John Curl's Blowtorch preamplifier part III

Status
Not open for further replies.
It's not mixed signal though? Since ESS doesn't make an ADC currently and didn't at the time of the presentation, they are talking about their Sabre DAC. I am missing where the analog domain enters this at all before the final 6-bit converter or whatever it is in the Sabre.

It may depend on the complexity of the digital device. In theory they are deterministic, but CPUs have gotten so complex it can be impossible due to time constraints to simulate every single sequence that might be subject to error.

In a dac, don't know how much trouble simulation would be in terms of simply exercising all possible bit patterns vs how hard it is to know what digital bit pattern sequence conditions will lead to audible artifacts for some percentage of very perceptive listeners. Even if a manufacturer knew about some issues or potential issues of that nature, it might be unwise to disclose details publicly.
 
Unless there is an ADC inside the DAC feeding back into the modulator, then it is purely digital and it is not the case that it is difficult to simulate. You cannot even remotely begin to compare a DSM with tiny transistor count to a CPU with 4B + transistors. You can implement a DSM on a microcontroller...

It is basically software. They write Verilog or VHDL and synthesize it to an FPGA. You can FULLY and accurately simulate the HDL. This happens every day.

Simulate isn't even the right word. You can fully exercise the math / algorithm separate from the implementation. This is what is done.

Here you go, an open source package in Python.

python-deltasigma
 
Those who don't hear imperfections in dacs argue that there are no imperfections, and any claimed imperfections must therefore be imaginary (if the imperfections are insufficiently stationary to show up in measurements that work well for other types of audio devices).

Wrong. Those who don't hear imperfections in DACs do not claim there are no imperfections in DACs. They only claim that those imperfections are not audible and those claiming they can hear them must provide proof of audibility before identifying the imperfection source. That, if they want to be taken seriously.

Also not sure what an "insufficient stationary imperfection" is (stationary has a certain meaning in physics/engineering which seemingly does not apply here) and why you think there are imperfections that don't and/or won't show in measurement. Perhaps you intended to say that certain imperfections cannot be measured by enthusiast amateurs. Otherwise, I would only remind you that the Deep Space Communication Network is able to analyze and retrieve data from digital signals arriving with a power of 10^-20W.

One question remain mysterious to me: Why the people, in this forum, that reject in audio any form of emotion or sensibility in the way to approach the technology, refuse the right to others to use a different approach, are so arrogant and aggressives, shoot at the player instead of the ball, use authority arguments.
Are-they so unsure of their positions, or is-it their lack of ability to feel and enjoy anything, music included, that frustrate them so much ?

You do realize I could turn your question into an equivalent, about people that reject any educated discussion about technology, and prefer to use emotions or sensibility when dealing with precise engineering facts, right?

Well, one big difference is that (for example) my emotions and sensibility are unknown to you and anybody else, while yours and your peers in "emotions and sesitivity" ability to deal with precise facts is quite obvious, isn't it?
 
Disabled Account
Joined 2012
The mixed signal simulation at this level is VERY difficult, we network workstations together just like they do at LHC and run some sims for days.

some sims run on the fastest computers ever created -- the super computers -- completely ties up the machine so others cannot share it. So, running code over night shift. At their calcs per second speed, running for hours, it is mind boggling and often just to see any small change in the lsb area of differences/changes being looked at.


-RNM
 
Last edited:
My point is chris is talking about the purely digital aspects of the problem, all of your quotes refer to the mixed signal aspects of the issue. When you are simply adding, multiplying, dividing numbers you don't get different answers depending on some poorly understood conditions.

Sure, in digital, no bits are changed & this again the usual mindset - ignoring the system whole. In this case the 2nd order effects from the modulator processes which can generate internal dynamically changing noise in the DAC & its effects appearing on output stages.

It's a pity that we have to listen to the analogue output otherwise the issues on the digital side could simply be ignored.
 
Sure, in digital, no bits are changed & this again the usual mindset - ignoring the system whole. In this case the 2nd order effects from the modulator processes which can generate internal dynamically changing noise in the DAC & its effects appearing on output stages.

It's a pity that we have to listen to the analogue output otherwise the issues on the digital side could simply be ignored.

It’s a pity this has nothing to do with the discussion or what Mallinson was taking about. You should stick to your $1000 power cords.
 
The modulator is a non-linear system - both Philips papers & Ess state that quantizer in SDM is non-linear which makes modelling of this very difficult, if not impossible

The SD modulator uses feedback (or feed-forward) to attempt to correct for the quantization noise generated in this process & shift it out of band.

I'm interested in Chris's understanding of what Mallinson & JohnW stated.
 
Last edited:
Is that really what he has said? ;)
That what Mark says.

No definition of "emotion" in reference to research into audio reproduction?

There’s differences that can’t be measured according to Geddes?

Or he just guesses there’s 5% that ‘thinks’ they hear things that aren’t measureable?
It's an educated guess, he has done a lot of actual testing
 
You mean the discussion according to your definition of what it's about? Explain then what was Mallinson talking about.

Sorry, I do not have any more time for you. There are many resources online including those I have linked where you can learn. You'll need to if you if you want to sell any digital products alongside your $1000 power cables.

If you do not understand why or how an algorithm can be exercised completely, we are done here.
 
Last edited:
Status
Not open for further replies.