In terms of subjective preferences, there are a lot of good reviews for all the TotalDACs on Audiostream. The tube output one had a glowing review too. That's fine and it got some of my interest, but from experience I know to take reviews with a grain of salt.
Following Vincent's reply I understand if I were to summarize:
- the maximum amplitude is limited to 14 bit and there is a relation to 0.01% distortion as a limiting factor
- ENOB not clear/unconfirmed at the moment
- I don't dispute the measurements, but it's still unclear to me how the above allows for such a high DR
It seems quite clear also from this article the relationship between ENOB and DR:
http://www.newelectronics.co.uk/electronics-technology/stretching-the-dynamic-range-of-a-d-converters/46404/
ENOB = (dynamic range – 1.76)/6.02
As it stands either the DAC has just over 26 bits (ENOB) or the DR should be much lower.
Very interestingly someone pointed me to this link: http://www.totaldac.com/boitier_stereo.htm
Even for what was the entry level the exciting DR figures come in again: 134dBFs which should correlate 22 ENOB. If we compare this to MSBs Analog DAC's 133dB, the Yggdrasil's 122dB (measured by atomicbob) or the Theta V A's 110dB, the TotalDAC A1 at a fraction of the cost to the MSB could look like a winner in these areas.
Now as it happens I have actually heard the TotalDAC A1, at home and at length, and my comments are posted elsewhere. Not only that it didn't come anywhere within sight of the performance of the MSB Analog or Yggdrasil, but 3 of us independently on various rigs compared it with the Theta V A. The TotalDAC A1 came awfully short in resolution, DR, clarity or precision to the very old Theta V A. The Theta V A has a DR of around 108-110dB, which gives 17-18 effective bits. Did I think the A1 sounded nice - yes. It was organic, engaging, had good tonality etc. But at the same time it could not perform or resolve even remotely well compared to the Theta. The final proof, if any was needed, was that the TotalDAC A1 owner has purchased a Theta V A.
I am not saying the TotalDAC D1 and monos will not be much better. I am sure they are. But in this only example of hearing a TotalDAC, the DR spec, which is (about) the only performance related spec published, unfortunately did not translate to anything remotely comparable when hearing the DAC.
I look forward hearing the TotalDAC D1 (hopefully before the EOY) to know first hand if it can actually match the MSB Analog's resolution, clarity, background blackness, and go beyond. That would be impressive and I would be glad there is more worthy competition out there, from much closer to me.
In terms of subjective preferences, there are a lot of good reviews for all the TotalDACs on Audiostream. The tube output one had a glowing review too. That's fine and it got some of my interest, but from experience I know to take reviews with a grain of salt.
Following Vincent's reply I understand if I were to summarize:
- the maximum amplitude is limited to 14 bit and there is a relation to 0.01% distortion as a limiting factor
- ENOB not clear/unconfirmed at the moment
- I don't dispute the measurements, but it's still unclear to me how the above allows for such a high DR
It seems quite clear also from this article the relationship between ENOB and DR:
http://www.newelectronics.co.uk/electronics-technology/stretching-the-dynamic-range-of-a-d-converters/46404/
ENOB = (dynamic range – 1.76)/6.02
As it stands either the DAC has just over 26 bits (ENOB) or the DR should be much lower.
Very interestingly someone pointed me to this link: http://www.totaldac.com/boitier_stereo.htm
Even for what was the entry level the exciting DR figures come in again: 134dBFs which should correlate 22 ENOB. If we compare this to MSBs Analog DAC's 133dB, the Yggdrasil's 122dB (measured by atomicbob) or the Theta V A's 110dB, the TotalDAC A1 at a fraction of the cost to the MSB could look like a winner in these areas.
Now as it happens I have actually heard the TotalDAC A1, at home and at length, and my comments are posted elsewhere. Not only that it didn't come anywhere within sight of the performance of the MSB Analog or Yggdrasil, but 3 of us independently on various rigs compared it with the Theta V A. The TotalDAC A1 came awfully short in resolution, DR, clarity or precision to the very old Theta V A. The Theta V A has a DR of around 108-110dB, which gives 17-18 effective bits. Did I think the A1 sounded nice - yes. It was organic, engaging, had good tonality etc. But at the same time it could not perform sufficiently good compared to the Theta. The final proof, if any was needed, was that the TotalDAC A1 owner has purchased a Theta V A.
I am not saying the TotalDAC D1 and monos will not be much better. I am sure they are. But in this only example of hearing a TotalDAC, the DR spec, which is (about) the only performance related spec published, unfortunately did not translate to anything remotely comparable when hearing the DAC.
