Audio-Gd Reference 10 - Page 6

And the Audio-Gd  Reference 10  has a MUCH THICKER  front panel.... and I am a sucker for thick front panels!    

Seriously, there is an assumption in high-end circles that high current, linear DC supplies  give better sound than switching supplies; I think the "reasons given"  are twofold:

• Switching supplies may contaminate other circuits with their RF noise, either through their DC output or through capacitive or inductive coupling.
• Linear regulators offer lower impedance outputs, especially at higher frequencies, which provides better sourcing AND SINKING of currents which D/A converters and analog stages draw.

If you look at designs like the Conrad-Johnson preamps, they understand that power supplies are more than just a block on a diagram from which current is drawn when needed; the power supply forms an integral part of the audio circuit, supplying as well as sinking currents originating in the active devices of the preamp. Therefore, C-J  has turned to using all film and even teflon type capacitors in their preamp power supplies to squeeze that 'last margin of audio goodness' out of the signal. (That's right, some C-J preamps have NO electrolytics in the B+ supply, only great big film caps.)  Audio-Gd follows a similar idea with their overkill class-A regulators- they are taking steps to keep power supply related factors from hindering audio performance.  The power supply doesn't have as much impact on the sound as the parts directly 'in series' with the audio signal, but they DO have an impact, so once all the "direct audio signal" designs are as linear as possible (DAC and analog stages) the next area to improve is the power supplies- to bring the sound "that much closer" to sonic nirvana. We all know that the last 5~10% of sonic improvement often accounts for 75% of the cost.

R-Core transformers, used in the more costly Audio-Gd products,  do usually measure lower in line noise transmitted through them than other types of transformers. (R-core transformers have lower primary-to-secondary capacitance than other designs, hence less high frequency hash is coupled into the power supply from the line.) I don't think switching supplies are particularly good at suppressing RF and other noise coming in on the power line, especially common mode noise.  They DO filter it out to a degree, but I think that R-core transformers followed by linear supplies would do a better job.

Also, Audio-Gd doesn't use IC op-amps in their analog stages.  Here, the thinking is:

• You can hand-select and hand-match parts in a discrete gain stage, not possible in an integrated circuit
• IC op-amps rely on negative feedback, sometimes quite heavily, and there is an opinion (some would say a "fashion") that negative feedback can hamper the dynamic operation of audio circuits.  Rather than use a 'good' circuit made to measure "great' by use of negative feedback, the trend is to make circuits that are very linear, wideband and have low output impedance to begin with, and eschew negative feedback as much as possible. Often this requires hand-selecting semiconductors.  Audio-Gd doesn't use negative feedback in some instances where it's been common practice to use it.

Using ACSS, and placing the volume control in the ACSS circuit, should have an advantage. I think this is because solid state circuits offer better linearity in current gain than in voltage gain, though my understanding of this is incomplete- I really don't know the topology of the ACSS designs and I am shooting from the hip when I try to understand the advantages of current-domain designs like ACSS.  I know using ACSS (or CAST as used by Krell) offers advantages in signal transmission over cables- cables between a CAST  output DAC and a CAST input preamp should have nearly zero impact on sonics - but we have no ACSS (or CAST) cables in the Reference 10, just internal signal wiring. The Reference 10 does have ACSS outputs for use with Audio-Gd power amps and preamps, but the advantages of ACSS when used in a stand-alone DAC must lie elsewhere than in terms of signal transmission.  I know Mark Levinson also used volume controls in current-gain stages in some of their well-regarded preamps. I would welcome more explanation and discussion of ACSS and the similar CAST concepts from any engineers out there who are familiar with these things! 

Does all this engineering and philosophy result in better sound?  ONLY LISTENING COMPARISONS CAN ANSWER THAT QUESTION!!!!  I would love to be able to do some blind A/B comparisons of these high-end DACs. Unfortunately, although I might buy an Audio-Gd Reference 10 sometime soon -  I do not have other similarly-priced DACs, like the Fostex or Jade to compare. 

OK, so I don't KNOW if the "Audio-Gd Way" results in better sound, but I am going to bet it doesn't result in WORSE sound, and I LIKE the implementation in the Audio-Gd products. This is a question of engineering aesthetics, not of sonics. Yeah  I DO listen to this gear, but aside from listening pleasure, I get satisfaction from the 'objectness' and the design philosophy implementation because I am a technical guy and just like technical things. I really get a rush from opening an Audio-Gd product and looking at what's going on in there.  This is maybe 40% of the enjoyment of the gear for me.  It's like looking at a painting, in a way.  It's a kind of art.

Quote:

Originally Posted by Telstar 

 

They keep AC noise out but they create their own. So in the end a SOTA switching PSU will cost about the same as a linear one, while being much more complex to build. Only advantage is efficiency.

