The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
32-bit converters??? Is ESS Sabre better than Apogee 16x??? Audio Interfaces
Old 1st July 2016
  #1
Gear Addict
32-bit converters??? Is ESS Sabre better than Apogee 16x???

Recently somebody mentioned 32-bit converters. As far as I knew 32-bit technology did not exist. Am I behind the times? Do we have converters capable of recording and playback at 32 bits? I looked at this ESS Sabre DAC. It does say 32-bit. And lists some insane dynamic range of 135 db! Am I misreading the specs? Are these chips available in studio converters? Which ones? What am I missing here?
Old 1st July 2016
  #2
Gear Guru
 
kafka's Avatar
Looks like it's been out, and that Apogee has been using the ESS SABRE32 as the basis for this generation of converters.

ESS Sabre32 DAC - Apogee Electronics

Depending on the chip, it looks like some are 32-bit for internal DSP, but using 24-bit externally. I'm not aware of any software that's using more than 24-bits. I'm certain that I don't have any mics quiet enough to take advantage of that kind of resolution.
Old 1st July 2016
  #3
Lives for gear
 
IanBSC's Avatar
It's a little complicated. There are 32 bit chips, but there still aren't ADC/DAC circuits that have greater than 130db dynamic range. And even the newest ESS 9038 DAC chip a dnr of 140db, which is huge, but still about 24bit spec. Any analogue circuit will have less range. The best ADCs chips are around 130db.

That I know of, the upcoming Mytek Brooklyn ADC and DAC are the only converters capable of 32 bit integer recording/playback. Presumably the benefit is more from DSP precision in the filters, and for an ADC , not truncating the output down to 24 bit. Maybe you get a cleaner digital signal path for processing. They claim it sounds better. I haven't actually heard it, so I can't comment on the sound quality. I've used a number of DACs with the ESS chips and they seem to be very good.
Old 2nd July 2016
  #4
Gear Addict
Quote:
Originally Posted by kafka View Post
Looks like it's been out, and that Apogee has been using the ESS SABRE32 as the basis for this generation of converters.

ESS Sabre32 DAC - Apogee Electronics
Hm... Did not realize Symphony has a 9 db higher dynamic range than my 16x's. Damn, I might have to buy new converters...

Quote:
Originally Posted by kafka View Post
I'm not aware of any software that's using more than 24-bits.
I can't speak for others but Cubase has always had 32-bit as an option. It is a waste of space to record at 32 bit but you certainly can.

Quote:
Originally Posted by kafka View Post
I'm certain that I don't have any mics quiet enough to take advantage of that kind of resolution.
I think any mic would benefit from that kind of resolution. And I would love to hear my Earthworks QTC50's through a 32-bit converter.
Old 2nd July 2016
  #5
Gear Addict
Quote:
Originally Posted by IanBSC View Post
That I know of, the upcoming Mytek Brooklyn ADC and DAC are the only converters capable of 32 bit integer recording/playback. Presumably the benefit is more from DSP precision in the filters, and for an ADC , not truncating the output to 32 bit. Maybe you get a cleaner digital signal path for processing. I haven't actually heard it, so I can't comment on the sound quality. I've used a number of DACs with the ESS chips and they seem to be very good.
I gotta research these Mytek Brooklyn converters... Thanks.
Old 2nd July 2016
  #6
Lives for gear
 
deuc647's Avatar
 

Quote:
Originally Posted by azzzy View Post
Hm... Did not realize Symphony has a 9 db higher dynamic range than my 16x's. Damn, I might have to buy new converters...



I can't speak for others but Cubase has always had 32-bit as an option. It is a waste of space to record at 32 bit but you certainly can.



I think any mic would benefit from that kind of resolution. And I would love to hear my Earthworks QTC50's through a 32-bit converter.
How is 32 bit a waste of space, not doubting you but ive heard this more than once and am curious why? Thanks
Old 2nd July 2016
  #7
Gear Addict
Quote:
Originally Posted by deuc647 View Post
How is 32 bit a waste of space, not doubting you but ive heard this more than once and am curious why? Thanks
Because the hardware cannot use the extra 8 bits. The quality of sound does not change from 24 bit to 32 bit on the current hardware. The only time it *hypothetically* makes a difference is when your plugins run at 32 bits and you want to keep the entire chain with no upsampling. I know a mastering engineer who works at 32 bits but it's only inside the box. As soon as the audio hits the converters it goes back to 24 bits.
Old 2nd July 2016
  #8
Lives for gear
 
