Trinity DAC

[esldude]

I've seen this video quoted a number of times, and there have been plenty of threads in the past on it (which I only purused)... Before we get carried away, I think it's best to address the statement that "there is nothing inherently wrong with the resolution of the 16/44 format". My response is, THERE ABSOLUTELY IS, and Monty's video actually indirectly points out the issues. Here they are:

1. Did you notice toward the end that the square wave input is not a square wave anymore at the output, after digital processing and reconstruction? He clearly says why: the signal is band-limited, thus the summation of all harmonics is "incomplete", causing the ripple effect he shows. This is a MAJOR inherent issue with digital, and consequently with RBCD, and higher sampling rates will reduce the band-limiting effect.
2. When it comes to sineware processing, he correctly claims and demostrates that the difference between higher and lower number of bits is noise; he then discusses dither and noise shaping (which pushes noise up to higher frequencies). But what he does not say and show is what happens to noise with higher-than-16 bits, and naturally you would expect it to be even lower.
3. He correctly states that lower noise results in wider dynamic range (letting you get closer to the theoretical limit of ~96dB for RBCD, not necessarily cross it); naturally, you would expect that even higher-than-16 bit resolution will result in even higher dynamic range - you might argue we don't need more, but I'd say this is yet another limitation of RBCD. [BTW, related to noise attenutation and dynamic headroom gain, under Steve's system thread, there is a discussion of how could he and Marty perceive louder overall sound with the Shunyata Typhons plugged in - and both Shunyata and I said basically the same thing: lower noise will result in higher dynamic headroom].

I like Monty's video, but as he says at the end, he has carefully chosen his demos - he's no fool. But let's not draw the wrong conclusions from his video.

But your ears are bandwidth limited. You can reproduce a few more harmonics, but your ears won't hear the extra harmonics.

And wider dynamic range, with peaks or signals out of the noise floor wouldn't sound louder. It has less average level and if the difference is enough to notice it sounds quieter. What many think of as wider dynamic range, is a more dynamic punchier sound. It actually is the reverse. That idea is where the loudness wars came from though they got out of hand and compress things so much all life is gone.

 

[Julf]

Did you notice toward the end that the square wave input is not a square wave anymore at the output, after digital processing and reconstruction? He clearly says why: the signal is band-limited, thus the summation of all harmonics is "incomplete", causing the ripple effect he shows. This is a MAJOR inherent issue with digital, and consequently with RBCD, and higher sampling rates will reduce the band-limiting effect.

Sure - but the difference only matters if you like to look at square waves on a scope. What I always suggest as an eye-opener is to use a maths simulation package such as mathlab, octave or mathematica to generate the square wave by adding a series of sine waves, but with one of two of the lower harmonics phase shifted by 90 or 180 degrees. The result doesn't look anything at all like a square wave - but it will sound exactly the same to your ear.

When it comes to sineware processing, he correctly claims and demostrates that the difference between higher and lower number of bits is noise; he then discusses dither and noise shaping (which pushes noise up to higher frequencies). But what he does not say and show is what happens to noise with higher-than-16 bits, and naturally you would expect it to be even lower.

Sure, but there is no "even more inaudible". If something isn't audible, it isn't.

He correctly states that lower noise results in wider dynamic range (letting you get closer to the theoretical limit of ~96dB for RBCD, not necessarily cross it); naturally, you would expect that even higher-than-16 bit resolution will result in even higher dynamic range - you might argue we don't need more, but I'd say this is yet another limitation of RBCD.

And where do you get source material with more than 96 dB dynamic range?

 

[Al M.]

And where do you get source material with more than 96 dB dynamic range?

Precisely. You could endlessly continue to play the game, like in re-writing the statement into a hypothetical discussion of 24 bit audio:

"He correctly states that lower noise results in wider dynamic range (letting you get closer to the theoretical limit of ~144dB for 24 bit audio, not necessarily cross it); naturally, you would expect that even higher-than-24 bit resolution will result in even higher dynamic range - you might argue we don't need more, but I'd say this is yet another limitation of 24 bit audio."

So what's the point, exactly?

 

[LL21]

Hi,

We received our unit of the Trinity DAC.
I read all of this thread and must give our tribute to Dietmar for designing this product.
Also want to say that the words and praise that some users described here are in line with our first listening tests done in our studio.

The equipment that we used to listen to the dac is in the foto below (sorry for the quality but was taken by mobile phone).
View attachment 13630

Warmest Regards
Miguel

Beautiful system!!! How do these particular speakers compare with TAD Ref One and Altairs? If you prefer to PM, please do so. I am aware these are not their flagships, but nevertheless, most curious. Thanks for any advice.

 

[Julf]

So what's the point, exactly?

Right. Pretty much what I wrote, "There is no "even more inaudible". If something isn't audible, it isn't.". A signal-to-noise ratio (or dynamic range) way beyond the signal-to-noise ratio of your source material (or your listening room) is pretty much pointless. What is the background noise level of your listening room?