 

That depends on how is the input stage of the amplifier included in the Ref.10. Current output (ACSS/CAST) can be used successfully to drive a differential stage. 99.9% amplifiers are made of at least two stages.

The volume control has nothing to do with ACSS.

 

>>The volume control has nothing to do with ACSS.<<

   Here's a quote from the Audio-Gd description of the Ref 10 - While the volume control is not in the ACSS circuit,  I don't think "nothing to do with ACSS" is quite accurate. It's at the output side of ACSS stage- and, according to Audio-Gd the volume control is being fed a current signal; it's operating on a current signal.

 in the Reference 10 , we applied the I/V conversion volume control , the volume control just  a variable passive I/V conversion place at the ACSS amp output ,where the output is the current signal, not the voltage signal. The volume control is I/V conversion, and the volume depends on the current (I) conversion to the voltage (V). (Like R-2R D/A chips output passive I/V conversion)  It can keep the signal frequency band flat and not lost any detail. It can't degrade on sound quality in every volume level . After the volume , there are four groups  diamond non-feedback buffer output stages offer the very low output impedance  .           Mark Levinson also knows that current volume control has great benefits, so in their Top End preamp NO.32 volume control, they use many components to change the signal to current signal (I), then implement the volume control , and then change back to voltage signal (V) again. But Reference 10 is simply a more complete work in terms of the current signal, and technically, it is superior to conventional technology. The less conversion bring much neutral sound .

And here's some more on ACSS volume control, from Audio-Gd's description of the NFB-10ES

ACSS technology, which is very different from conventional technology. In conventional technology the gain is fixed, like 3DB or 6DB.
      But in ACSS technology the gain changes with volume, and when you turn off the volume, the gain is -80DB or -100DB (in conventional technology it can't work lower than 0DB).
      While you are listening in normal volume, the gain may be around -50DB, or between -80 to +12DB.
      The volume control is also different from conventional technology. Conventional technology volume control is placed in the input, and attenuates the signal.
      But in ACSS, it is placed at the ACSS modules output, where the output is the current signal, not the voltage signal. The volume control is I/V conversion, and the volume depends on the current (I) conversion to the voltage (V). (Like D/A chips output I/V conversion)
      As you can see, ACSS circuits have no gain, as it is a completely different circuit.
              

Quote:

Originally Posted by Telstar 

 

They keep AC noise out but they create their own. So in the end a SOTA switching PSU will cost about the same as a linear one, while being much more complex to build. Only advantage is efficiency.

 

 

 

I can't find any information about R-Core transformers creating noise.  What I find is that they have very low flux leakage, meaning their EMI performance would be superior to other types.  Rather than being cited as a source of noise as you seem to imply, they are touted as being a low noise transformer. See http://www.custommag.com/products/r-core.shtml , http://www.kitamura-kiden.co.jp/english/products_e.html ,   http://www.dibao-transformer.com/R-core-transformer/R-core-transformer_productlist4_1.htm  etc

Can you point me to some info on noise created by R-core transformers?  I'm not finding any.

Quote:

Originally Posted by Telstar 

 

 

I was talking about switching power supplies.

Ah, I see.

So, I guess no one has one of these DACs yet.

Typically, as you "up the line" from Audio-Gd's lower cost items to the more expensive ones, you see the following:

• The more expensive gear has more individual power supplies, for example, separate supplies for left and right channels instead of a combined supply.
• The more expensive gear uses "Class A" voltage regulators in the supplies, meaning they are drawing max current at idle, etc. Kingwa feels these sound best due to their quick, very linear response to current sink or supply demands.
• The more expensive stuff uses R-Core power transformers instead of torroids or other designs.
• The more expensive stuff uses relay-actuated discrete step circuits for volume control, the less expensive stuff uses pots.
• The more expensive stuff tends to have heavier chassis construction with multiple internal shielded compartments.
• The Reference 10 uses a considerably more expensive DAC, the PCM1704 UK.  The R2R ladder type DAC is not only more costly as a part, it costs somewhat more to implement.

As has been my theme all though this discussion of he Reference 10, while all this engineering is interesting and even somewhat heroic, Inquiring Minds Want To Know:  DOES IT SOUND BETTER?

I'm looking forward to someone doing a comparison.  If I sell enough of the junk- er, I mean, CLASSIC gear I have laying around, I may even get one and do some A/B listening. (Right now selling a nice old Shure V15 TYPE III and just sold a pair of Bybee "purifiers" that I took out of a phono preamp I got. I prefer things that are impure.  More sales to come- just have to get on the stick and list the stuff.  It will eventually mount up to enough to buy either a Ref 10  DAC or a used Subaru.)