Best DAC ever heard

[AudioExplorations]

I found the 1.06v setting to be too high when using the internal digital preamp.

0.4-0.5v would've been spot on. Maybe this could be added in a firmware update?

 

Same here, I have the medea set at 0.49V.

 

You do realise that your invicta DAC has 4.6V on the XLR and 2.3V on the RCA?

[spacey]

It works differently IIRC Keith. Have look here, its a review of the prototype INVICTA: http://www.6moons.com/audioreviews/resonessence/2.html

 

C&P'ed from the link above, it looks to be variable, i may email Resonessence and ask them, apprears to be some kinda hybrid:

 

with 16bit audio its bit accurate all the way down to -96db. With 24bit audio its bit accurate to -48db. Even then, the most it ever lobs off is max -4db S/N.

 

Output voltage on XLR/RCA vs. readout equals 5/2.5V for zero attenuation, 4/2V for -2, 2/1V for -8, 1/0.5V for -14, 0.5/0.25V for -20 and 0.2/0.1V for -28. The manual volume control proceeds in 0.5dB intervals from -125.5dB to zero.

 

The volume range on the production model goes to -127.5db

[Moiz Fidelity]

Nobody mentioned this? Lavry DA2002: Lavry Engineering

 

Discrete R2R design. Supposedly among the best. I don't have or listen to one though.

 

 

Dan Lavry is one of the best designers there is, but beware that the reference DACs always use special parts and may in the future not repiarable. Some of the old Lavry Gold DACs and ADC's can not be repaired because of lack of parts

[AudioExplorations]

It looks to me like it's fixed at 4.6V (XLR) and 2.3V (RCA) and the variable output is digital attenuation only.

[spacey]

Maybe at 0db its 4.6v, but at -20db the output voltage is as low as 0.2v. there must be more at work than just simple digital attenuation. They claim it to be bit perfect down to -96/48db @ 16/24 bit respectively. If it was solely digital attenuation it should surely be lobbing off LSB's way before then??

 

Even beyond -96/48db it only ever looses 4db max. Which leads me to think there's some kind of output voltage variation going beyond the digital attenuation.

[earflappin]

Weiss DAC's are the only products that overcome this hurdle with the adjustable output voltage.

 

Audio, I don't believe this statement is correct. The Lynx Hilo (and other pro DACs AFAIK) has this same adjustability.

[AudioExplorations]

ear- just checked the hilo manual and you are right. This is the first time I have seen this outside of weiss and Barrow's DIY DAC. Any others you are aware of?

[barrows]

Note: the invicta uses the volume control built into the ESS 9018. This volume control runs at 48 bits, so with 24 bit music, there still 24 bits to work with before any data is lost. 24 bits of attenuation is -144 dB, so it is safe to say that one will never have any data loss with the Invicta, regardless of the output V.

For other DACs (besides Weiss), it is not all that difficult to add a simple voltage divider to lower the output voltage a little to match an amplifier, if you are not handy, you could get an electronics tech to do this for you. And do not get all worried about this negatively affecting the sonics (what do you think an analog volume control does, and what do you think the Weiss produtcs do internally). If you want the voltage divider to be "perfect", order 0.1% tolerance TX-2575 resistors from Texas Components (about $10.00 each, you will need two per signal leg, so a total of eight for balanced stereo connections). The voltage divider might be able to be built into your DAC if there is room, but it is better to put it at the input to the amplifier (this allows the cable to be driven at a higher voltage, nice). Another option is making this adjustment internally in the amplifier, this can get a little more difficult, but is possible with many amps just by changing some resistors on the input board. Or it could be added as a small plug in right at cable inputs to the amp. There are some of these already on the market, but I would recommend making your own, or using a good tech, so that you can use the best resistors-with TX-2575s you will never hear them.

For manufacturers I think the best soultion is allowing for a few gain settings at the input of the amp. Most amps designs could accommodate this by just having a couple of resistors per channel mounted in screw terminal sockets. My Pass has two gain levels, adjustable internally by a jumper plug: 26 dB and 30 dB.

So, there are ways to adjust two components to match for gain quite nicely, just find a sympathetic electronics tech in your area. Of course, always contact the manufacturer of your gear before making mods if you are concerned about warranties, they might be willing to make these mods for you in some cases...

[spacey]

The course volume setting is nothing special, it’s simply an analog -db pad. No big deal. In pro devices this is normally done with internal -db jumpers. The Benchmark DAC1 has these and small potentiometers to adjust further on the rear. The Antelope is the same also. The difference with Weiss is from there the rest is done in the digital domain.