I look forward hearing the TotalDAC D1 (hopefully before the EOY) to know first hand if it can actually match the MSB Analog's resolution, clarity, background blackness, and go beyond. That would be impressive and I would be glad there is more worthy competition out there, from much closer to me.
These are the challenges of lab measurements, aren't they? They can provide some objective quantification of what we are hearing or what we should hear but they really can't tell us if something sounds good. My comments on resolution were only in response to your query and it was a valid question. In the end, with respect to MSB vs TotalDac or other fine DACs at this level, it will amount to splitting hairs rather than major deltas. I have always valued your comments and I look forward to hearing your take on the TotalDac D1 once you hear it.
In an interview on Stereo and Mono, here is what Vincent had to say on this matter:
“The truth is in listening, not in the theory or standard measurements. So the listening test decide what is good, even if it is unexpected or unbelievable. I like working like an instrument maker, not only as an electronic engineer. An "engineer only" work leads to very good measurements but most time a not so good sound.”
I share your curiosity about this, Darryl. I have no explanation for how some of these players exceed both yours and my highly tweaked, purpose-built PC. All I can say is the proof is in the listening. With the Aurender N10, for example, you will hear it in the first 30 seconds. If you don't, then your baseline setup is better than mine.
With the USB Regen, I believe this is only useful for fixing issues related to the USB signal. It will not, for example, make a 96k mp3 file sound like anything more than it is but I know that you know this. I believe there is so much molestation occurring to a file within the PC chassis that something like a USB Regen amounts to nothing more than a band-aid rather than a proper fix. I have the USB Regen on my TotalDac + Server as we speak and I am barely hearing a difference. With the Aries on my Bricasti, the difference was significant.
The best I can tell you is to call Brian and setup an audition. I told him you would probably be contacting him.
Brian and I are engaged. I'm guessing I'm after you. Thanks a lot for that information, it's going to be really interesting testing them out.
The TotalDAC Server has the re clocker in it - correct? If so I see why adding the Regen would yield no improvements. You put a re clocker on top of a re clocker - shouldn't expect to yield any positive results from that. One the other hand it makes sense adding the Regen to the M1 / Aries and getting good results.
30 seconds is a bit of a stretch. But I'm going to put it through all the paces.. If nothing else it looks like it's very user friendly, so that's plus right there. If I could get grid of this big monitor for an iPad, then I'm all for that..
Brian and I are engaged. I'm guessing I'm after you. Thanks a lot for that information, it's going to be really interesting testing them out.
The TotalDAC Server has the re clocker in it - correct? If so I see why adding the Regen would yield no improvements. You put a re clocker on top of a re clocker - shouldn't expect to yield any positive results from that. One the other hand it makes sense adding the Regen to the M1 / Aries and getting good results.
30 seconds is a bit of a stretch. But I'm going to put it through all the paces.. If nothing else it looks like it's very user friendly, so that's plus right there. If I could get grid of this big monitor for an iPad, then I'm all for that..
Sure. I'm literally not kidding about 30 seconds. This is how long it took me. The N10 didn't even have to warm up much but again your baseline system may be better then mine. I'll be interested to hear your impression. As for the iPad user interface, it is excellent.
Aurender uses an iPad interface called Conductor to control it. It is the only way to control it. It is the equivalent of MPAD for the TotalDac but unlike MPAD, Conductor can also play Tidal. Roon capability is also expected.
about the time resolution, just to give a sense of what something like 7µs can mean, 50µs is the period of a 20khz frequency, so of course a 16/44 format that cuts everything past 22khz wouldn't bother to collect information about waves that are faster than 22khz...but the science part clearly had some problems.
Sorry, it's hard to completely understand what you've said but I believe you misunderstand what auditory time resolution is because it has nothing to do with our auditory frequency spectrum (20-20k Hz) or anything beyond 20kHz. It merely describes the time interval by which we can discern 2 sounds. If 2 sounds occur 10 µs apart, our ears can probably discern it. If 2 sounds occur 5 µs apart, our ears probably won't be able to discern the 2 sounds. That's it.
so the way you explain the "need for speed 192" (soon in your theater), and using the meridian lolz numbers, is making amalgams as good as saying you can hear 140khz or something like that(7µs period).
because you can't talk about speed as frequence resolution only when it's convenient to show how highres is better, I believe we should not do that at all, but if we do we must do it all the way. on a side note, I believe you must have misplaced something. 20.8µS just so happens to be the period of a 48khz frequency, not 44khz.
