• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

RME changes DAC-chip

devink

Member
Joined
May 21, 2021
Messages
13
Likes
14
Right, that’s a good point. Though as a “pro” who needs the ADC I’d also want the best performing DAC.

I really hope they expand the use of ESS chips especially into their multichannel interfaces.
 
OP
Mulder

Mulder

Active Member
Joined
Sep 2, 2020
Messages
273
Likes
342
Location
Gothenburg, Sweden
About 120 THN + N, or 20 bits, is to my understanding maximum what can be achieved with regard to limitations in electronic components. RME ADI-2 is perhaps 3 or 4 db below that limit. Is it an example of good engineering to squeeze out those last numbers, or is it better engineering to ignore it, because no one still hears this difference? It may be a wiser strategy to put your efforts on things that actually matter - such as DSP, reliable firmwares and long-term and reliable support, etc. What really matters in the end? Engineering is not practised in a vacuum. It must always be set in relation to a cost and a purpose. At the level we are talking about here, where the RME ADI-2 is located, the DAC is the least of the concerns in the audio reproduction chain. Other parts affect more.
 
Last edited:

Skeptischism

Active Member
Joined
Sep 6, 2019
Messages
229
Likes
117
yeah nah :D

-120dB THD+N is about where it sat with the ES9018S nearly a decade ago, even with that chip you could get ever so slightly better. these days, with a well specified and built design with 9028 and 9038 pro and 2nd and 3rd harmonic 'predistortion' we are able to eek somewhere around -125-6 ish dB THD +N (with all 8 outputs summed to mono balanced output in dual mono) on the 28 and some 38pro dacs getting closer to -130. I believe syn08 was able to get something like -126dB SINAD with a dual 9038PRO DIY, which is incredible. (the now unavailable AK4499 was a touch better again)

with the 9038q2m I believe IVX and others have gotten better than -120, maybe -123 or so and like this 9028q2m its a low power chip for mobile phones and daps, so hardly state of the art (although honestly fabulous performers, incredible really to say that these days you can get this performance with a mobile chip). amplifiers and analogue stages can do much better again. an extremely high performance headphone amp for example can manage 130-140dB. Measuring starts to become difficult, but better is absolutely possible. Noise is limiting the performance, the distortion is much lower.

I had a look at the PCB to refresh myself of what was going on. if i'm not mistaken, it looks like perhaps they are using solid state output relays/protection. of course, given the number of outputs that need switching. when looking at ONSEMI's highest performance chips of this type just recently for our dac and IIRC the best was around -116dB THD+N, so perhaps those output switches are 'limiting' the output performance in both cases, so perhaps the ESS will indeed measure the same, as the output switch is the bottleneck for distortion. I could be wrong there after looking on my computer, they could just be the 6 parallel opa or the headphone out. standard relays, not solid state. they will cause some limitation too.

I wont argue with the rest of your post. I have no doubt it's an extremely well engineered unit.
 
Last edited:
OP
Mulder

Mulder

Active Member
Joined
Sep 2, 2020
Messages
273
Likes
342
Location
Gothenburg, Sweden
The best THD+N among those DAC:s reviewed here is about 123 db. I have never seen any DAC measured here with a THD+N reaching 125-6 db.
 

JohnYang1997

Master Contributor
Technical Expert
Manufacturer
Joined
Dec 28, 2018
Messages
7,017
Likes
16,459
Location
China
The best THD+N among those DAC:s reviewed here is about 123 db. I have never seen any DAC measured here with a THD+N reaching 125-6 db.
The THD+N is limited by APx555's self noise which is a couple of times higher than D90se. It should at least measure -127dB and perhaps towards -130dB. There's still more performance to be had in electronics. Headphone amplifier for example can already achieve near -150dB THD+N at 20Vrms.
 

Skeptischism

Active Member
Joined
Sep 6, 2019
Messages
229
Likes
117
and yes, I had written close to -150dB for headphone amps, but I couldnt remember 100% what commercial amp to mention, since i'm not with my finger on the pulse with current offerings. So I edited it back to 140dB
 
OP
Mulder

Mulder

Active Member
Joined
Sep 2, 2020
Messages
273
Likes
342
Location
Gothenburg, Sweden
Well, ok, so it can´t be measured? I am not convinced of anything, but I doubt if there is any gain from say 125 db compared to 115 db. Our hearing has it´s limits.
 

JohnYang1997

Master Contributor
Technical Expert
Manufacturer
Joined
Dec 28, 2018
Messages
7,017
Likes
16,459
Location
China
Well, ok, so it can´t be measured? I am not convinced of anything, but I doubt if there is any gain from say 125 db compared to 115 db. Our hearing has it´s limits.
Yeah no really any benefit other than looking impressive.
It can be measured tho just not directly with AP.
 

Skeptischism

Active Member
Joined
Sep 6, 2019
Messages
229
Likes
117
Well, ok, so it can´t be measured? I am not convinced of anything, but I doubt if there is any gain from say 125 db compared to 115 db. Our hearing has it´s limits.

We werent talking about that and nobody was making audible claims. its a red herring. I agree, although I do enjoy the technical endeavour and using a digital crossover and volume I prefer the highest performance I can muster in the DAC.

It's very difficult to measure with THD+N, because the measurement is noise limited and the best tool Amir has here is the APx555, so anything equal to, or exceeding its own input self-noise performance measures the same. It can be measured though (albeit not directly), some helper circuits and calculations are required.
 