Brian Campbell's Avatar
 

The best you can do on a good day is about 21 bits of dynamic range, the rest is thermal noise that you cannot avoid here on planet earth.
You could do better at zero degrees Kelvin
Old 2nd July 2016
  #9
Lives for gear
 
IanBSC's Avatar
Quote:
Originally Posted by azzzy View Post
I can't speak for others but Cubase has always had 32-bit as an option. It is a waste of space to record at 32 bit but you certainly can.
And this actually 32 bit float point, 24 bit with 8 extra added, not a raw 32 bit integer input.
Old 2nd July 2016
  #10
Lives for gear
AudioQuest Dragonfly DAC using the ESS Sabre as well and only with USB thumb drive size audiophile on the go DAC format. I'm using to mix my music on the go and it's sounding great...for the money...
Old 2nd July 2016
  #11
Lives for gear
 
carloff's Avatar
Quote:
Originally Posted by azzzy View Post
Because the hardware cannot use the extra 8 bits. The quality of sound does not change from 24 bit to 32 bit on the current hardware. The only time it *hypothetically* makes a difference is when your plugins run at 32 bits and you want to keep the entire chain with no upsampling. I know a mastering engineer who works at 32 bits but it's only inside the box. As soon as the audio hits the converters it goes back to 24 bits.
32bit floating-point for most users amounts to being able to red-line channels without clipping.It is almost impossible to digitally overload in the box with 32bit floating-point.It doesn't mean you cannot go out of the box then.Simply any 24 bit DA takes 32 bit floating signal like 24 bit. 12 years only with 32 bit floating point here.
Old 3rd July 2016
  #12
Gear Addict
Quote:
Originally Posted by carloff View Post
32bit floating-point for most users amounts to being able to red-line channels without clipping.It is almost impossible to digitally overload in the box with 32bit floating-point.It doesn't mean you cannot go out of the box then.Simply any 24 bit DA takes 32 bit floating signal like 24 bit. 12 years only with 32 bit floating point here.
I never said that you couldn't use 32-bit, let alone go out of the box once you used it. As for clipping, I don't do much in the box, but on the way in a hot 32-bit clips just as 24-bit does. A hot signal is a hot signal. If you're talking about redlining without clipping once the track has been recorded (as in during mixing), personally I don't see the point - it's easy to keep a recorded signal at a level where it doesn't clip. Again, not talking anyone out of using 32-bit - just personally don't see the point (until the hardware can handle it).
Old 3rd July 2016
  #13
Lives for gear
Quote:
Originally Posted by azzzy View Post
Again, not talking anyone out of using 32-bit - just personally don't see the point (until the hardware can handle it).
Most line inputs have 10 kOhm input resistance. A 10 kOhm resistor produces 1.79896 uV of noise at 20 °C which equals to -112.68 dBu. Gear with 17 V supply rails can't output more than 26.83 dBu, so in theory you could reach a signal to noise ratio of 139.51 dB with line level devices which is more than 4 dB less than the 144 dB that the 24-bit system offers. In reality you won't come even close to those values. You'll usually get a SNR of less than 90 dB from most mics. So no matter how good the converter technology gets, you won't get any advantage with the additional bits.

But while you wouldn't gain anything by using the 32-bit format in Cubase to record 24-bit data from an ADC and play it back, it makes sense in certain situations.
-You have a mix that doesn't clip the stereo out, because you reduced the level on the master fader. If you want to record steams from buses, they might clip. If you record them in 32-bit format, they won't clip.
-You do processing on the actual file. If you attenuate a 24-bit file by 48 dB and amplify it by 48 dB, the resulting file won't be able to have more than 96 dB of dynamic range and 16 bit of usable information. If you record 24-bit stream into a 32-bit FP file, you can do the same procedure without any loss of information. You could also amplify certain parts of the file without clipping that would otherwise occur.
Old 3rd July 2016
  #14
Lives for gear
 
toledo3's Avatar
 

Another side of it is that each app has a limited memory footprint, depending on a few factors. If all recorded files were 32bit float - which is way in excess of the spec it really needs to be to reproduce the dynamic range - that memory footprint runs out faster. So it's less tracks and plugins, or 24 bit files... which still have more dynamic range than the converter. So you just have to weigh out what is more important; as many tracks and as much DSP as you want, or overkill bit depth.

It's also worth noting that it would be possible to contrive a scenario where you record with 24 bit ADC, and get something like 21~22 bits of real dynamic range (really, much less because of room ambient noise, HVAC, etc).

THEN, through careful automation and/or DSP or that recorded signal, you could theoretically output greater dynamic range than you even recorded to the DAC. So that's already a neat little hack, outputting more dynamic range than you could possibly even record.

