Any benefit to dual chip DACs?

[watchnerd]

I guess they simply do not believe it's any better to use two?

My point was the opposite -- consumers usually think DAC chips are more expensive than they are. So adding a 2nd chip can allow the maker to raise the price of the device quite a bit without raising the costs to build very much.

 

[invaderzim]

Same with RME ADI-2 Dac with its AKM 4490

I think it also appeals to consumers, 'dual chip' waowww

Yep, more is better. More watts, more megapixels, more horsepower, more numbers always looks better. And in all of those things the way they are implemented can make them much worse than the same item with lower numbers.
I work with a guy that partially judges speakers by how high their watt rating is and nothing I can say can convince him that as long as it is high enough for his amp then that number is perfect.

 

[wiggum]

There is no practical benefit for audio. It serves as a prescription to those suffering from audiophilitis disease.

 

[watchnerd]

Yep, more is better. More watts, more megapixels, more horsepower, more numbers always looks better. And in all of those things the way they are implemented can make them much worse than the same item with lower numbers.
I work with a guy that partially judges speakers by how high their watt rating is and nothing I can say can convince him that as long as it is high enough for his amp then that number is perfect.

"Simply the biggest, baddest tower on the market, the XLS-215 features two cast frame high excursion 15-inch woofers and delivers earth-shaking bass down to 36 Hz. With guaranteed "eviction notice" performance, the XLS-215 also features a 6.5-inch mid-range speaker and a 1-inch soft dome tweeter coupled to a proprietary waveguide

Power Capacity: 500 watts (Peak )"

[of course, no mention of distortion, excursion, etc]

http://www.cerwinvega.com/home-audio/floorstanding-speakers/sl-15.html

 

[pos]

I work with a guy that partially judges speakers by how high their watt rating

Sounds like the perfect client for a 4500W Devialet Phantom

 

[andymok]

Wait, are we talking about the the advantages of "1 chip-1 channel", or "1/multi DAC-1 channel"?

Coz 1 chip can have many DACs in it.

As to multi-DAC-1 channel playback, you can lower the noise floor by 3dB, by doubling no. of DAC
(1/2/4/8). You do not see Pro gears implement and take this advantage however, probably due to masking the real noise floor in recordings if doing so. Correct me if I'm wrong.

 

[Dialectic]

"Simply the biggest, baddest tower on the market, the XLS-215 features two cast frame high excursion 15-inch woofers and delivers earth-shaking bass down to 36 Hz. With guaranteed "eviction notice" performance, the XLS-215 also features a 6.5-inch mid-range speaker and a 1-inch soft dome tweeter coupled to a proprietary waveguide

Power Capacity: 500 watts (Peak )"

[of course, no mention of distortion, excursion, etc]

http://www.cerwinvega.com/home-audio/floorstanding-speakers/sl-15.html

As vulgar as the marketing is, these probably are not that bad, especially at their price point, though I don't know exactly what they cost. If they can provide some dynamic range at a sub-$10K price tag, they have already beaten most audiophile speakers.

 

[watchnerd]

As vulgar as the marketing is, these probably are not that bad, especially at their price point, though I don't know exactly what they cost. If they can provide some dynamic range at a sub-$10K price tag, they have already beaten most audiophile speakers.

$1500/pair

 

[Killingbeans]

There are a lot of higher end DACs that use separate DAC chips for each channel. I was wondering if this forum had any opinions as to the benefits (or potential issues) with that setup over a single chip operating on both channels?

One potential issue would be that it's practically impossible to produce DAC chips with identical output impedances. The chips are designed to be internally matched, but impedances of individual chips are never the same. I've seen it mentioned as something in a dual chip implementation that could cause problems outweighing the benefits, but I'm not sure whether it's actually so?

What makes these a bit different is that each one of those multibit chips are quite expensive - about $400 for 4.

It's been said a million times before. An expensive component is pointless without proper implementation, which again is not the same as throwing other expensive components at it. The amount of expensive parts in a system tells you nothing about how well it measures

 

[Draculr]

Lots of great responses so far. I'm just wondering as one example if a Schiit Modi 3 would sound much different to a Gungnir or if it's just marketing fluff in the best case (or the Gungnir having additional issues over the Modi 3 in the worst case). Not singling Schiit out just using them as an example in this case, the upcoming Topping D70 with dual AK4497 chips could also be compared to the D30/D50.

