• 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!

Delta-sigma vs “Multibit”: what’s the big deal?

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Getting 4, 5, 6 or even 8 bits to have excellent linearity is rather easy.
Its the signal processing that makes the DS DAC a DS DAC. The fact that it can be done using several bits is another matter and has advantages with post filtering and linearity.

Beyond 16 bits is the real challenge. Even using tricks there has its limits.
 

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
3,427
Likes
3,981
Location
SoCal
Why do you think so?

Compare the then top of the line Pcm1704 specs to that of top of the line ESS or AKM chips today. Apparently it took a lot of time and effort for TI to get that PCM1704 to the market, yet it could barely beat the older generation of chips such as PCM63 in real applications. That technology maxed out. The folklore about multibit superiority lives on however.
 

mocenigo

Major Contributor
Joined
Dec 8, 2018
Messages
1,274
Likes
1,034
Getting 4, 5, 6 or even 8 bits to have excellent linearity is rather easy.
Its the signal processing that makes the DS DAC a DS DAC. The fact that it can be done using several bits is another matter and has advantages with post filtering and linearity.

Beyond 16 bits is the real challenge. Even using tricks there has its limits.

I know this of course. Also, sometimes these 5 bits are implemented with 32 independent switches that are load balanced, so any tiny resistor or capacitor variance is averaged out. But this also proves that "pure 1-bit ΔS" has been surpassed as well (and in fact the folks that do very high 1-bit ΔS modulation in SW to feed DSD dacs that internally reduce the frequency and use multi-bit ΔS puzzles me a bit).
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
AFAIK there are no off the shelf pure 1bit DS DAC chips.
I do remember a SONY DAC chip that was 1 bit at the time Technics had MASH, a 4 bit DS alike chip that some thought MASH was a 1bit DAC. (I was working at Technics at that time)
This was decades ago (around 1990). This was at the same time 18 bit DACs were becoming the rage.
 

mocenigo

Major Contributor
Joined
Dec 8, 2018
Messages
1,274
Likes
1,034
AFAIK there are no off the shelf pure 1bit DS DAC chips.
I do remember a SONY DAC chip that was 1 bit at the time Technics had MASH, a 4 bit DS alike chip that some thought MASH was a 1bit DAC. (I was working at Technics at that time)
This was decades ago (around 1990). This was at the same time 18 bit DACs were becoming the rage.

Yes, today off the shelf pure 1bit DS DAC chips either do not exist or they are very rare. In the past it was different of course.

My point was not about their availability, but the fact that some people perform 1-bit DSD upsampling to DSD256, 512 and even 1024 and then feed it to DACs that support these formats, but it is not clear at all whether they convert them vanilla or they use an internal multibit DSD format - it would be indeed fun if each 32 consecutive bits of a DSD512 bit input were just added to form a 5 bit 2.8mhz stream.

This is probably the problem AKM was working on before the fire with the 4498: not only separation of digital and analog domains, but also offering SW developers the possibility to actually generate the stream the DAC internally converts. Who knows
 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
3,337
Likes
3,278
Location
.de
AFAIK there are no off the shelf pure 1bit DS DAC chips.
The CS4334/5/8/9 family is still around. But those are a basic 8-pin stereo DAC first introduced in 1998, 23 years ago, making them the longest-serving family of audio DACs I know of. No idea who still uses these, I guess it's only the SO-8 package that kept them alive for this long (the CS4344/5/6/8 family with a multibit design comes in 10-pin TSSOP).

But that's pretty much the proverbial exception to the rule... I doubt anyone has released a new purely 1-bit design in the last 20 years. The first multilevel modulators appeared as early as ca. 1994 (Burr-Brown PCM1710U) and swept through the industry in the late '90s.