Pretty contrived, since most of us reduce dynamic range as is. Full scale 24 bits of dynamic range is enough to deafen. In all likeliness your output data can fit in much less than 24 bits, forget about 32 bit float. That's what the reduction to 16bit "CD quality" can often be pretty painless.

Many DAWs pass around 32 bit float buffers for DSP (all nowadays?), which then ensures that a full 24 bit dynamic range can be spit back out. If we could chain countless analog devices and not decrease dynamic range, it would be considered a miracle. Some DAW operations happen at even higher bit depths, 48, 56, etc., depending on the software in question.

In many ways, even higher sample rate isn't too necessary with modern oversampling converter design, which gets around nyquist sampling rate artifacts right from the start. You also potentially get rid of ultrasonic squeals that can saturate the signal chain without it being as easy to realize, because of the lower LPF.

I think the reason that you don't see a big push for 32bit floating converters is because it has to come from hardware and software people, who in their heart of hearts know that it just doesn't make sense. Most programmers, hardware designers, etc., have to think logically in order to accomplish their goals.
Old 3rd July 2016
  #15
Gear Addict
Quote:
Originally Posted by jetam View Post
Most line inputs have 10 kOhm input resistance. A 10 kOhm resistor produces 1.79896 uV of noise at 20 °C which equals to -112.68 dBu. Gear with 17 V supply rails can't output more than 26.83 dBu, so in theory you could reach a signal to noise ratio of 139.51 dB with line level devices which is more than 4 dB less than the 144 dB that the 24-bit system offers. In reality you won't come even close to those values. You'll usually get a SNR of less than 90 dB from most mics. So no matter how good the converter technology gets, you won't get any advantage with the additional bits.
I vehemently disagree. Tape inherently is a lot noisier than digital recording, yet it sounds to a lot of people more musical (and some would say more dynamic, even though tape compresses the signal when pushed, which is somewhat counter-intuitive). Human ear is not designed to hear segmented snapshots of a continuous wave, even if they are spaced very closely together (OK, granted everything ultimately breaks down to some discrete segments but we are very far from that resolution even at 24 bits). The idea is not just to keep the noise down but to have a natural reproduction of the sound.

Quote:
Originally Posted by jetam View Post
But while you wouldn't gain anything by using the 32-bit format in Cubase to record 24-bit data from an ADC and play it back, it makes sense in certain situations.
-You have a mix that doesn't clip the stereo out, because you reduced the level on the master fader. If you want to record steams from buses, they might clip. If you record them in 32-bit format, they won't clip.
-You do processing on the actual file. If you attenuate a 24-bit file by 48 dB and amplify it by 48 dB, the resulting file won't be able to have more than 96 dB of dynamic range and 16 bit of usable information. If you record 24-bit stream into a 32-bit FP file, you can do the same procedure without any loss of information. You could also amplify certain parts of the file without clipping that would otherwise occur.
All my mixing is done OTB. And if I need to print stems I do it from the buses on my hardware mixer (although I hardly ever print stems actually - once a mix is done it's done ). I have enough outboard hardware not to use plugins as well (even though I do own a full Waves suite and almost all UAD plugs). So for me going over what the hardware can handle in term of bits makes little sense. A mastering engineer has pointed out to me that he prefers 32-bit mixes, 'cause that's what he runs them at in his Pyramix. I might consider doing 32-bit for that reason in the future. But as far as the benefits of 32 bits within my studio, there are none. That being said, I totally understand that in some ITB cases 32-bit has advantages. So if it works for someone that's cool.
Old 3rd July 2016
  #16
Gear Addict
Quote:
Originally Posted by toledo3 View Post
I think the reason that you don't see a big push for 32bit floating converters is because it has to come from hardware and software people, who in their heart of hearts know that it just doesn't make sense. Most programmers, hardware designers, etc., have to think logically in order to accomplish their goals.
I don't know if there isn't a push for a 32-bit converter. Lavry once wrote a paper demonstrating that using the 192/24 is detrimental to the sound and 96/24 is better. In order to take advantage of 192 you need to step up to 32 bits. I won't pretend to be an expert in this but I have heard the same thing from a scientist studying all these things. Which is why I was wondering whether the 32-bit has been achieved in a converter. I guess not yet...
Old 3rd July 2016
  #17
Lives for gear
Quote:
Originally Posted by azzzy View Post
I vehemently disagree. Tape inherently is a lot noisier than digital recording, yet it sounds to a lot of people more musical (and some would say more dynamic, even though tape compresses the signal when pushed, which is somewhat counter-intuitive). Human ear is not designed to hear segmented snapshots of a continuous wave, even if they are spaced very closely together (OK, granted everything ultimately breaks down to some discrete segments but we are very far from that resolution even at 24 bits). The idea is not just to keep the noise down but to have a natural reproduction of the sound.
You can use all the adjectives you want to describe how tape sounds to you, but you don't seem to understand how digital sampling works. There are no segments when a digital signal is converted back to analog and the increased "resolution" that you are referring to with more bits is just a reduction of the noise floor. Nothing else. I know that it's hard to understand, but at least try to find some serious articles about sampling, noise shaping, dither etc.
If digital isn't continuous enough for you, tape isn't much better. There is an AC bias used when recording.