You misunderstand the practical significance of oversampling. Yes, it increases bandwidth beyond 20kHz which serves us no purpose as we cannot hear any sound beyond 20kHz but oversampling also serves to increase time resolution and this is why oversampling is done. At 44.1 kHz (CD quality), a sample is taken every 22.7 µs. I had quoted a value of 20.8 µs and you are astute to have realized that this is the sampling frequency for 48 kHz and so thank you for the correction. At 96 kHz, the sampling time is 10.4 µs as I previously stated. At 192 kHz, the sampling time improves to 5.2 µs. Our auditory system can react very quickly to frequency changes, in the order of 5-10 µs although the literature frequently cites 7µs. 192 kHz was never an arbitrary target but a very strategic one based on this information. What is a curiosity to me is the DXD standard which oversamples at a rate of 384 kHz. Someone has to explain the logic of this one to me.
music still happens to be made of waves and not squarewaves, so a significant change in amplitude that would be as fast as 7µS, that would create an ultrasonic content in the music, not some fancy transient on the first impact of a cymbal that would feel like speed.
Again, I believe you misunderstand the practical significance of time resolution based on this statement. The amplitude of a sound wave results in volume and the frequency of a sound wave is what determines pitch. The rateof change of both the amplitude and frequency is time resolution. If a cymbal is hit, the leading edge of that sound hits our ear first and the signal that follows helps us to determine the direction and distance of the sound. It is this 3D quality that oversampling hopes to achieve.
You're right. Analog media has its issues including wow and flutter in the same way that digital systems have to deal with jitter but this doesn't impact time resolution. Again, you are misunderstanding what time resolution is. There is no need to oversample a record because it has no gaps of information. With digital, you oversample to fill in the gaps.
about vinyls recorded at 20bit and how that's why vinyl can sound more dynamic, how about the effective resolution of using a vinyl? as in the sound going out of a turntable and the funny SNR and distortions it gives? no more 20bit out there I tell you that. or simply mention how most masters on vinyls must be limited in dynamic as to avoid having the needle jumping out of the groove on too big amplitudes at certain frequencies.
Sure, I agree with you here and these are the most valid points in your entire response. I am not trying to make any statement about which format is superior, only to say that analog is often a reference point of comparison. Since the 1970s, most vinyl mastering has been done with digital delay lines instead of analog delays on the signal going to the lathe that cuts the spiral groove so even if the recording, mixing and mastering was done using analog gear, at some point, A/D conversion has to take place and most A/D converters are 20 bit. That is where the 20-bit figure is derived. What happens after that is dependent on many factors including the quality of the turntable, tonearm, etc. just as you've stated and yes, the effective dynamic range is likely to be less than 20 bit. However, I have heard implementations where digital has been compared side by side to analog using a 16/44 file against the identical track on vinyl and to all in the room, the vinyl presentation was considerably more dynamic.
Don't feel bad about this one, you just have to look at it from a different perspective. If you look at it from the standpoint of octaves, 20-20,000 Hz represents 10 octaves. From 10,000-20,000 Hz is 1 octave. At 16.5 Khz, you stil can hear 96.5% of the audible frequency spectrum. That's not so bad
I'll leave your subjective impressions out for others to interpret as they please. I don't know the DACs and feelings are personal. but the science part clearly had some problems.
I imagine this should probably move to sound science, but until then, here I go.
I effectively do not understand what you're talking about. sure I get the general concept, but I fail to imagine it in actual audible soundwaves. concept, that a percussion has a very fast rising time on impact, so to be reproduced correctly it needs fast sound reproduction and stuff like that.
but in practice the sound is in the audible range and will last more than a period of an audible frequency, so we don't really need super small data points do we? for a cymbal, I would imagine the first part of the sound(before it start going up in frequencies) would last a few ms. that's several magnitude bigger than what you've been arguing about. and again if something of audible amplitude variation was to rise faster than some 20µs, then it would express ultrasonic content right? I can understand the concept, and in very rare extreme test tracks passages, I even could hear something that could go with the idea. but that was with mp3 not 16/44 PCM. I have no concrete situation of CD music not sounding like highres(aside from remastering), so of course when I read about how we are able to identify stuff smaller than the sample size of a CD, I wonder what could those small stuff be?
here is how I conceptualize super short time events that could possibly be recognized by ear:
1/ we have an audible tone, we instantaneously stop it( would a headphone be able to stop fast enough?) for a few µs, and start it again with the same tone rising calmly as to become an audible frequency. in such a situation I imagine I could probably get to hear something out of a very small time delay. but what instrument does that, what transducer can record and replay it accurately?