 

[TRINITY]

Hi all,
somebody triggered me to watch the mentioned Video, unfortunately I have not the time to view the whole Video so I jumped to 17:50 where the "wizard" starts to explore the ringing in the digital System. At least that Point is wrong explained. The fact that he could play a 20kHz sine wave in the beginning let us assume that his demo Setup is based on a regular digital Signal path with digital and analog filter. Of course we all know the Fourier series and its effect on the rise time, but his conclusion is wrong. The visible ringing Comes from the digital filter and is not caused by a bandlimiting System. I donot know if other faults are in this Video, but please be carefull what he proclaims.
It is his opinion, which is obviously not always based on the correct technical knowledge or understanding.
Just to stay in front of a measurement Instrument makes you not to an expert.
The TRINITY DAC has an ideal Impulse Response in the "pure LIANOTEC" mode. You can playback a rectengular pulse. It is already described in the Manual.


In addition you can see what the loudness war can create, in the most cases the compressed digital recordings have a Peak Level of 0dBFS and will than be clipped by the digital filter. See the highlighted part.
That is valid for all DAC converter systems, which use a digital filter and is not a behaviour of the TRINITY DAC alone.
On the other hand you can see, that the TRINITY DAC can playback 0dBFS recordings in the "pure LIANOTEC " mode without any clipping.
Reference recording or Chesky are always recorded with a Peak Level well below 0dBFS, that is best recording technique.

 

[Julf]

Of course we all know the Fourier series and its effect on the rise time, but his conclusion is wrong. The visible ringing Comes from the digital filter and is not caused by a bandlimiting System.

Have a look at http://en.wikipedia.org/wiki/File:SquareWave.gif]this animation.

It shows how a square wave is built up from separate harmonics. The "wave" pattern is clealrly visible when the higher harmonics are missing. There are no digital filters involved. The "ringing" is not caused by digital filtering - it is a natural result of upper harmonics missing. It looks bad on a scope, but if the missing harmonics are above the limits of audibility, there will not be any difference in the sound.

In addition you can see what the loudness war can create, in the most cases the compressed digital recordings have a Peak Level of 0dBFS and will than be clipped by the digital filter.

Are you talking about inter-sample peaks clipping? As long as your analog stage has headroom, the normal interpolation filter will properly reconstruct the peaks going above 0dBFS.

 

[talibe801]

good morning to all,

refering to the trinity dac, we just received it 2 days ago here at Ultimate audio,and all i can say is that this is the best dac i heard in a very long time...and we deal with a lot of good stuff here.

 

[Julf]

Have a look at http://en.wikipedia.org/wiki/File:SquareWave.gif]this animation.

It shows how a square wave is built up from separate harmonics. The "wave" pattern is clealrly visible when the higher harmonics are missing. There are no digital filters involved. The "ringing" is not caused by digital filtering - it is a natural result of upper harmonics missing. It looks bad on a scope, but if the missing harmonics are above the limits of audibility, there will not be any difference in the sound.

An even better one is this flash presentation that allows you to step through the process (by using the round green arrow icon at the top right corner), adding individual harmonics. Still no digital filtering involved at all.

 

[Don Hills]

Hi Dietmar,
I am about to make some comments that may be interpreted by some people as being argumentative or dismissive of your work. I wish to say first that this is not my intention. I admire your work and the design ideas you have implemented. I think it is an interesting solution to some technical issues. I would buy one if I could afford it. (Not because I could likely hear a difference, but because I like implementations that "think outside the box".)

... the "wizard" starts to explore the ringing in the digital System. At least that Point is wrong explained. The fact that he could play a 20kHz sine wave in the beginning let us assume that his demo Setup is based on a regular digital Signal path with digital and analog filter. Of course we all know the Fourier series and its effect on the rise time, but his conclusion is wrong. The visible ringing Comes from the digital filter and is not caused by a bandlimiting System. I donot know if other faults are in this Video, but please be carefull what he proclaims.

My understanding is that if it were an effect unique to digital filtering, it would not be present on old NOS DACs with analogue (RLC) filters. But it is present... Also, ringing is a damped resonance, so it would it not also show up in the frequency domain as a peaking response?

The TRINITY DAC has an ideal Impulse Response in the "pure LIANOTEC" mode. You can playback a rectengular pulse. It is already described in the Manual.

I suspect that this is because you don't have a sharp cut-off reconstruction filter. At the effective sample rate that your DAC outputs, stray circuit RLC will form a filter (of rather indeterminate characteristics). I would expect the sound of your DAC to be quite sensitive to the cables used and to the RF sensitivity of the following electronics. Plenty of opportunity for audiophiles...

In addition you can see what the loudness war can create, in the most cases the compressed digital recordings have a Peak Level of 0dBFS and will than be clipped by the digital filter. See the highlighted part.
That is valid for all DAC converter systems, which use a digital filter and is not a behaviour of the TRINITY DAC alone.

This is a well-known issue, commonly called "inter-sample peaks". It's becoming less of a problem as filter designers become more aware of it, and most mastering engineers understand the requirement to allow some headroom (after any limiting and clipping) to avoid clipping in those filters that don't allow for it.

