The 2 philosophies in DAC design, hands off and hands on. Which is better?

[Orb]

Aha so that does confirm a transformer based filter, heck no idea how both me and Don missed that page
What made me think of the transformer being involved was the way they just involved it in their description of the product while mentioning no filters.
Muralman, if there was no filters then there is the possibility of really screwing up the equipment performance downstream of the DAC or generating random performance behaviour from the downstream equipment, it does go against the fundamentals of PCM technology (not just hifi but also it could be said what it was originally designed for and that was telecom world).

Don,
interestingly MSB Tech with their R2R ladder (ok some may feel this is not ideal) 24-bit DAC can be used as 16x oversampling or nonos with an analogue filter, no idea how current the DAC is though as I am not really up on these things, its their OEM product so some of the other high end manus seem to still offer a nonos route in a similar way you mentioned for ADI.

Cheers

Orb

 

[DonH50]

I looked at a couple of the products, didn't even notice the kits. The Audio Note site I found was a UK (I think) manufacturer; not even sure they are the same company!

Any type of DAC can be oversampled. Without a delta-sigma modulator, and assuming a constant filter to limit the bandwidth, you gain 1/2-bit SNR (reduction in in-band quantization noise) for every doubling of the sampling frequency. That is, take an ideal 16-bit 44 kHz "NOS" DAC with a 22 kHz output filter (brick wall) and you get about 98 dB SNR. Now, double the sampling frequency to 88 kHz, keeping the 22 kHz filter, and the you'll achieve about 101 dB SNR. The extra 3 dB (0.5 bit) is because you spread the quantization noise over a wider bandwidth, then filter out the "extra" high-frequency noise. Delta-sigma modulators ("OS" DACs) provide noise shaping that pushes the noise to the higher frequencies instead of spreading it flat across the band, so when you oversample and filter you gain even more in SNR, typically L+0.5 bits for an L-bit modulator. That is, a 3rd-order modulator (4 - 6 are typical now) provides an extra 3.5 bits (21 dB) for each doubling of the sampling rate. Note that since the noise is pushed to the high end of the band, you must oversample a delta-sigma DAC to achieve good in-band SNR; that is not a requirement for a conventional ("Nyquist") DAC.

HTH - Don

p.s. I am a bit uncomfortable straying too far off-topic ini muralman1's thread; we could continue this elsewhere if desired.

 

[muralman1]

DonH5O, Steve Williams will eventually hear what I have here. He will walk out a believer, and I won't even have to dunk him.

I don't think you are straying from the topic at all. I do have to admit, i read your glad OS tidings with a smile. My experience here is oversampling just doesn't sound like real music. The highs suffer the most. There is a lot of smearing, and detail loss. Ringing decay is sharply truncated. The stage collapses. Dynamics suffer a lot. There is a considerable amount of self noise, which is distortion to my ears.

Steve will bring a much better OS player than ever was tried here. If it is better, I will humbly accept. However, to best this system, OS players need something extra, like a splash of Holy Water, and a blessing from the Pope.

 

[DonH50]

Thanks, I'll try to be good.

My "glad OS tidings"? I have no dog is this hunt. I use what I can afford, which happens to be whatever's in my Oppo at the moment. I am aware of the limitations of OS (DS) and conventional Nyquist DACs. They each have their positives and negatives. One of the best players I ever heard was non-os (Accuphase) but was way out of my price range. It's a simple fact that it's much easier and cheaper to realize high accuracy and precision with a delta-sigma modulator-based (OS) DAC in today's technology. Not designed by me, however; not really my area of expertise. I haven't done much listening to high end DACs today; paying for the kid's college comes first!

I am very interested in how your trials go. - Don

 

[JackD201]

DonH5O, Steve Williams will eventually hear what I have here. He will walk out a believer, and I won't even have to dunk him.