2/ left and right ear differences that serve to estimate the direction of a sound. those differences would be rather small when the source is rather close to be on the axis right in front of us. and I'm willing to believe we can identify those very small left right delays and use them to place the sound in space. IMO what could be wrong would be in a worst case scenario, the first sample, and by as much as 1/44100s. the spacing between samples is much better than the sample resolution(super duper clocks) so the wave will still be created accurately over a few samples(as long as it's in the audible range), and phase difference will be properly estimated soon after the first few samples.
even if at some point we were to mistake a sound at 0.3° in front of us for a sound IDK maybe at 1°(and I see no reason for that to happen on a CD), I'm not so sure we humans have that much accuracy when placing sounds in space. and I doubt the guy doing the mastering is setting the panning of tracks within 2degrees. most likely he's turning a knob until the sound is approximately where he wants it. there is no losing the real sound in this matter even if we were to be misguided by 16/44 and I'm not sure at all that we are.
within the same kind of idea, if a phase error of 7µs is enough, then anything more than... let's go for 35° out of phase in the audible range, would potentially create such a delay. and our gears are usually minimum phase, not perfect phase. so this + that+ a great many headphones(mostly dynamic), and many of us end up with variations that are bigger than what could come because of 16/44. yet I don't see too many people crying about it and proving that they can distinguish the sound difference.
3/ the starting moment of a sound. will the album be ruined if the guitar starts up to 22µs later than it did in the studio? who will ever know, let alone notice? did all recording engineers in the 70's align all their tracks withing 1µS margin? to me that one is about splitting air, not getting a better audio.
4/ stuff like shockwaves from explosion(not my favorite music TBH ^_^)? but I know nothing about it except that it still needs to travel in the air, be recorded and replayed.
I wont say I've ruled out everything, but all my other ideas right now involve ultrasonic sounds, so I just dismiss those as "not heard by me".
conclusion, while I would never deny the potential for better accuracy in high res, I feel like I can deny the claim of 16/44 falling under what we can all distinguish because we can distinguish some 7µs stimulus. if only because of how few honest and controlled material we have showing that people can identify a 16/44 track(and not just the low pass filter of the DAC, or some massive IMD from loud ultrasonic content being badly handled by the sound system). I don't know under what circumstance a human can react to a 7µs stimulus, but I doubt it happens often in the waveform of audible music.
You're right. Analog media has its issues including wow and flutter in the same way that digital systems have to deal with jitter but this doesn't impact time resolution. Again, you are misunderstanding what time resolution is. There is no need to oversample a record because it has no gaps of information. With digital, you oversample to fill in the gaps.
well if errors in time huge enough to actually change the pitch of sounds doesn't impact time resolution, I wonder what does ^_^.
to me oversampling is mainly a mean to use more convenient low pass filters, not a way to fill the gaps(that said I'm 100% pro OS). if you question the number of samples and the need to reduce the gaps, I guess you question either Nyquist and pretty much all that is digital sound to this day, or our inability to hear ultrasounds. else 16/44 seems just fine for most uses.
Quote:
Sure, I agree with you here and these are the most valid points in your entire response. I am not trying to make any statement about which format is superior, only to say that analog is often a reference point of comparison. Since the 1970s, most vinyl mastering has been done with digital delay lines instead of analog delays on the signal going to the lathe that cuts the spiral groove so even if the recording, mixing and mastering was done using analog gear, at some point, A/D conversion has to take place and most A/D converters are 20 bit. That is where the 20-bit figure is derived. What happens after that is dependent on many factors including the quality of the turntable, tonearm, etc. just as you've stated and yes, the effective dynamic range is likely to be less than 20 bit. However, I have heard implementations where digital has been compared side by side to analog using a 16/44 file against the identical track on vinyl and to all in the room, the vinyl presentation was considerably more dynamic.
I would attribute your testimony of feeling more dynamic to 2 ideas:
1/ the coloration and added distortions/noise/crosstalk end up really sounding different and perhaps at times, more engaging?
2/ non linearity. the needle going far too high on some trebles and giving an output several db higher than the signal on the groove intended. but as this would happen on louder parts with the most magnitude, we can end up with a wider range of amplitudes. quiet part quieter and loudest part louder.
Sorry, I wish I could speak French because clearly you have good intelligence on the science of sound and I would love to engage you in meaningful conversation but I am simply struggling to understand you. I still believe you are understanding time resolution differently than I am. Perhaps another analogy I can use is soundstage, both width and depth. Have you ever tried listening to a large and complex orchestral performance recorded in 96k mp3? Quite likely, you will find it to sound very 2 dimensional. You might recognize the presence of woodwind instruments but you probably will struggle to know where on the stage the woodwind section is relative to the string section. To take it further, within the string section, even though they travel at different frequencies, you will probably struggle to differentiate the differences between an upper standup bass versus the lower registers of a cello. Then play that same recording via a properly mastered SACD which is 64x the resolution of a CD and all of a sudden, you know exactly where everything is on the stage. This 3D quality has everything to do with time resolution and what oversampling hopes to achieve.
