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

[JJB70]

My field of engineering is very different but there are similarities in that implementation is invariably more important than throwing a lot of good components into a basket and assuming it'll all add up to a market leading product. However, in terms of marketing being able to name check a few key things is highly prized by those who sell the things and those who buy. Wrong? Maybe, but we are where we are.

 

[DonH56]

Many years ago I was attending a conference and watching in disbelief as the presenter addressed one problem after another, many with well-known solutions, and described the extra circuits he used to fix the problem. And the problems caused by those circuits, etc. I scribbled "If it don't work, throw more transistors at it!" on my copy and showed it to my friend, who got a dirty look from the moderator when he laughed out loud. A well-known industry icon (Bob Pease) was there and I expected him to rip the guy apart, but he didn't say anything. I asked him later at dinner and he just shook his head and said something like "some people never learn, and some are so far gone it's not worth the time to explain". He went on to disparage (over)use of simulations (SPICE) and said something along the lines of forcing people to do it on paper would make them realize just how simple a good design can be. The Widlar bandgap, based on a challenge to build the simplest reference, is a great example (Widlar won that one).

 

[THW]

Many years ago I was attending a conference and watching in disbelief as the presenter addressed one problem after another, many with well-known solutions, and described the extra circuits he used to fix the problem. And the problems caused by those circuits, etc. I scribbled "If it don't work, throw more transistors at it!" on my copy and showed it to my friend, who got a dirty look from the moderator when he laughed out loud. A well-known industry icon (Bob Pease) was there and I expected him to rip the guy apart, but he didn't say anything. I asked him later at dinner and he just shook his head and said something like "some people never learn, and some are so far gone it's not worth the time to explain". He went on to disparage (over)use of simulations (SPICE) and said something along the lines of forcing people to do it on paper would make them realize just how simple a good design can be. The Widlar bandgap, based on a challenge to build the simplest reference, is a great example (Widlar won that one).

i mean, simple solutions are sometimes the last things that come to mind, has been my experience

 

[bravomail]

the way multibit is implemented - u get square shaped signals aka more noise . with deltasigma u get free linear interpolation and less noise.

not an expert in engineering, but from what I’m gathering from others outside here, “multibit” is supposed to be better because it “preserves the original samples”... but if said “multibit” design distorts the signal, wouldn’t that mean that it also isn’t really good at preserving the original signal in the first place?

and delta-sigma is supposedly inferior because it’s a simulation... but then what’s so bad about a very well executed simulation?

based on this, wouldn’t the implementation of both design types and the design as a whole matter more than whether or not one is “delta-sigma” or “multibit”?

just a horse sense take on this, lol

 

[bravomail]

solderdude explained it better

 

[garbulky]

Multibit has issues with linearity at low level signals due to (I think) resistor matching.

<Subjective impressions - no dbt tests performed here>
I've used both. I liked the sound of both. I felt multibit had a "timing" thing that felt right to me and was easier to listen to. But I couldn't nail down why. Since my exposure to multibit units is low - I've listened to PCM 63 and Schiit's Gungnir multibit solutions, I felt that my DC_1 had a bit better detail but the music felt easier to follow to with multibit.

 

[DonH56]

I have no idea what "free linear interpolation" means and my knowledge of delta-sigma designs does not include that phrase... And they also put out rectangular pulses, same as multibit. Noise shaping and oversampling reduces in-band noise and pushes it above the audio band. You can oversample conventional multibit designs, too, but you only gain 1/2 bit for each factor of two in oversampling since there is no noise shaping.

Multibit data converter linearity is related to resistor matching, transistor matching, thermal gradients, nonlinear output impedance of current sinks/source and switches, and many other things, and is not just at low levels unfortunately. For that matter output buffer linearity for multibit or delta-sigma designs is worse at higher levels. And many of the same things affect delta-sigma designs. There are too many variables to draw simple conclusions or make sweeping generalizations IME/IMO.

 

[The Dragon]

Congratulations, you have just discovered that the vast majority of things said on audio forums is total ignorant nonsense.

LMAO.

The differences in how multi-bit vs. delta-sigma DACs decode the same stream of digital numbers gives us audiophiles something to debate and talk about.

 

[The Dragon]

Speaking of measurements, it just seems strange to me that “subjectivists” would reject them outright, claiming that you can’t say how good something is if you only look at the measurements

But most of the reviewers that have been reliable to me so far have been able to link their listening experience with what is shown in the measurements, which makes that line of argument even stranger to me

I have never understood the defense of some DACs that don't measure well when the very first requirement for a DAC is to be able to interpret the stream of digital data ACCURATELY. Sure, some DACs may "sound" good that don't have great test data. But, that doesn't mean they are reproducing the signal accurately. There are many things that can influence the sound. The quality of the power supply for example. We have seen instances of some of the more popular, so called "high end" DACs with power supply modulation issues. Then there are the filter implementations, some of which are proprietary. These filters absolutely affect the sound. Then there is the analog buffer section which must be of high quality.

A truly great DAC should both measure well AND sound good. One without the other defeats the premise of High Fidelity. I don't care what the implementation (multi-bit, delta-sigma, or custom). This does not necessarily seem to correlate with cost directly in many cases.