I don't think you are straying from the topic at all. I do have to admit, i read your glad OS tidings with a smile. My experience here is oversampling just doesn't sound like real music. The highs suffer the most. There is a lot of smearing, and detail loss. Ringing decay is sharply truncated. The stage collapses. Dynamics suffer a lot. There is a considerable amount of self noise, which is distortion to my ears.

Steve will bring a much better OS player than ever was tried here. If it is better, I will humbly accept. However, to best this system, OS players need something extra, like a splash of Holy Water, and a blessing from the Pope.

This thread is like watching an HBO pre-fight 24/7. Let's hope the fight itself will be as entertaining. I would be mindful that not all fights end up in K.O.s. Split decisions are pretty common.

 

[muralman1]

But, I have the home court advantage! And, I think I can get Ali's, "Thrillah in Manilla," corner man.

 

[muralman1]

The H2O Fire preamp, a very different preamp.

Frequency Response 2Hz-500KHz -3Db
Max Output Voltage 10VRMS
Firepower Supply 2 250VA Toroidal Transformers
Filter Capacitance 78000uF
Fire Decoupling Capacitance 396000uF
Input Impedance 22Kohms
Output Impedance 20 Ohms
Inputs 4 single ended
Outputs 2 Single ended
Remote Yes
Remote Features Volume, Mute
FIRE Dissipation 100watts

 

[Orb]

I looked at a couple of the products, didn't even notice the kits. The Audio Note site I found was a UK (I think) manufacturer; not even sure they are the same company!

Any type of DAC can be oversampled. Without a delta-sigma modulator, and assuming a constant filter to limit the bandwidth, you gain 1/2-bit SNR (reduction in in-band quantization noise) for every doubling of the sampling frequency. That is, take an ideal 16-bit 44 kHz "NOS" DAC with a 22 kHz output filter (brick wall) and you get about 98 dB SNR. Now, double the sampling frequency to 88 kHz, keeping the 22 kHz filter, and the you'll achieve about 101 dB SNR. The extra 3 dB (0.5 bit) is because you spread the quantization noise over a wider bandwidth, then filter out the "extra" high-frequency noise. Delta-sigma modulators ("OS" DACs) provide noise shaping that pushes the noise to the higher frequencies instead of spreading it flat across the band, so when you oversample and filter you gain even more in SNR, typically L+0.5 bits for an L-bit modulator. That is, a 3rd-order modulator (4 - 6 are typical now) provides an extra 3.5 bits (21 dB) for each doubling of the sampling rate. Note that since the noise is pushed to the high end of the band, you must oversample a delta-sigma DAC to achieve good in-band SNR; that is not a requirement for a conventional ("Nyquist") DAC.

HTH - Don

p.s. I am a bit uncomfortable straying too far off-topic ini muralman1's thread; we could continue this elsewhere if desired.

Heya Don,
although that is theory about all dacs being nos capable?
Edit:
OK I think we are saying the same thing and my point earlier was identifying a chip/architecture that also manages nos; I agree all DACs can be oversampled, but not necessarily implemented ideally in a nos architecture .
End Edit:

From what I understand Sigma Delta DACs have performance issues (not ideal) when used in nos, this was partially shown in the previous paper I linked in this thread and also can be seen in some other stuff on the internet.
I could be wrong but my understanding is that the audionote nos DACs/chips used are also R2R architecture, same route as MSB Technology although I think MSB Tech developed their own chip - hence the OEM option to buy.
Any idea what the Accuphase architecture was Don?

Anyway, would be interesting comparing the Audionote to an MSB Technology R2R DAC in nos implementation, Muralman you ever had the chance to compare and if so curious how they differed in sound.
Personally from my experience I am more a fan of the FPGA/bespoke chip architecture when it comes to oversampling DACs, I guess it is to do with the potential of better performance on noise shaping/filtering.

