Next Level of Performance DACs.

[themystical]

The highest performance DAC chips on the market currently are the ES9038Pro and the Ak4499eq. The current implementations of these chips are achieving a SINAD of 120/121dB, irrespective of whether a single DAC chip is used or a dual mono arrangement with two chips. What is the next level of implementations that is likely to improve these figures to 124/125dB. Is anybody aware of any DACs planned or in development that might achieve this next level? I’m pinning my hopes on the Ackodac which appears to have the best implementation of the AK4499 I have seen so far.
https://www.ackolabs.com/dac.html
http://www.ackolabs.com/
Anybody else aware of any other hopefuls.?

 

[Wes]

If I thought there was an psychoacoustic benefit to a "better" DAC, I would toss out my Oppo and buy an Okto.

Based on measurements here, and assuming they cover the multi-factorial space defining SQ, that extra 5 dB is likely a will o' the wisp.

 

[themystical]

If I thought there was an psychoacoustic benefit to a "better" DAC, I would toss out my Oppo and buy an Okto.

Based on measurements here, and assuming they cover the multi-factorial space defining SQ, that extra 5 dB is likely a will o' the wisp.

I realise that this might be an academic issue as far as audibility is concerned but there is a strange delight in the progress of technology and engineering prowess?

 

[BDWoody]

I realise that this might be an academic issue as far as audibility is concerned but there is a strange delight in the progress of technology and engineering prowess?

I agree with that...but I hope it doesn't take time away from someone building a decent AVR/AVP! Let's get a few of those up into the 3 digit range and I'll be a happy camper...

It will be interesting to see what the charts look like in another year...

 

[Spocko]

If you are scoring "highest performance DAC" purely on SINAD you may be chasing a fool's errand here. It appears that the only way Chi-Fi companies can separate themselves is to implement measurements beyond audible hearing as their number one defining quality. At some point, SINAD measurements and distortion measurements will reach pointless levels of performance. It's like arguing my car engine is better than yours because my engine can handle 500 MPH for 12 hours before blowing up while yours can only do 450 MPH, but both of our cars are minivans for driving our kids around. We have reached this point for DAC chips. Established companies like RME and Hegel (whose clientele are both discriminating and demanding) have publicly stated that these new DACs simply do not sound any better than the older DACs, despite amazing SINAD measurements - the chip upgrade is not expensive, and yet, they've chose older chips. And they're not alone. So, to call one DAC higher performing than another makes little sense when these digital differences are only measurable on special instruments and the analog section unrelated to the chip adds more noise than the differences between the DAC!

 

[Spocko]

I realise that this might be an academic issue as far as audibility is concerned but there is a strange delight in the progress of technology and engineering prowess?

Not when commercialization brings no money to the table. Audio companies (other than Chi Fi) are electing not to buy these amazing feats of engineering that does nothing for sound quality.

 

[Tks]

I would imagine constantly aiming higher is so when implementations occur in various products, there is headroom for things like DSP and other factors to be less of a detriment.

What I mean is, if you start off with a DAC design that is 120dB SINAD or w/e, it pays to have such level of performance if for instance for whatever reason the AVR or some specialized device you're making has tons of distortion that gets fed in to the processing, can now have less of an impact on the overall measurement in the end.

Personally speaking though, I think DAC's have conquered ages ago performance-wise, with respect to noise/distortion. Now all I care about is seeing proper measurements elsewhere, and featuresets. By that I mean, I want the eventual device to be free of jitter, also one with proper filters (ESS does these better than AKM even with there 4499 out now), and from manufacturers, I would like to see them compete on a field that RME has created (insane creature comforts and tons of options). Performance regressions should be kept to a minimum while doing this, and having these newer DAC chips, allows any regression to be less felt (like if for instance the FPGA implementation you build around your eventual DAC degrades performance by like 5 or 10db, that's much less of an issue than if the same DAC was using a chip that performs at best something like 80db SINAD).

I realise that this might be an academic issue as far as audibility is concerned but there is a strange delight in the progress of technology and engineering prowess?

Yes to this, I don't mind rewarding companies with my money for pursuits of pushing engineering ahead. I literally wouldn't understand the point of life (or a fulfilling felt one) if everyone said "good enough" to everything in their lives.

I only wish I had actual knowledge where I could technically appreciate these advances from a specific design view. Seeing 120dB SINAD is great, but seeing, or knowing what the designer actually did to delineate themselves from others to get there would taste sweeter. Only problem is, I'm uneducated in many of the concepts involved in electronic engineering, and am slowly learning out of just genuine interest and from the insanely well experienced people on this forum.

 

[pozz]

Hegel

Disagree at least for DACs. Noise floor of -150dB? Not even clear what they are referring to.

Otherwise I'm with you and you comments. I'd love to see their rackmount power amps tested.