[4est]

The course volume setting is nothing special, it’s simply an analog -db pad. No big deal. In pro devices this is normally done with internal -db jumpers. The Benchmark DAC1 has these and small potentiometers to adjust further on the rear. The Antelope is the same also. The difference with Weiss is from there the rest is done in the digital domain.

 

Are we sure that they are only voltage dividers at the outputs? I am not saying they are or are not, but there are very good reasons why one typically sees the attenuation between stages in a typical preamp. In part this is to avoid interactions with downstream devices.

[barrows]

Are we sure that they are only voltage dividers at the outputs? I am not saying they are or are not, but there are very good reasons why one typically sees the attenuation between stages in a typical preamp. In part this is to avoid interactions with downstream devices.

 

4est. We cannot be sure how the output level is adjusted without looking inside the case, or getting confirmation from the manufacturer. In the case of Benchmark, considering their circuitry, it likely is just pots on the output. But with Weiss, they may, perhaps, adjust the actual gain of the circuit. This can be an important consideration, as using a voltage divider directly on the output of the DAC will raise the output impedance somewhat.

If one is interested in attenuating the signal to obtain a better match when running amp direct, the best way to do it is at the input to the amplifier, this way the output impedance of the DAC is preserved, and the input impedance of the amplifier is actually increased.

[Miska]

using a voltage divider directly on the output of the DAC will raise the output impedance somewhat

 

And it will also increase noise... So best is to go straight to suitable level without involving attenuation.

[earflappin]

ear- just checked the hilo manual and you are right. This is the first time I have seen this outside of weiss and Barrow's DIY DAC. Any others you are aware of?

 

My LIO-8 has software adjustable output voltage, but not sure about the other pro DACs. It is a nice feature.

[barrows]

And it will also increase noise... So best is to go straight to suitable level without involving attenuation.

 

Hmmm, not quite sure what you mean Miska. I guess you are referring to the thermal self noise of the resistors. If you use TX-2575s this will be inconsequential in any system context I am aware of, and in attenuating the signal, you will reduce the output quiescent noise of the DAC itself by whatever amount you attenuate, a 4 dB pad would reduce the DACs noise floor by 4 dB (but S/N of course will remain the same).

 

One "problem" of digital volume controls is that they do not reduce the noise floor as they reduce the level of the signal, so if one has to use a lot of digital attenuation, the S/N ratio suffers. If the amplifier requires 2 V for full output, then best to attenuate the DAC signal so it is close to 2 V at 0 dB, then use the digital volume from there. If you consider some of the "hot" DACs out there, with 4-6 V outputs at 0 dB, using just a digital control will result in a substantially higher noise floor, better to pad the output down by 6 dB, reducing the noise floor by 6 dB as well, and actually increasing the S/N ratio in the process.

 

Ear: the MH stuff has onboard analog chip attenuators, so by adjusting the overall level in software, all you are doing is turning down the volume anyway, it is not changing the gain of the circuit. This discussion is really about matching system levels to allow for use of a digital volume control without using too much (digital) attenuation, sinch MH relies on an analog volume control, it is a different thing

[Miska]

I guess you are referring to the thermal self noise of the resistors. If you use TX-2575s this will be inconsequential in any system context I am aware of, and in attenuating the signal, you will reduce the output quiescent noise of the DAC itself by whatever amount you attenuate, a 4 dB pad would reduce the DACs noise floor by 4 dB (but S/N of course will remain the same).

 

For example noise of Sabre's output is roughly noise of 600 ohm resistor. Resistor noise depends directly on the resistance value[1]. In most cases for example potentiometers are much higher resistance than 600 ohm, so they can only increase output noise of the DAC.

 

[1] Johnson

 

 

One "problem" of digital volume controls is that they do not reduce the noise floor as they reduce the level of the signal, so if one has to use a lot of digital attenuation, the S/N ratio suffers.

 

Well, if the noise floor is already lower than what you can get from the analog stages after the DAC, then DAC's noise floor staying the same as attenuation is increased is not a problem, since the output noise is in any case dominated by the analog electronics, not the digital. 32-bit DAC has digital noise floor around -190 dBFS, even without noise shaping. Even with 24-bit's -144 dBFS you are going to have hard time matching it with analog electronics...

[joeinid]

I'm reading great things about the MSB Diamond dac, but it can get pretty pricy with all the options. I have a friend trying to buy one and I am looking as well.

[Miska]

For 24-bit, 119 nV RMS re 2 V and for 32-bit 465 pV RMS re 2 V.