I mean, they go through all sorts of extra engineering to double up on everything that's in the Modi/Bifrost. It's a lot of effort, I just wonder if it is actually done for the purpose of better sound or just because they know there's a market for that.

 

[wiggum]

I mean, they go through all sorts of extra engineering to double up on everything that's in the Modi/Bifrost. It's a lot of effort, I just wonder if it is actually done for the purpose of better sound or just because they know there's a market for that.

It is purely for marketing. You need to remember only one thing. THD of headphones/speakers average around 0.5% in the midrange. So,reducing the THD,IMD of amplifier and DAC from 0.001% to 0.0005% or even lower is of no real use since the distortion of the speaker dominates. This is one of the reasons even poor amplifier/DACs(by modern standards) are being sold & discussed. Nobody can pick them apart in a blind test.

 

[Sangram]

There are a lot of higher end DACs that use separate DAC chips for each channel. I was wondering if this forum had any opinions as to the benefits (or potential issues) with that setup over a single chip operating on both channels?

At a purely technical level you get 3dB increase in S/N ratio (increase in resolution) for every doubling of output stages (internal to the DAC chip). This is because there is more current available to drive the output, so you can drive the outputs to higher voltage. S/N ratio is a ratio between two voltages so you can either push the floor lower or the top higher. Both result in an increase in S/N.

High output voltages are one way to get excellent S/N ratio however it's a little useless because you will never be able to use the entire range of output available. Most units quote S/N ratio at 0dB FS, which for a DAC putting out more than ~4V is useless. For apples to apples comparison you need to look at noise vs say 2Vrms output signal.

Even in cases where you get twice the output current and are able to achieve higher resolution, the analog circuitry is the limiting factor. Real resolution rarely exceeds 20 bits, and 21-bit is considered state of the art. Thermal noise in our environment is about -160dB, and once current starts flowing the -144dB floor for true 24-bit resolution becomes impossible to reach. In a purely digital state it is possible, but not once you convert it to analog.

 

[DonH56]

At a purely technical level you get 3dB increase in S/N ratio (increase in resolution) for every doubling of output stages (internal to the DAC chip).

That is only true for some types of noise, depends upon the circuit topology, and unless the number of bits is changed does not improve the quantization noise. Not saying you are wrong, just that other things also play into it. As you noted -- the stages after the DAC itself must handle the extra current, and e.g. if the output signal amplitude is not changed then there is less gain to the final outputs so you might not even get the 3 dB you'd expect.

A point unrelated to circuit noise is the fact that, when two or more DAC outputs are combined, there are very stringent requirements on timing and level matching between (among) the DACs to prevent the combination from raising the distortion floor. Interleaving errors can be a major PITA.

 

[jsrtheta]

Accuphase has long put lots of DAC chips in its devices (often four or eight, which operate in parallel), claiming that it reduces distortion. If I recall correctly, Accuphase calls this their MDS technology. It may reduce distortion in a way that Amir could measure, but is such a reduction audible?

In light of the superb performance of some single-chip DACs (notably the RME ADI-2 DAC), I have my doubts.

IIRC, Accuphase's top of the line at one point had 16 PCM63P-Ks per channel. I have no idea why.

 

[FireLion]

So DAC's do not need two to be balanced? I am aware a standalone DAC chip can output true balanced although I'd like some clarification?

 

[Blumlein 88]

So DAC's do not need two to be balanced? I am aware a standalone DAC chip can output true balanced although I'd like some clarification?

Variations are probably covered earlier in the thread.

Some DAC chips put out a balanced signal so one per channel will work. Some put out more than that and a single chip can put out both balanced channels. There may be some larger ones, but some can put out 8 balanced channels. So one chip could be used for 8 channels or could be paralleled for 2 channels with each channel having somewhat lower noise. In fact there are chips that are 8 balanced channels of ADC and DAC on the same single chip. And those can give fairly good performance.

 

[solderdude]

https://www.audiosciencereview.com/forum/index.php?threads/paralleling-dacs.2212/

I did notice that the chips with the IMD hump seem to benefit from paralleling if I remember correctly.

For stereo separation it is utter nonsense for a bunch of reasons.