Quote:
All my mixing is done OTB.
In this case it doesn't matter.
Old 3rd July 2016
  #18
Gear Addict
Quote:
Originally Posted by jetam View Post
You can use all the adjectives you want to describe how tape sounds to you, but you don't seem to understand how digital sampling works. There are no segments when a digital signal is converted back to analog and the increased "resolution" that you are referring to with more bits is just a reduction of the noise floor. Nothing else. I know that it's hard to understand, but at least try to find some serious articles about sampling, noise shaping, dither etc.
Let's dial down the condescension a little, shall we? I perfectly understand that there are no segments in what comes out of the DAC. I was talking about what goes into the DAC before it gets converted to analog. Which is a series of samples.

Honestly, I have no desire (nor time) to get into an extended debate over who has better understanding of digital audio. I was asked a question about my view of the 32-bit recording, and I gave an answer applicable to my situation. Never said it was wrong to use 32-bit, only that for me it's a waste of hard drive space. You like your audio at 32 bits? Good for you. Let's leave it at that.

Cheers.

Quote:
Originally Posted by jetam View Post
In this case it doesn't matter.
That's all I was saying.
Old 3rd July 2016
  #19
Lives for gear
 
Brian Campbell's Avatar
 

It's worth noting that a good analogue tape deck is equivalent to about 13 or 14 bits in the digital world.
Old 3rd July 2016
  #20
Lives for gear
 
Friedemann's Avatar
 

Quote:
Originally Posted by bcwiz View Post
It's worth noting that a good analogue tape deck is equivalent to about 13 or 14 bits in the digital world.
And less then 48 kHz sampling rate...
Old 3rd July 2016
  #21
Gear Addict
Quote:
Originally Posted by Friedemann View Post
And less then 48 kHz sampling rate...
Now that's just nonsense. Tape has an infinite sampling rate.
Old 3rd July 2016
  #22
Lives for gear
Quote:
Originally Posted by azzzy View Post
Now that's just nonsense. Tape has an infinite sampling rate.
What?
Old 3rd July 2016
  #23
Gear Guru
 
kafka's Avatar
Quote:
Originally Posted by azzzy View Post
Now that's just nonsense. Tape has an infinite sampling rate.
Quote:
Originally Posted by jetam View Post
What?
This subject really should be banned on GS.
Old 3rd July 2016
  #24
Gear Addict
Quote:
Originally Posted by kafka View Post
This subject really should be banned on GS.
In fairness this was not the subject. The subject was whether there are converters capable of 32 bid depth.
Old 3rd July 2016
  #25
Lives for gear
 
Brian Campbell's Avatar
 

Quote:
Originally Posted by azzzy View Post
Now that's just nonsense. Tape has an infinite sampling rate.
tape does not have infinite bandwidth
Old 3rd July 2016
  #26
Lives for gear
 
Brian Campbell's Avatar
 

Quote:
Originally Posted by azzzy View Post
In fairness this was not the subject. The subject was whether there are converters capable of 32 bid depth.
yes to capture all that noise and thermal hash beyond 21 bits
Old 3rd July 2016
  #27
Lives for gear
 
Brian Campbell's Avatar
 

Quote:
Originally Posted by kafka View Post
This subject really should be banned on GS.
It's kinda fun for a minute but really been hashed over and over in many other threads...
Old 3rd July 2016
  #28
Gear Addict
Quote:
Originally Posted by bcwiz View Post
tape does not have infinite bandwidth
Where did I say that tape has infinite bandwidth???
Old 3rd July 2016
  #29
Lives for gear
 
Brian Campbell's Avatar
 

Quote:
Originally Posted by azzzy View Post
Where did I say that tape has infinite bandwidth???
post #21 (infinite sampling rate)
Old 3rd July 2016
  #30
Gear Addict
Quote:
Originally Posted by bcwiz View Post
post #21 (infinite sampling rate)
Post #21 does not mention bandwidth. What are you talking about???
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump