New DAC Chipset from AKM

[samsa]

This Press Release from AKM announces a new flagship DAC chipset. The AK4499 is still the flagship DAC chip But now they're talking about splitting the functionality of a ΔΣ DAC between two chips: the AK4191 which contains the digital reconstruction filters and the ΔΣ modulator, and the AK4498, which contains the analog filter.

They say:

By using two ICs, AKM has minimized the effects of digital noise within the analog output, resulting in a perceived improvement of the ratio of signal to noise. This improvement, while not easily quantifiable via traditional measurement techniques, is easily experienced during controlled listening tests.

(emphasis mine).There aren't (AFAIK) any commercial products yet. But perhaps @amirm had better gear up with some non-traditional measurements.

 

[Max]

The AK4498's specifications look like more to AK4497 than to AK4499 (not to say it is in fact exactly the AK4497's ones). Not a flagship but perhaps a new strategy for a better/quicker use by electronic designers : it is easier to work with a new "small simple" chip than to "train a new big beast" chip.

 

[Max]

Thus, IMHO, the target is clearly the electronic designers community ... but AK can't stop targeting "audiophiles". As they say :

"This improvement, while not easily quantifiable via traditional measurement techniques, is easily experienced during controlled listening tests."

 

[mansr]

AKM has achieved full audiophool, a journey they started when they dreamed up the Velvet Sound name.

 

[Jimbob54]

So, you give your R&D dept a budget and a brief. They come back with 2 ickle chips to replace one big chip (which I guess increases manuf. costs?) and you ask, "what's the improvement, folks?"

And they bust out the Diana Krall CD.. "weeeeelllllll"

 

[taner]

The reasoning is kind of in contradiction with measurement favoring people. Why separate chips if there is not a measurable difference? Because it sounds better!

Things will become more interesting in DAC business.

 

[samsa]

So, you give your R&D dept a budget and a brief. They come back with 2 ickle chips to replace one big chip (which I guess increases manuf. costs?) and you ask, "what's the improvement, folks?"

I would have guessed that these are the first in a series of mix-'n-match DAC chipsets from AKM. If you have (say) 3 different ΔΣ modulator chips and 3 different analog filter chips, you can offer DAC designers 9 different combos while only having to manufacture 6 different chips (with larger production runs for each). And the yield is probably higher on the "simpler" chips. Hence cheaper.

The downside is that two chips take up more space and consume more power than a single "integrated" design. That would be a killer in the mobile and embedded markets, but probably not in the audiophile DAC market.

 

[CedarWind108]

AK4499 and AK4493 are almost getting old by today’s standards. I’m impatiently waiting on a ES9048

 

[RayDunzl]

I’m impatiently waiting on a ES9048

I wonder when we will get quantum DACs.

How many qubits will be "enough"?

 

[Tks]

You know for a fact the engineers working on this were just shaking their heads when that claim of: "while not easily quantifiable via traditional measurement techniques, is easily experienced during controlled listening tests.".

Let's see the tests -.-

Oh while you're at it, get your filters to properly attenuate to 22.05 with proper brickwall(even ESS and their IMD hump dumbasses got the filters correct). Would've thought they'd have nailed that part down at least by the 4499's release, instead we get the same old filters. You want to appease audiophiles? Give us things like that, that can at least be measured even if not heard. But don't give us "heard, but not measurable", and then not pony up the supposed controlled listening tests you allude to (which I'd wager anything aren't ABX, and if they were, then these tests don't exist beyond the declaratory statement you assert I would claim baselessly).

 

[raif71]

I'm sure AKM chip designers know what they are talking about, right?... no need the emojis to downplay what they say. Go go go...AKM...I just love your velvety sound

 

[Tks]

Problem is, we don't know what they're talking about. (Or I should say I, perhaps others do).

 

[sinski]

Maybe separate powers supplies for each chip can have positive effect ?

 

[mansr]

Separating the digital circuitry from the analogue parts can certainly help reduce noise in the latter. There may also be some advantage to using different silicon processes for the two chips, though that could be had with two dies in a single package too.

 

[HpW]

... and reducing/skipping channel cross talk using two analog chips... as seen in RME ADI-2 Pro xx using two DAC chips (connection at the front)

 

[pierre]

If you asked eng. from AKG they would likely make a brief around: it is already inaudible and the current DACs are perfect for what you want to do but to justify our salaries we have improve stuff a tiny bit such that marketing can talk about something. Then it goes to marketing department that remove all of the above and rewrite it. That's life and that's reasonable from a business PoV.

 

[frogmeat69]

I'm sure AKM chip designers know what they are talking about, right?... no need the emojis to downplay what they say. Go go go...AKM...I just love your velvety sound

LOL, If AKM were a cake,

 

[renaudrenaud]

I wonder when we will get quantum DACs.

How many qubits will be "enough"?

22k ?

 

[despoiler]

You know for a fact the engineers working on this were just shaking their heads when that claim of: "while not easily quantifiable via traditional measurement techniques, is easily experienced during controlled listening tests.".

Let's see the tests -.-

Oh while you're at it, get your filters to properly attenuate to 22.05 with proper brickwall(even ESS and their IMD hump dumbasses got the filters correct). Would've thought they'd have nailed that part down at least by the 4499's release, instead we get the same old filters. You want to appease audiophiles? Give us things like that, that can at least be measured even if not heard. But don't give us "heard, but not measurable", and then not pony up the supposed controlled listening tests you allude to (which I'd wager anything aren't ABX, and if they were, then these tests don't exist beyond the declaratory statement you assert I would claim baselessly).

Do you honestly think that the engineers at AKM missed the chapters on Nyquist, yet still got hired, or that a 24k stopband is a conscious engineering tradeoff? Wait until I tell you that ESS chips passband doesn't go to 20k, but ends at 19.88.

 

[Tks]

Do you honestly think that the engineers at AKM missed the chapters on Nyquist, yet still got hired, or that a 24k stopband is a conscious engineering tradeoff? Wait until I tell you that ESS chips passband doesn't go to 20k, but ends at 19.88.

I think the latter. The only issue being, why would I care if it was either/or (ignorance, or a engineering trade-off). They put out their latest and most complex chip design, yet still suffer the identical "tradeoff". I just fail to understand why ESS can do it, but they aren't able.

As for the passband ending at 19.88, I don't see much of an issue there (sure it's not "ideal"), but for the nitpicky realm of discussion here, I'd rather that, than it going out to 24kHz just for the sake of equipment not having to bother trying to output such signal (which is basically the only reason I care for filters in the first place, not so much for the sake of audibility at all really, seeing as how the only difference I hear is when I use a non-oversampling filter for example, but the rest? Yeah they're basically identical to me).