Cheers
Orb

 

[muralman1]

Heya Don,
although that is theory about all dacs being nos capable?
Edit:
OK I think we are saying the same thing and my point earlier was identifying a chip/architecture that also manages nos; I agree all DACs can be oversampled, but not necessarily implemented ideally in a nos architecture .
End Edit:

From what I understand Sigma Delta DACs have performance issues (not ideal) when used in nos, this was partially shown in the previous paper I linked in this thread and also can be seen in some other stuff on the internet.
I could be wrong but my understanding is that the audionote nos DACs/chips used are also R2R architecture, same route as MSB Technology although I think MSB Tech developed their own chip - hence the OEM option to buy.
Any idea what the Accuphase architecture was Don?

Anyway, would be interesting comparing the Audionote to an MSB Technology R2R DAC in nos implementation, Muralman you ever had the chance to compare and if so curious how they differed in sound.
Personally from my experience I am more a fan of the FPGA/bespoke chip architecture when it comes to oversampling DACs, I guess it is to do with the potential of better performance on noise shaping/filtering.

Cheers
Orb

Hi Orb, no, I have not heard any MSB players. I have heard the AMR CDP 77 in SF which has variable oversampling options. The owner and I found NOS setting was best. Now, he has moved to a SET system, keeping only the AMR player out of the system I heard.

That brings me to another reason NOS is better. I know of several SET systems ranging upwards to $150k. They ALL use NOS DACs. I think you would be hard pressed finding an SET system that has an OS player.

Another guy brought his expensive APL player here. The OS was switchable out. We both found the setting with the OS off was the best, by far. We called the guy (isn't that Modwrite?) and he argued with us, even though we had the proof right here. That period, I was using Pass Labs gear.

 

[JackD201]

Dan modded AMR's?

 

[muralman1]

Dan modded AMR's?

No, the modded player was the Denon APL. That was a neighbor of mine. The AMR is used in a different system enjoyed by a guy in SF.

 

[Orb]

Muralman,
thanks for the heads up, very interesting in a good way
The AMR CD/DACs also have a pretty good reputation themselves.
IS it possible to say how the various nos dacs sound different to you?
Just curious how the NOS sound differentiates between various NOS competitors, I appreciate that this may be very subtle and comes down to long term preferences but would be interesting.

Cheers
Orb

 

[JackD201]

Then that would be Alex not Dan.

 

[muralman1]

Muralman,
thanks for the heads up, very interesting in a good way
The AMR CD/DACs also have a pretty good reputation themselves.
IS it possible to say how the various nos dacs sound different to you?
Just curious how the NOS sound differentiates between various NOS competitors, I appreciate that this may be very subtle and comes down to long term preferences but would be interesting.

Cheers
Orb

When it comes to NOS, my experience is limited. That would be the case for nearly everyone, I would wager. NOS is a rare breed. I have heard the Audio Note CD player (which one I can't remember), the AN One.1, the AN 2.1, the AN 2.1 modded, the AN 3.1, the AN 4.1 balanced, the AMR, and finally, the APL Denon.

The APL was the worst of them on my system. The 4.1 balanced gave a very graceful sound. Detail was very good. Someday, I hope to hear that DAC in conjunction with the Flatfish transport. With the diode swap, that would be a killer DAC.

AN DACs have rolled off highs. The bass is indistinct too. The 2.1 modded is my present DAC. I can't tell you how important the diode change is. This sound is the best of all the mentioned DACs and players. That is because of all the accurate leading edges. Rim hits, all percussion including piano are far realer. The bass is deeper and more focused.

 

[mep]

But, I have the home court advantage! And, I think I can get Ali's, "Thrillah in Manilla," corner man.

Funny you mention Ali. You kind of remind me of what Ali said about Howard Cosell. Ali once remarked that every time Howard Cosell opens his mouth he should be fined for air pollution. That's what I think of your take no prisoners attitude you have about NOS DACs and stinky fish transports, super thin speaker wire with almost no insulation, and magic diodes. Everybody that comes to your house leaves crying knowing that their stereo sucks compared to the glory of yours. Wow Zen master. I will be interested to see what Steve thinks of your system.

