Jump to content

Mutibit vs Delta Sigma (old vs new)


Recommended Posts



Why all the focus on the type of DAC chip and almost nothing on the receiver chip that feeds the DAC?

That chip may actually be more important to the sonic signature, no?

It might be, yes. In my experience of developing a DAC design, I originally assumed this. But so far I have

focussed all my attention on other aspects because the (annoying) artifacts I was hearing turned out to be

DAC-chip, layout and output stage derived. That's not to say that the receiver chip makes no difference,

rather its to say it hasn't so far been the major difference in my experience.

Some receiver chips are better than others where measurements are concerned - for example technically

the WM8805 is better than the CS8414. It also has a lot more features for the money. Whether I can hear

this difference in measurements in practice is yet to be determined.

Link to comment
Share on other sites

For all you doubters out there who don't believe in noise modulation - here are a couple of FFTs from the AD1955 datasheet which show it clearly. What to look out for is the difference in the noise floor between when a signal is present (first plot) and when its not (second one). The difference is >10dB in the region where our ears are most sensitive (around 4kHz) and this is an averaged measurement so short term differences may well be greater.

post-131552-0-20679200-1333671586_thumb.

post-131552-0-92854400-1333671607_thumb.

Link to comment
Share on other sites

Here's a plot from the AK4399 datasheet - another manufacturer's S-D part. The noise modulation can be inferred from the full scale

THD+N measurement of -105dB when set up against the dynamic range (unweighted) figure of 120dB. The FFT plot verifies that the

full-scale THD+N figure given is indeed dominated by the noise as the 3rd harmonic distortion shows about -109dB. (Vertical scale I

cropped - its 10dB per division).

post-131552-0-99705300-1333677264_thumb.

Edited by techspurt
Link to comment
Share on other sites



I2s is the only way to go.................

That's not quite true! What it should read is:

I2s the way its done by Mario is the only way to go! Everything else is crap... This is a proven and non-disputable fact - proven by years of experience and confirmed again and again by the many followers who worship at the temple of Kaj.

  • Like 1
Link to comment
Share on other sites

Guest regal

Glad to see this discussion. The biggest problem is audio manufacturers in the business of integrating DAC chips are at the mercy of the big chip foundries. Of course they and the jounals have to defend the modulator technology, they have no choice as you can't develop a business plan based on an integration of an onsolete or "planned obsolete" chip. So I empathize with the attempts to change the subject and direct this read off tangent.

An important point was made, one can't design a voltage regulator that does anything with "noise" above 1Mhz. Today we have DSP's commonly running at 400 mhz, some 600 mhz. We have asynch usb sharing ground with a computer running at 2.4 GHZ. Then the cheap modulating chips themselves run well above any speed that can be dealt with. With DC, the ground is just as important as the regulator side. These high frequencies are inaudible in themselves but they 'cause instability in the wonderful analog circuitry we design, and that we hear.

I am by no means an expert but know that a PCM1704 with the BPO & servo defeated (big slow caps on the pins) with good discrete power regulation paired with a antique PMD100 and a modern <3ps spdif transport (Jocko Legato) is at a level where a S-D can't compete for my ears. Why? Maybe because a pulse transformer separates my analog from megahertz oscillation inducing garbage.

We all have different preferences and subjectivity plays a huge role with audio, but it is time for the DAC integrators to start pushing back to their suppliers and listening to their customers.

Tell TI " we don't want these modulators." Tell Analog Devices "your Sharc + AD1955 model isn't interesting our customers". Tell "ESS nice try but not what we want."

Instead of all the audio integrators trying to outsmart each other with jitter babble, maybe they should team up, listen to customers, and tell the big chip companies to wake the f up ;)

Edited by regal
Link to comment
Share on other sites

The biggest problem is audio manufacturers in the business of integrating DAC chips are at the mercy of the big chip foundries.

As a DAC designer I see this as an opportunity, not a problem. I agree a lot of the designers think like this - but then there are

others who break the mould. Vincent Brien's 'TotalDAC' is one such. Cees at Metrum got away from using a DAC targeted at

audio - its still from a big foundry, but it demonstrates that designers aren't limited to what the big chip companies design

specifically for audio. A lot of the fun comes from re-purposing alternative and non-mainstream parts in my experience.