When AKM's AK4393 appeared at the end of 1998 with a dynamic range of 120 dB and THD+N almost as good as a -K grade PCM1704, it would have been obvious that all-in-one delta-sigma DAC chips had finally triumphed over previous concepts even at the high end. You have to remember that before you would need 3 chips - one digital filter and two DACs. Fancier concepts often used a whole bunch of DAC chips to shift glitches away from digital zero and increase dynamic range, and those buggers weren't cheap (something like $20 for one PCM1704 I think?). As a final nail in the coffin, 1999 brought delta-sigma DAC models with integrated HDCD decoding (so you no longer needed a Pacific Microsonics digital filter for that) and DSD support. With quad speed support and new high-end DACs piling up in 1999 and 2000 (AD1853, CS4396/97, AK4394, AD1852, CS4392, PCM1738, CS43122), multibit was quickly left in the dust. Better performance with less board space and at substantially lower cost? No-brainer. (NPC still made multilevel delta-sigma jobs for conventional setups with external filters, the SM5865BM/CM, but those apparently weren't a big hit.)

Now I think the very best multibit converters in the world still were hybrid multibit jobs at this point, but then we are talking what was used in beasts like the Pacific Microsonics Model Two which reportedly could do >120 dB in loopback and cost as much as a small car (while still not turning a profit).
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Whenever I see threads like this the debate is usually those that use delta-sigma modulation vs. those that do not, not whether there is a multibit DAC (or ADC) inside the DS loop. Noise shaping is the real debate IME. I doubt many audiophiles know the details of the circuit topology inside their DACs. Multibit is anything from fully unary to fully binary, though most are actually segmented designs with unary MSBs and binary lsbs, and oversampled designs are any flavor of delta-sigma converters even though you can oversample conventional designs as well (albeit only gaining 1/2-bit per doubling without noise shaping).
 

Esotechnik

Member
Joined
Mar 9, 2019
Messages
72
Likes
5
Location
Russia
That technology maxed out. The folklore about multibit superiority lives on however.
What is a settling time for modern cheap "millibit" DACs?
If the DAC can't reproduce a single pulse well, it can't accurately reproduce digital audio (sample stream) either.
  • "There is a small group of devices with higher precision around -69 dB: Texas Instrument PCM4222 Evaluation Module, Focusrite Blue 245, Forssell MDAC-4, MSB Platinum IV Plus DAC, MSB Platinum Studio ADC, SPL Madison, ReQuest Audio The Beast DAC, Swissonic DA24 MkII"
 

Esotechnik

Member
Joined
Mar 9, 2019
Messages
72
Likes
5
Location
Russia
that all-in-one delta-sigma DAC chips had finally triumphed
r1300-ak4498block.png

50 dB better than "conventional" converters. Back to the future :)

69+50 = hi-fi?
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,205
Location
North-East
If the DAC can't reproduce a single pulse well, it can't accurately reproduce digital audio (sample stream) either.

A "pulse" is not a valid audio signal, so why would you want to reproduce it accurately and what does accurately mean? Audio is band-limited, as is the ear, but a pulse is not. And, as far as I know, the goal of a DAC is to reproduce audio signals for human consumption.

I think you need to explain a bit more why you're mentioning DF/Gearspace or AKM chips, it's not clear to me how these are related to the topic.
 

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
What is a settling time for modern cheap "millibit" DACs?
If the DAC can't reproduce a single pulse well, it can't accurately reproduce digital audio (sample stream) either.

Read the audience. There are actual experts here, not self appointed arm chair experts. People who have developed the actual technology that this industry runs on.

It's cute when someone tries to tell us "how things work".
 

Esotechnik

Member
Joined
Mar 9, 2019
Messages
72
Likes
5
Location
Russia

Attachments

  • 3-s2.0-B9780750672917500145-f12-66-9780750672917[1].jpg
    3-s2.0-B9780750672917500145-f12-66-9780750672917[1].jpg
    13.8 KB · Views: 65

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,749
Location
Alfred, NY

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,749
Location
Alfred, NY
Commercial digital music is anti-piracy demo version of vinyl.
4-6 bit "enough for anybody."
Hi-res over-compresssed than mp3:
Credibility is of concern only to enthusiasts.
I suppose switching from "ignorant' to "irrelevant" is progress. Also, you apparently don't know the difference between dynamic compression and data compression. But don't let that stop you.
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,749
Location
Alfred, NY
Top Bottom