• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Best spec ADC Chip currently.... ??

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
hi bennetng, that was very informative

im wondering if you have seen the Sound Devices patent for their triple ADC that claims 168 dB of dynamic range?

https://patentimages.storage.googleapis.com/a7/6b/f5/77e31e68cca8b7/US9654134.pdf

Their implementation in the Mixpre-II series states 142dB of dynamic range in the ADC section. That series uses 32-bit float files. In your opinion would such a setup theoretically perform better writing to 32-bit fixed? The EIN of the unit is -128dBU which both float and fixed can contain.

I think you will find bennetng's posting is usually very informative.

As for EIN, there are physical limits to how good that can be. It is usually applied to microphone pre-amps with a source impedance of 150 ohms. A maximum value is 130.9 dbU over 20hz-20,000 hz. This is how much thermal noise a 150 ohm resistor would create.

The advantage to floating point is you won't clip the signal. Mixpre's are meant for portable on location recording often of sound for video. So doing it via floating point means you don't get clipped levels for live events. Writing to fixed 32 bit is plenty of range, but doesn't give you the options to avoid clipping or the noise floor that floating operations do. You'd still need to have your max level set not to exceed it and while the noise floor of 32 bit fixed is plenty low enough your analog circuitry handled that way won't make use of the 32 bit range anyway. The advantage of float is being able to scale up and down without clipping so it fits optimally in the dynamic range of the rest of the analog circuitry.

I'm probably explaining it poorly.

Sound Devices explains it.
https://www.sounddevices.com/32-bit-float-files-explained/
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
hi bennetng, that was very informative

im wondering if you have seen the Sound Devices patent for their triple ADC that claims 168 dB of dynamic range?

https://patentimages.storage.googleapis.com/a7/6b/f5/77e31e68cca8b7/US9654134.pdf

Their implementation in the Mixpre-II series states 142dB of dynamic range in the ADC section. That series uses 32-bit float files. In your opinion would such a setup theoretically perform better writing to 32-bit fixed? The EIN of the unit is -128dBU which both float and fixed can contain.
Hard to tell as dBFS (relative) and dBu (absolute) are different things, therefore the analog parts plays an important role here. In terms of safety floating point is always better since you can never clip in the digital domain. For the analog part doing real measurements would be fairer.
 

jerryfreak

Active Member
Joined
Jan 12, 2020
Messages
125
Likes
131
Location
Reno, NV
I notice that hot line level signals (>+24 dBU) get high voltage rather quickly, i imagine that voltage is the limiting factor of the dynamic range on an upper end

so is it fair to say that no analog microphone input has a dynamic range over 130dB? or is there a way to push to upper limit up

so if you match 0dBFS of the ADC to the clipping point of the analog input, wouldnt a 144dB 24-bit integer have a quantization noise well below the 130 dB EIN? enough that it would be essentially masked? and of course a 186dB 32bit integer would be well below
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,198
Likes
16,981
Location
Riverview FL

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
I notice that hot line level signals (>+24 dBU) get high voltage rather quickly, i imagine that voltage is the limiting factor of the dynamic range on an upper end

so is it fair to say that no analog microphone input has a dynamic range over 130dB? or is there a way to push to upper limit up

so if you match 0dBFS of the ADC to the clipping point of the analog input, wouldnt a 144dB 24-bit integer have a quantization noise well below the 130 dB EIN? enough that it would be essentially masked? and of course a 186dB 32bit integer would be well below
That makes sense. One can always "push the limit" by throwing the mic on the floor:)
So basically it is a pissing contest, as long as you are recording at a sensible digital level.
 

andymok

Addicted to Fun and Learning
Joined
Sep 14, 2018
Messages
562
Likes
553
Location
Hong Kong
I'm not grown enough to know a lot about the history of DAW :), but from my understanding Merging's been using 32-but floating point right from the start for their Pyramix mixing engine 20 years ago. What I see is they're trying to provide a complete 32-bit workflow end to end, all the way from acquisition to distribution.