Of course they and the jounals have to defend the modulator technology, they have no choice as you can't develop a business plan based on an integration of an onsolete or "planned obsolete" chip. So I empathize with the attempts to change the subject and direct this read off tangent.

Businesses don't though have to depend on business plans (and VCs, etc. etc.). They

can grow organically. But sure there's a good helping of defensiveness around trying to

support established habits with all kinds of smokescreen BS. Rather than say 'yep, its

the only way we could raise finance by doing it this way.'.

With DC, the ground is just as important as the regulator side. These high frequencies are inaudible in themselves but they 'cause instability in the wonderful analog circuitry we design, and that we hear.

Yep - the myth of 'ground' is widespread.

I am by no means an expert but know that a PCM1704 with the BPO & servo defeated (big slow caps on the pins) with good discrete power regulation paired with a antique PMD100 and a modern <3ps spdif transport (Jocko Legato) is at a level where a S-D can't compete for my ears.

If you listen for yourself you're more of an expert than someone relying on 'established practice' and

measurements.

We all have different preferences and subjectivity plays a huge role with audio, but it is time for the DAC integrators to start pushing back to their suppliers and listening to their customers.

Tell TI " we don't want these modulators." Tell Analog Devices "your Sharc + AD1955 model isn't interesting our customers". Tell "ESS nice try but not what we want."

Its certainly one approach, but I doubt that they're in the mood for listening. They're driven by the majority

who don't much care about the sound quality (perhaps the 'iPod generation' ?). Big companies are compelled

by law to put shareholders before customers in the USA.

Instead of all the audio integrators trying to outsmart each other with jitter babble, maybe they should team up, listen to customers, and tell the big chip companies to wake the f up ;)

I myself prefer the approach of disrupting their businesses by gradually stealing their customers. Then when

they wake up its because their company has lost its revenue stream...

Link to comment
Share on other sites

If those who still think that newer (Delta Sigma) is better than R2R (multibit), and can't see that it's advertised to be better so the big companies can reap the larger profits, these three links may sink in and start some doubting the direction dac chip development has gone. Complete I'm sorry with more measurment graphs.

http://www.6moons.co...otaldac2/1.html

http://www.totaldac.com/ready_dac.htm

http://www.6moons.com/audioreviews/totaldac/1.html

Cheers George

Edited by georgehifi
Link to comment
Share on other sites



So, George, is your argument that the R2R measures better therefore it sounds better?

No as I said in my first post, properly implemented R2R (multibit) has always sounded better to me than properly implemented Delta Sigma (nothing mentioned about measurments).

But it's nice to know the the measurments made are backing up my listening preference of R2R Multibit sounding better to my ear than Delta Sigma.

As always no dac amp or any audio electronics worth it's salt has ever been developed without the use of measurments, it's nice to see them back each other up, and how it should be if the right measurments are made.

Cheers George

Link to comment
Share on other sites

If those who still think that newer (Delta Sigma) is better than R2R (multibit), and can't see that it's advertised to be better so the big companies can reap the larger profits, these three links may sink in and start some doubting the direction dac chip development has gone. Complete I'm sorry with more measurment graphs.

http://www.6moons.co...otaldac2/1.html

http://www.totaldac.com/ready_dac.htm

http://www.6moons.co...totaldac/1.html

Cheers George

George,

I don't want to buy into the DS versus multibit - but just want to correct a few things here wrt noise measurements.

These noise floor measurements are not really very well done and are ambiguous.

That noise floor measurement on the second link is NOT an RMS measurement and would not be close to -134dBFS RMS period.

The designer does not appear to correctly understand the nature of FFT measurements and how they work. This is a very common misconception.

When an FFT sweep is performed, the measuring device does a narrow bandwidth sweep, usually from 20Hz to 20kHz.

The narrower the bandwidth of the scan, the less effective noise power and so the lower the 'grass' on the FFT graph.

This is exactly how we can see a small level signal through much higher effective RMS noise.

Refer to the attached FFT below, which is of Anedio DAC that uses Sabre DAC chip. The quoted dynamic range is around 126dB (unweighted) but

the 'grass' in that FFT is below -150dB.