 

[THW]

I have never understood the defense of some DACs that don't measure well when the very first requirement for a DAC is to be able to interpret the stream of digital data ACCURATELY. Sure, some DACs may "sound" good that don't have great test data. But, that doesn't mean they are reproducing the signal accurately. There are many things that can influence the sound. The quality of the power supply for example. We have seen instances of some of the more popular, so called "high end" DACs with power supply modulation issues. Then there are the filter implementations, some of which are proprietary. These filters absolutely affect the sound. Then there is the analog buffer section which must be of high quality.

A truly great DAC should both measure well AND sound good. One without the other defeats the premise of High Fidelity. I don't care what the implementation (multi-bit, delta-sigma, or custom). This does not necessarily seem to correlate with cost directly in many cases.

Yeah the idea that something that measures well won’t actually sound well is quite strange to me, I have yet to see a good reason for completely disconnecting measurements from actual listening experience, assuming that your priority is accuracy

 

[DonH56]

Yeah the idea that something that measures well won’t actually sound well is quite strange to me, I have yet to see a good reason for completely disconnecting measurements from actual listening experience, assuming that your priority is accuracy

A look at many reviews will reveal that is often not the highest priority, and some go so far as to say the opposite.

 

[THW]

A look at many reviews will reveal that is often not the highest priority, and some go so far as to say the opposite.

I mean, I’m not disagreeing there

Thing is though, I just find it strange how it’s coming from the same people who say they strive for “high-fidelity”

 

[solderdude]

High-Fidelity means 'the least altered / most accurate' to some folks.
High-Fidelity means 'the most pleasant/natural' sound to other folks.
These may go hand in hand for some folks but may not be so for others, because of audio religion or personal taste.

The same could be said for a lot of other 'audio terminology'.

 

[JJB70]

I am probably like a record with a scratch on this subject but the attitude I really can't understand is liking a euphonic distorted sound simultaneously being obsessed with preserving a pure signal path with no tone controls, EQ, DSP etc. I can actually get why people like a warm or distorted sound which sounds better to their ears. That is a purely subjective preference but I get it. I also get why people want the best possible measured performance on the basis that it is not the job of replay equipment to punch its own stamp on music. I also get why people don't care either way if they can just listen to music and enjoy it. But once you purposefully embrace distorted sound because you prefer it then it strikes me as completely irrational to then have a total aversion to tone controls etc as you have already thrown accuracy away. And personally I don't buy the arguments that tone controls and EQ compromise SQ unless catastrophically badly implemented.

 

[THW]

Yeah I’m using EQ more as of late to better tune my headphones and I actually think my music sounds noticeably better overall so I also don’t particularly agree with the sentiment that EQ “ruins” the sound or anything, my music still sounds as clear as ever after EQ

It’s certainly a lot better than trying to spend even more money for marginal returns

 

[Dogen]

I remember back in the 90s I bought a Yamaha CD player with a multibit DAC. Curiously, the salesperson (from a Circuit City or similar) warned me I wouldn’t like it, but it had features I wanted so I bought it. It wasn’t a bottom of the line player. Anyway, I returned it within days. It smoothed everything out into a bland, easy listening type sound. No guts whatsoever. My wife, who had great ears but no audio vocabulary, picked up on it independently. I guess since then I’ve had a bias against multibit. Totally irrational I’m sure. It’s the only time I can remember hearing big differences from non-broken CD players.

 

[Deleted member 65]

Not much to do with multibit vs deltasigma.

Had to switch speakers from main room to media room due to energetic 4 yo grandchild who threatened to flip my Gauder Akustik Arcona 100 speakers.

Anyway we're happy with the switch. Evolution Acoustics MicroOnes now in our main listening room for the first time. They used to reside there a couple of years back driven by 2 * Hypex nCore monoblocks and an AudioLab MDA DAC/Pre. Never sounded good to my ears. Tried them with an Audio-Gd Ref 5 partnered with ShengYa pre and amps, no luck as for SQ.

Anyway, here we are a few years later, Evolution Acoustics MicroOnes speakers in main room (blame kid in my avatar) partnered with my Norma integrated amp, Topping D50 DAC for the first time and they sound sublime. Now just need to find some subs ...

 

[restorer-john]

You can oversample conventional multibit designs, too, but you only gain 1/2 bit for each factor of two in oversampling since there is no noise shaping.

The Sony CXD-1244S 8fs digital filter used in their TOTL CD players employed 45 bit noise shaping in 1989.

 

[restorer-john]

I remember back in the 90s I bought a Yamaha CD player with a multibit DAC. Curiously, the salesperson (from a Circuit City or similar) warned me I wouldn’t like it, but it had features I wanted so I bought it. It wasn’t a bottom of the line player. Anyway, I returned it within days. It smoothed everything out into a bland, easy listening type sound. No guts whatsoever. My wife, who had great ears but no audio vocabulary, picked up on it independently. I guess since then I’ve had a bias against multibit.

By early 1991, Yamaha's entire range was their 'S-Bit' (single bit). The last of their multibit machines (Hi-Bit) was sold in 1990.

 

[Dogen]

By early 1991, Yamaha's entire range was their 'S-Bit' (single bit). The last of their multibit machines (Hi-Bit) was sold in 1990.

Then it was earlier than I thought. Definitely Hi-Bit.