Skeptischism

Active Member
Joined
Sep 6, 2019
Messages
229
Likes
117
It really did seem like you were convinced though ... I mentioned it, you countered it by mentioning the list here. I explained how, when and with what it has been exceeded, you mentioned the numbers and measurements here again.
 

devink

Member
Joined
May 21, 2021
Messages
13
Likes
14
Well, ok, so it can´t be measured? I am not convinced of anything, but I doubt if there is any gain from say 125 db compared to 115 db. Our hearing has it´s limits.
I see this sentiment a lot on ASR (that anything better than ~115 dB doesn’t matter), and while it might be true for consumers, it’s not true for professionals who work in audio production / post-production.

For pros, who aren’t working with already-mastered music, the dynamic range can be huge, and you typically need to leave headroom in the monitoring chain to have enough extra volume when working on quiet material. 12 to 20 dB is common.

In the past this was normally handled by an analog monitor controller (many of the popular one’s had a SINAD of ~100 dB), so that the DAC could always run near its maximum output and you’d change the analog gain as needed. But DAC’s are so good now that such an analog monitor controller would probably only harm performance.

These days, many pros use digital attenuation when monitoring, especially for complicated monitoring chains with subs, surround speakers, Atmos, etc. But if you’re leaving 12-20 dB of headroom, that means whatever SINAD your DAC has, you can essentially subtract 12-20 dB. So, a 120 dB DAC suddenly becomes a 100 dB DAC. (side note, this is one of the things RME did extremely well with the ADI-2: they change the gain staging with relays as you change the output volume, thus optimizing the signal-to-noise of the DAC and avoiding this problem / making it very easy to pair with any speaker/amp setup).

This is why it does actually make sense to over-engineer audio converters a bit and shoot for a little better than what we can actually hear — it allows for more wiggle room with gain staging and headroom. This is also an important consideration when printing audio through an analog processing chain, for example.

You might ask why a professional is posting on ASR in the first place. Well, in my opinion, professional DAC’s have sort of stagnated, and some of the best ones now happen to be in the consumer space. For example, there are no professional DAC’s that I know of that use an ESS9038PRO. Apogee’s Symphony 2x6 SE is the closest I know of, which I believe uses an ESS9028PRO.
 

Skeptischism

Active Member
Joined
Sep 6, 2019
Messages
229
Likes
117
^^ exactly my reasoning. I dont often do much production, although I dabble with analogue synths and logic, for fun, but the fact that I do, as well as using digital crossover, DSP/correction and digital attenuation, means that -115 doesn't quite cut it and i'm always looking for more headroom.

I think these days with more and more people using DSP for room correction and bass management, as well as just using the convenient and perfectly channel matched digital volume control in their dacs, this is becoming more and more widely applicable to audio enthusiasts in general, not jus producers.
 

Matias

Major Contributor
Forum Donor
Joined
Jan 1, 2019
Messages
3,201
Likes
5,795
Location
São Paulo, Brazil
 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
2,094
Likes
1,975
Location
.de
I see this sentiment a lot on ASR (that anything better than ~115 dB doesn’t matter), and while it might be true for consumers, it’s not true for professionals who work in audio production / post-production.

For pros, who aren’t working with already-mastered music, the dynamic range can be huge, and you typically need to leave headroom in the monitoring chain to have enough extra volume when working on quiet material. 12 to 20 dB is common.
Fine, let's do the maff.

So let's say you listen at 85 dB SPL (a standard level in the film industry, probably on the high end of music mixing levels). Plus 20 dB headroom = 105 dB SPL. Noise level should be <4 dB SPL to be inaudible. Required dynamic range: 101 dB.

Or maybe you even need 95 dB (not recommended except short-term)? That's 111 dB then.

This obviously assumes that your monitors have a somewhat sensible input sensitivity / control range. Still, if your DAC manages 126 dB you'll have at least 15 dB of leeway. 115 dB with selection in output stage gain will also do the trick. I bet not one user of the RME Fireface UC or UCX has ever complained about having too little dynamic range on the D/A side, despite it being a "measly" 113 dB(A) - the level choices of +2 dBV, +13 dBu and +19 dBu should be enough to make most happy.

People had figured the requirements out 30 years ago. The first commercial ADCs to manage 130 dB, at least A-weighted, appeared in the mid-late '90s. The first DAC chips capable of exceeding 120 dB(A) in the early 2000s. In practical use, you can get away with considerably less. About 110 dB seems to be the level you want for an ADC in typical studio use, where adjustment of mic level is possible. You'll be hard-pressed to get more through a fair few mic preamps anyway... and things are rapidly thinning out past 120 dB.

If you only want to record spoken voice, my own setup when (normally) peaking at -9 dBFS makes it to about -83 dB minimum noise (white noise equivalent) or ~14 bits...
 

BlueAlpha

New Member
Joined
Aug 18, 2021
Messages
2
Likes
0
Hey Guys—

I’m new here but have followed the forum for quite a while….

Anyway, I was on a wait list for the RME ADI-2 to get back in stock in one of the big venue stores. As soon as I was notified that it was back in stock I ordered it and it just arrived. It has a “B” at the end of the serial #. Does this mean that I have the AKM chip?

If so, is this preferable to the ESS chip? (I’m aware of RME’s position)
 

xaviescacs

Major Contributor
Forum Donor
Joined
Mar 23, 2021
Messages
1,015
Likes
1,120
Location
La Garriga, Barcelona
Hey Guys—

I’m new here but have followed the forum for quite a while….

Anyway, I was on a wait list for the RME ADI-2 to get back in stock in one of the big venue stores. As soon as I was notified that it was back in stock I ordered it and it just arrived. It has a “B” at the end of the serial #. Does this mean that I have the AKM chip?

If so, is this preferable to the ESS chip? (I’m aware of RME’s position)
Don't even think about it, just enjoy your device.
 
Top Bottom