says wired according to IEC 268-12
Pin2: signal, Pins 1 and 3: connected to ground

Is it OK to connect them to another power amp while the normal RCAs are connected to the normal power amp?
(I'm experimenting with a second pair of speakers running a variation of the main signal simultaneously)

Wouldn't want to short out anything!

Cheers

This cable is for a balanced INPUT converted to RCA i.e. balanced input of an amplifier to receive a single ended sgnal.
and NOT balanced OUTPUT converted to RCA i.e. balanced output of a preamp or DAC to single ended input of an amplifier.
If it makes any sense!