[barrows]

Miska: here is what I mean:

 

"Well, if the noise floor is already lower than what you can get from the analog stages after the DAC, then DAC's noise floor staying the same as attenuation is increased is not a problem, since the output noise is in any case dominated by the analog electronics, not the digital. 32-bit DAC has digital noise floor around -190 dBFS, even without noise shaping. Even with 24-bit's -144 dBFS you are going to have hard time matching it with analog electronics... "

 

DAC chip w digital volume-DAC I/V output stage-Amplifier....... vs:

 

DAC chip-DAC analog output stage-preamp w analog volume-Amplifier

 

In the first case, reducing the volume will not reduce the noise floor of the following I/V output stage, so the S/N of the DAC suffers as the digital volume is reduced.

In the second case, reducing the analog volume control of the preamp does reduce both the signal, and the noise equally, so the S/N ratio of the DAC is preserved.

 

In any case, most of this is academic as long as one ahs a decent gain match between DAC and amplifier. With my DIY DAC I just adjust the circuit so I never use more than 25 dB of digital attenuation, no problem doing this, and having no preamp and no analog attenuator is a benefit. But if someon is using a DAC with 6 V out, and a sensitive amp/speaker combo, they may end up using 70 dB of digital attenuation-this could be a problem and they woudl be better off putting a pad at the input of the amp.

 

Also. Noise from different stages is cumulative, so being below the noise floor of a subsequent stage is not enough.

[sdolezalek]

At my age...you'd opt for the Porsche.

 

Yeah! Especially when it comes with a Burmester system inside....

 

I guess that means no DCS in the house. ;-(

[The Computer Audiophile]

Bump ......

 

Hi Chris,

 

Not trying to stir the pot but any updates on the new review? Thanks!

 

Joe

 

Hi Joe - The review is finished and has been sent to the manufacturer for an accuracy check. I hope they can get it back to me before the weekend so I can publish it sooner :~)

[Miska]

Barrows,

 

Since the pad likely has noise 10 - 100x of the IV stage, adding it will make sense for large attenuations, otherwise the additive effect will just make things worse.

 

With 70 dB attenuation there is likely too much gain in the following gain stages and the SNR would be poor anyway because pad's thermal noise would get amplified so much.

 

In any case, if DAC+IV has noise of 400 nV RMS, adding an analogapad that has self noise of 1000 nV RMS doesn't make much sense to me in anyway...

[joeinid]

Hi Joe - The review is finished and has been sent to the manufacturer for an accuracy check. I hope they can get it back to me before the weekend so I can publish it sooner :~)

 

Holy Cow Chris! What personal service. I am honored. Thank you! There are many CA members anxiously waiting to read it. Thanks again for all the great info. Looking forward to it.

 

Joe

[AudioExplorations]

ear- just checked the hilo manual and you are right. This is the first time I have seen this outside of weiss and Barrow's DIY DAC. Any others you are aware of?

 

To answer my own question (DAC's with digital volume and adjustable circuit gain) Wadia also includes this. The list is now Weiss, Lynx, Wadia, and DIY.

[jtwrace]

Barrows

 

Question for you...

 

Have you had an opportunity to compare your Buffalo III to any of thses dacs that we're talking about?

[barrows]

Barrows

 

Question for you...

 

Have you had an opportunity to compare your Buffalo III to any of thses dacs that we're talking about?

 

Not directly. Among the best DACs which I have heard are: Weiss Medea (not Plus), dCS Debussy, and full Scarlatti stack, and then everything else. By memory, I would say that my Buffalo build is performing at least close to those DACs, if not on par. Every DAC sounds different though, but determining "better" is going to be system dependent.

With DIY, it is also a process: my Buffalo build is not stock Twisted Pear, my power supplies are different, and both my analog stages are modded (different filter coefficients) and different parts than what Twisted Pear Audio supplies. I can tune the sound of the DAC to suit my system perfectly, and I keep improving things, so the resulting sound is rarely fixed, but is constantly evolving. Also, once you start DIYing, you find every passive part makes a difference, things like choice of XLR and output wiring matter, layout of the boards inside of the case matters, wiring layout can be critical, transformer choice, etc, etc. All I can say is that if you have some time, and are not afraid to learn, and can solder well, you can achieve superb perfromance for a fraction of the expense of top tier DACs, but you will spend 1000s of hours getting there...

 

One other thing, the computer interface is critical. For best performance you want a really good async USB interface. I am using an SOtM (XMOS based) which is modified with Crystek CCHD oscillators-I power the each oscillator from its own Twisted Pear Trident shunt regulator, and these shunts are fed from a separate pre reg, with its own transformer, then the SOtM USB board is powered from its own custom shunt pre reg and transformer winding. There is another really good XMOS based USB receiver, the Wave IO, made by diyaudio member "Lucien", which would be another good choice. Twisted Pear will also, someday, release their own async USB receiver (also XMOS), which I expect to be something special-they may end up supporting DSD as well (I have my fingers crossed for that), as it is taking them a long time to offer this product...