Practical or not, their just trying to take away the limit. As to clipping, engineers should always avoid that anyways, and they can now adjust the gain on the fly anyways.
 

jerryfreak

Active Member
Joined
Jan 12, 2020
Messages
125
Likes
131
Location
Reno, NV
so can line inputs with higher impedance have wider dynamic range than 130dB? or is it there assumed there is always *some* gain stage between the input and ADC that will amplify thermal noise?
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
so can line inputs with higher impedance have wider dynamic range than 130dB? or is it there assumed there is always *some* gain stage between the input and ADC that will amplify thermal noise?
You have it backwards. Lower impedance has lower noise output. And it is the output impedance of the source of the signal, not the input impedance of the input.
Play with this thermal noise calculator. It will help it make sense.
http://www.sengpielaudio.com/calculator-noise.htm

You can set the gain it can be zero or 60 or 75 or more db. Gain will always amplify the noise level.

Zero gain and the source impedance set a noise floor you don't get below. There are ways to investigate below that however.
 
OP
S

SoZo

Member
Joined
Apr 21, 2020
Messages
21
Likes
9
I looked at them before, but with this new card Im going Merging 100% thx all....
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,340
Likes
688
Since this thread is about A/D, I'd like to ask about multibit versus one-bit. I know on the DAC side, 1-bit came to dominate, and now there is some revival of multibit and sometimes non-oversampling. How about on the ADC side? What has been the historical trend over time?

And has anyone implemented an ADC without the antialiasing lowpass filter (considering that at higher sampling rates and with real microphones, there might not really be anything to alias).
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
I know on the DAC side, 1-bit came to dominate, and now there is some revival of multibit and sometimes non-oversampling. How about on the ADC side? What has been the historical trend over time?
Multibit delta-sigma is the norm for DAC and ADC, 1-bit delta-sigma is a thing in late 90s and early 2000s.
BTW, the old Benchmark ADC with "obsoleted" 24-bit technology showed better measurements than these stacking 32-bit ADCs.
https://www.audiosciencereview.com/...e-10t-welcome-to-add-others.16332/post-528540
https://www.audiosciencereview.com/...e-10t-welcome-to-add-others.16332/post-533028
 
Last edited:

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,340
Likes
688
Multibit delta-sigma is the norm for DAC and ADC
That's interesting, since when I worked for giant audio company we were promoting multibit delta-sigma. I described it as 3 bits hiding inside a 1-bit, or something like that. My boss HATED it when I said that...I was never really sure why but I flat said "well that IS what is going on!" He eventually acquiesed after I invented some catchy trade name (so catchy that I no longer remember it :rolleyes: but it worked at the time)

So nobody is making specialty multibit A/D?
 

IVX

Major Contributor
Joined
Mar 3, 2019
Messages
1,407
Likes
2,771
Location
South of China, SHZ area, - Слава Україні
As I mentioned, ESS docs are always full of chaos, even regarding the input voltage you can find different value!
2021-01-04_15-37-36.png
 

earlevel

Addicted to Fun and Learning
Joined
Nov 18, 2020
Messages
545
Likes
776

IVX

Major Contributor
Joined
Mar 3, 2019
Messages
1,407
Likes
2,771
Location
South of China, SHZ area, - Слава Україні
I guess it is apostate from Canada BC. The docs quality is exclusively awful, even impossible to get any clue about how to turn the chip On. I did turn On the I2S interface to the master mode and set 48/3072kHz but can not get serial data anyhow. A lot of default values mismatch, see attached. All mnemonics are unique and zero comments on it, many of them mentioned just once or twice in the entire pdf, so I feel like a thief against safe-box. And I did sign the NDA, what for! I believe the most important ESS secret is not the product's info but the quality of the info.

PS: try to guess, why they gave me the reg#63;64? Do I need to turn Off some integrator1 or 2 or 3, how to set its current? Why I need to On/Off the comparator or "logic"?? What a hell is DAC does in the ADC, do I need that DAC or f..k it Off? Use the state of the comparator as a logic output or not?? Is somewhere is a candid camera, is it a prank?
2021-01-04_19-27-08.png
 
Top Bottom