... Reference recording or Chesky are always recorded with a Peak Level well below 0dBFS, that is best recording technique.

My point exactly.

Regards,
Don.

 

[Al M.]

Right. Pretty much what I wrote, "There is no "even more inaudible". If something isn't audible, it isn't.". A signal-to-noise ratio (or dynamic range) way beyond the signal-to-noise ratio of your source material (or your listening room) is pretty much pointless. What is the background noise level of your listening room?

Around 25 dB -- and I live in an extremely quiet area.

 

[TRINITY]

Hi all to make a Long Story short.
If the ringing is not caused by the digital filter, why is the ringing different between the two digital filters and why can TRINITY with a bypassed digital filter create a rectangular pulse?

 

[Julf]

Dietmar,

If the ringing is not caused by the digital filter, why is the ringing different between the two digital filters

What I was saying is that the "ripple" or "ringing" shown in Monty's video wasn't caused by a digital filter. I didn't say some digital filtes don't cause ringing - some definitely do. But that is not what is shown in the Monty video.

why can TRINITY with a bypassed digital filter create a rectangular pulse?

Because it allows through enough of the (inaudible) upper harmonics that make a square wave look good. Relevant only if you like to watch square waves on your scope.

To make a long story short - with your technical knowledge, how do you explain the appearance of "ripple" or "ringing" in the two links I posted, where there is definitely no digital filter involved?

 

[Elberoth]

good morning to all,

refering to the trinity dac, we just received it 2 days ago here at Ultimate audio,and all i can say is that this is the best dac i heard in a very long time...and we deal with a lot of good stuff here.

talibe801 - thanks for joining the discussion. Would you mind sharing some info on what digital is/was your current reference ?

As you can imagine, without any references to known references, 'the best dac i heard in a very long time...' statement brings very little to the table.

 

[analog brother]

which power and digital cables do you use with the trinity?

 

[tunes]

Hi,

We received our unit of the Trinity DAC.
I read all of this thread and must give our tribute to Dietmar for designing this product.
Also want to say that the words and praise that some users described here are in line with our first listening tests done in our studio.

The equipment that we used to listen to the dac is in the foto below (sorry for the quality but was taken by mobile phone).
View attachment 13630

Warmest Regards
Miguel

Miguel,
Good to hear that you've got the dac in portugal at UA. Looking forward to your thoughts on the unit after you've had a listen over time. Will you have the trinity phono in as well? Bet it would sound superb with the tw acustic.
Chris

 

[ack]

Precisely. You could endlessly continue to play the game, like in re-writing the statement into a hypothetical discussion of 24 bit audio:

"He correctly states that lower noise results in wider dynamic range (letting you get closer to the theoretical limit of ~144dB for 24 bit audio, not necessarily cross it); naturally, you would expect that even higher-than-24 bit resolution will result in even higher dynamic range - you might argue we don't need more, but I'd say this is yet another limitation of 24 bit audio."

So what's the point, exactly?

The point, Al, is that at some point you have enough dynamic range and needn't go higher (notice, I did not really cap at any particular word-length), but to me 96dB isn't it - I assume you've been to Symphony Hall and understand dynamic range. But I can see where this is going and you can mock my arguments all you want, but up until this point I had a different impression of you. Done with this thread.

 

[Al M.]

The point, Al, is that at some point you have enough dynamic range and needn't go higher (notice, I did not really cap at any particular word-length), but to me 96dB isn't it - I assume you've been to Symphony Hall and understand dynamic range. But I can see where this is going and you can mock my arguments all you want, but up until this point I had a different impression of you. Done with this thread.

My intention, Peter, was not to mock your argument, but to point out, albeit perhaps in a quite drastic manner, that it is flawed. There is a huge difference between the two, and I hope you can see that. Others have pointed out the flaw as well.

Yes, I think I understand dynamic range, but perhaps not. The background in Symphony Hall is at least 25 dB, like in my living room, but probably considerably higher (I would be hugely surprised if in the most optimal case it would be below 35 dB with audience present), and I would be surprised if I had ever heard there anything topping 110 dB (which would be really screamingly loud and close to ear-damaging territory even at just brief exposures). That would leave an operable dynamic range of about 75 dB (at max 85 dB, with a putative and unlikely 25 dB background).

That guestimate of mine would correlate with what I just looked up in Wikipedia to confirm:
The dynamic range of music as normally perceived in a concert hall doesn't exceed 80 dB, and human speech is normally perceived over a range of about 40 dB.[9]

[9]^ Eargle, John (2005). Handbook of Recording Engineering. Springer. p. 4. ISBN 0-387-28470-2.

 

[Julf]

Dietmar,

I guess you didn't have a chance to answer this one? I am struggling to understand what makes LIANOTEC different from normal upsampling.

So are the LIANOTEC delayed oversampling points simply delayed, and thus repeated copies of the previous sample, or are they calculated interpolation values?

 

[Elberoth]

Dietmar,

I guess you didn't have a chance to answer this one? I am struggling to understand what makes LIANOTEC different from normal upsampling.

They are delayed. Please read the patent if interested. Dietmar have supplied the patent #.