Sorry, I wish I could speak French because clearly you have good intelligence on the science of sound and I would love to engage you in meaningful conversation but I am simply struggling to understand you. I still believe you are understanding time resolution differently than I am. Perhaps another analogy I can use is soundstage, both width and depth. Have you ever tried listening to a large and complex orchestral performance recorded in 96k mp3? Quite likely, you will find it to sound very 2 dimensional. You might recognize the presence of woodwind instruments but you probably will struggle to know where on the stage the woodwind section is relative to the string section. To take it further, within the string section, even though they travel at different frequencies, you will probably struggle to differentiate the differences between an upper standup bass versus the lower registers of a cello. Then play that same recording via a properly mastered SACD which is 64x the resolution of a CD and all of a sudden, you know exactly where everything is on the stage. This 3D quality has everything to do with time resolution and what oversampling hopes to achieve.
oh don't worry, it's also horrible to read my posts in french. ^_^
how about you explain to me, or point me toward those 7µs tests? if I can get my head around an actual situation where such time difference can be noticed, maybe I'll get it. my only personal experience of minute changes being audible is when changing phase only in one channel. but it's out of context here.
your example of a 96kbps mp3 has IMO too many other components to be a valid example. just removing so much content in the trebles as 96kbps does, will do what you talk about. we're bound to lose some sense of precision because trebles are easier to pinpoint in space than low frequencies. so by losing some audible trebles we do lose some of that pinpoint accuracy. also the artifacts become clearly audible and must be shouting to our brain "something isn't right!!!!".
but at least I understand how missing some audible frequencies can do what you explain. but that doesn't apply to 16/44(at least not to my ears^_^).
Does anyone have any suggestions ? I would like to use the D1 Dual with headphones and to drive my mono blocks . I dont want to keep changing XLR Cables since there is only 1 pair of XLR outputs . In order to take full advantage of the D1 I need to use the XLR outputs
My Mono blocks are single ended so I use the XLR to RCA Transformers from Totaldac
oh don't worry, it's also horrible to read my posts in french. ^_^
how about you explain to me, or point me toward those 7µs tests? if I can get my head around an actual situation where such time difference can be noticed, maybe I'll get it. my only personal experience of minute changes being audible is when changing phase only in one channel. but it's out of context here.
your example of a 96kbps mp3 has IMO too many other components to be a valid example. just removing so much content in the trebles as 96kbps does, will do what you talk about. we're bound to lose some sense of precision because trebles are easier to pinpoint in space than low frequencies. so by losing some audible trebles we do lose some of that pinpoint accuracy. also the artifacts become clearly audible and must be shouting to our brain "something isn't right!!!!".
but at least I understand how missing some audible frequencies can do what you explain. but that doesn't apply to 16/44(at least not to my ears^_^).
I wish I could speak French half as good as you can speak English. There are actually numerous studies and this threshold has been reported to be anywhere from 5-10µs. Here is a link to one of those studies and it references others:
http://boson.physics.sc.edu/~kunchur/temporal.pdf
Here is also a video by Hans Beekhuyzen that you might find helpful:
With the flip of a switch and the swap of a chip, I can convert the monobloc to a d1-dual. Once everything is burned in, I will be comparing monobloc (which requires the reclocker) vs d1-dual + reclocker vs d1-dual by itself.
Does anyone have any suggestions ? I would like to use the D1 Dual with headphones and to drive my mono blocks . I dont want to keep changing XLR Cables since there is only 1 pair of XLR outputs . In order to take full advantage of the D1 I need to use the XLR outputs
My Mono blocks are single ended so I use the XLR to RCA Transformers from Totaldac
Yes, this is somewhat of a dilemma, and you might want to base your decision on which system you listen to the most. With the 6.35mm (1/4") jack in the back, I found the SQ to be quite good, just under powered for certain headphones. If you intend to drive an LCD-X which has an impedance of 20 ohms, the power output from this jack will probably be adequate given its high sensitivity although you can A/B and decide for yourself. For the HD800, that headphone port will drive it just fine. For something like the HE-1000, I found that port to not be powerful enough. Even though the output from that jack is unbalanced and only takes advantage of half of the ladder, I found it to be excellent.
The alternative would be to drive your monoblocs single ended from the RCA outputs and keep your headphones connected via XLR. Personally, this is probably what I would do.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