 

[solderdude]

There is a practical limit when looking at noise floors. It can only be increased further by substantially increasing 'line levels'.
Then consider an amp will not reach those noise levels either.
Next, consider there isn't a single recording on this planet, where recordings were done acoustically that will ever reach levels such as DACs, amplifiers and transducers will ever be able to reach in real world home audio applications.

Sometimes enough is enough. Manufacturers love to push the envelope so they can sell more or show theirs is 'longer' than those from the competition. That's where the SINAD race is fueled from, further aided by SOTA measuring equipement that is able to show it.

 

[Blumlein 88]

Don't use chip DACs. Here is the product you want.
https://www.mola-mola.nl/tambaqui.php

It ain't cheap.

Top of the SINAD list on ASR currently.

To my knowledge, the best measuring audio DAC available. Anyone know of better?

 

[JohnYang1997]

Actually we already have dacs of 124-125db. The numbers are limited because APx555's performance is too high at 4V input.

 

[Objectivist01]

If I thought there was an psychoacoustic benefit to a "better" DAC, I would toss out my Oppo and buy an Okto.

Based on measurements here, and assuming they cover the multi-factorial space defining SQ, that extra 5 dB is likely a will o' the wisp.

I think at this point better numbers is more of a psychological benefit !

 

[Pluto]

Moving on, great leaps in performance are unlikely for the simple reason that we are now close to the theoretical limits imposed by the random movement of electrons in real world materials. The only (vaguely) practical way of reducing so-called Johnson noise is by cooling (as this noise is, more or less, proportional to absolute temperature) which is why the head amplifiers of radio telescopes are cooled by liquid nitrogen and stuff.

The interfacing will probably continue to improve (a little) as techniques improve but, in extremis, this sort of thing is as much dependent upon PCB layout and quality as on the raw technology. My prediction is that the next generation will largely offer the functionality of more powerful DSP before the actual conversion.

The sort of product that would probably sell like hot cakes, given the chance, is something like a Dolby Atmos decoder with digital in and however many volume-controlled analogue outputs. But such products are unlikely as the patent holders are more concerned with keeping a grip on their IP than allowing their systems to become ubiquitous. In my view the answer is to put the playback decoding technology into public (or, say, a $1) license but seek royalty income on media produced using the system.

 

[themystical]

It will be interesting to see what the charts look like in another year...

Any ideas as to which products these might be?

 

[themystical]

There is a practical limit when looking at noise floors. It can only be increased further by substantially increasing 'line levels'.
Then consider an amp will not reach those noise levels either.
Next, consider there isn't a single recording on this planet, where recordings were done acoustically that will ever reach levels such as DACs, amplifiers and transducers will ever be able to reach in real world home audio applications.

Sometimes enough is enough. Manufacturers love to push the envelope so they can sell more or show theirs is 'longer' than those from the competition. That's where the SINAD race is fueled from, further aided by SOTA measuring equipement that is able to show it.

Do you think the future trends will focus on price reduction or optimisation instead of improvements in technical performance?

 

[themystical]

Actually we already have dacs of 124-125db. The numbers are limited because APx555's performance is too high at 4V input.

It would be interesting to know whether any test equipment manufacturer is working on equipment that will fuel the next set of improvements. I guess the two go hand in hand so if we are currently limited by the test equipment, until improved test equipment appears, it’s a futile endeavour?

 

[solderdude]

Do you think the future trends will focus on price reduction or optimisation instead of improvements in technical performance?

I can't predict the future but development costs money. I would say manufacturers will charge more for future high-end chips and perhaps offer lesser performing (but still improved over previous versions) perhaps a tiny bit cheaper.

It would be interesting to know whether any test equipment manufacturer is working on equipment that will fuel the next set of improvements. I guess the two go hand in hand so if we are currently limited by the test equipment, until improved test equipment appears, it’s a futile endeavour?

There is equipment that can measure lower noise than the AP555. The AP555 is an excellent top universal audio test device but not specifically dedicated to the lowest possible noise floor. There is always a trade-off somewhere when test devices can test a lot.

 

[themystical]

Established companies like RME and Hegel (whose clientele are both discriminating and demanding) have publicly stated that these new DACs simply do not sound any better than the older DACs, despite amazing SINAD measurements - the chip upgrade is not expensive, and yet, they've chose older chips. !

Are these so called established (western?) companies capable of engineering what chi-fi companies are achieving? If so, are they deliberately toning down their designs to achieve a worse engineering outcome. Could it be that Chi-fi engineering has now surpassed western engineering with SOA products at very low costs

 

[solderdude]

Choosing the used chips has to do with availability, price, discounts, and length of development path before newer chips hit the market.
It has nothing to do with sound quality or other aspects.

 

[Jimbob54]

If the quest for higher SINAD (or other benchmark) stops, what will be the next development? Its either features or performance that will incentivise someone to buy a new DAC. Multichannel may well be a next big thing but I suspect its a way off being a driver for mass sales .We have done the "balanced" thing. Suspect MQA is stagnating

I suspect the D90 V2 will be a spec bump, but no idea what the D100 will have that the D90 doesnt.