 

[mep]

But remember Tom, no one can know which component really sounds better, only the component which they prefer. I don't know how you and Ethan ever came to that conclusion, but I do find it astounding. Hey Ali, I need you to come and say a few words to Tom and Ethan. Seriously, you said you were looking forward "to the differences detected between the two cd players." If you can detect differences, doesn't that infer that you should be able to pick one that clearly sounds better than the other assuming there is a clear difference? And I would think there will be clear differences between a NOS and OS DAC. I know that Muralman thinks he will need to pass out crying towels once everyone hears his blowfish. What I am certain of is that one DAC will sound more like real music than the other and that DAC "wins" and therefore sounds better.

 

[muralman1]

Every kitchen has an old pot.

 

[Steve williams]

I guess for my $0.02 is whether muralman1's player/dac sounds better in my system. I have always felt that any system is a synergy amongst all of it's parts. There is no doubt that Vic's system sounds fantastic due to his synergy. Sometimes taking the same component and placing it in a different system only to find that the synergy is gone

 

[DonH50]

Heya Don,
although that is theory about all dacs being nos capable?
Edit:
OK I think we are saying the same thing and my point earlier was identifying a chip/architecture that also manages nos; I agree all DACs can be oversampled, but not necessarily implemented ideally in a nos architecture .
End Edit:

From what I understand Sigma Delta DACs have performance issues (not ideal) when used in nos, this was partially shown in the previous paper I linked in this thread and also can be seen in some other stuff on the internet.
I could be wrong but my understanding is that the audionote nos DACs/chips used are also R2R architecture, same route as MSB Technology although I think MSB Tech developed their own chip - hence the OEM option to buy.
Any idea what the Accuphase architecture was Don?

Anyway, would be interesting comparing the Audionote to an MSB Technology R2R DAC in nos implementation, Muralman you ever had the chance to compare and if so curious how they differed in sound.
Personally from my experience I am more a fan of the FPGA/bespoke chip architecture when it comes to oversampling DACs, I guess it is to do with the potential of better performance on noise shaping/filtering.

Cheers
Orb

I am not sure I followed this one, Orb.

The ADI DAC was designed to be oversampled; it is not just theory, it works in practice, with or without a delta-sigma modulator. I think a description of DAC architectures is outside the scope of this thread, or at least outside the time I have available at the moment. I have been asked and am thinking about it, but am swamped at work and have a concert next weekend so do not have the hours it takes to put something like that together. Later...

Delta-sigma OS, NOS Nyquist, they all have positives and negatives. As I said, a delta-sigma DAC must be oversampled to provide room for noise shaping and filtering (there are in fact NOS delta-sigma designs, but I have not seen a commercial version). Nyquist (NOS) DACs can be R-2R with voltage or current switches, are usually segmented (binary and unary stages) to improve linearity, and there are a host of trades that go into the technical design. Too deep for here, I think, or chalk it up to laziness on my part. It would take a lot to explain it all.

The ADI DAC in the Audio Note (if it is the AD1865 we are talking about) is a segmented R-2R DAC with four unary bits and the rest binary.

I have no idea the specific architecture the Accuphase used, just that it was a discrete DAC, so probably binary or minimally segmented and very likely an R-2R approach.

ASIC, FPGA, full-custom, or whatever, what matters is the algorithm and filter implementation. FPGA's allow reprogramming the filter but offer no other advantage; an EEPROM will accomplish the same goal and is cheaper, though an FPGA is more flexible.

 

[muralman1]

I guess for my $0.02 is whether muralman1's player/dac sounds better in my system. I have always felt that any system is a synergy amongst all of it's parts. There is no doubt that Vic's system sounds fantastic due to his synergy. Sometimes taking the same component and placing it in a different system only to find that the synergy is gone

Yes, Steve, this is perfect synergy. I hope to be able to drive again, and if you will have me, I would find the drive to Danville a delight. It would be fun to see what this source set up is like on another system. I have, "Wolf at the Ruins," by Forest Fang on now. You can let your imagination fly listening to this.