This is because the figure of -126dB is the RMS noise with a bandwidth of 20kHz. The bandwidth of the actual FFT sweep may have been only a few

Hz, as such it can 'see' the -140dB signal below the -126dB RMS noise.

The way those noise floor FFT's are presented, with grass around -130dB, I would assume a 20kHz RMS noise floor of around -100dB but it is

impossible to state without proper measurement conditions.

Z

post-110074-0-45260500-1334752042_thumb.

Edited by zenelectro
Link to comment
Share on other sites

Same fudging of specs goes on with all of them, Abraxalito says they need to measure differently.

Too many well healed "Golden Ears" prefer old school R2R multibit for Red Book playback and if you have HDCD decoding as well with it's excellent filtering charactistics your almost some say better than sacd, even guys like Nelson Pass.

The other proof is in the secondhand market, try to pick up a s/h dac or cdp cheap that has TDA1541 crown, or PCM63k, or PCM1702k, PCM1704k with PMD100 or 200 these all fast becoming investment property rather than clean up day junk, where they should be at 20 and more years old, people have ears and they know what gives the best Red Book sound.

I'm not interested in sacd or high rez down loads all I what is my hundreds of Red Book cd's to sound at their best and Multibit (R2R) does this better than any of todays hyper Delta Sigma chips can which are part bitstream ( single bit) and we remember how bland they sounded when they tried to push them onto us.

http://www.diyaudio.com/forums/digital-line-level/204065-have-you-seen-anything-like.html#post2853112

Cheers George

Link to comment
Share on other sites

Meh...

I have changed to a Marnatz CD65 and only changed the outout caps, it already has a bunch of Cerafine caps in it stock, and I must say that I really like the sound of this TDA1541 (non A) based player :party

Will do more mods as I go, Clock, Nos mod, seperated reg's...tube output.

Link to comment
Share on other sites



Definitely get rid of the SAA7220 - but that's not at all straight forward because its the clock master for the transport. A separate oscillator will be needed and you'll also lose a level of error correction, interpolation and soft muting. An easier alternative is to isolate its considerable power supply noise as far as possible and bypass the filter part to go NOS.

Link to comment
Share on other sites

Yeah, I was going to connect one of the pins from SAA7220 to one on the controller chip (forget off hand the numbers) to correct the muting after by-passing it. I'm not sure if those other aspects are audible, those that have done the nos mod report improvements without audible drawbacks :confused:

Will be adding separate transformers for power to some of the chips, as well as running the new clock off a battery.

Edit: I read up on the nos mod on the Lampi site http://lampizator.eu...mpling/NOS.html but this player does use a different controller chip (m4803a I think), so I need to look at that a bit more before doing it.

Reading a thread elsewhere i think It's SAA7220 Pin 23 to pin 11 on the m4803a

Edited by datafone
Link to comment
Share on other sites

I think you guys are missing the point of these older, simpler DACs. It has little to do with the S/N or the THD numbers. The lower these are, yes you will get a bit more detail in the playback. However the thing that makes these sound better has nothing to do with this. The natural sound of these has everything to do with the digital filtering or lack thereof IME. Take the digital filtering off any Sigma-Delta DAC (if you can) and find out for yourself. Really nice.

Steve N.

Link to comment
Share on other sites

Steve, while I see that you are saying... that if one was able to by-pass the filtering of a S-D chip there would be significant improvement?

But what are you implying regarding the relationship in the case of something like the TDA1541?

Sorry, and It might just be my lack of sleep, but I'm finding that part somewhat cryptic of sorts ;)

Edit: Are you agreeing, or disagreeing that the TDA1541 benefits with by-passing the filter chip?

Edited by datafone
Link to comment
Share on other sites



Might it be that this "jump factor" is down to the ability of multibit NOS to track & reproduce music transients better than the averaging of sigma-delta?

I doubt it. I have not experienced this jump factor correlation. I get great jump factor out of all D/A chips. Jump factor has more to do with jitter and power subsystem of the DAC IME. Maybe it was just coincidence that these were related by the poster.

Steve N.

  • Like 1
Link to comment
Share on other sites

Steve, while I see that you are saying... that if one was able to by-pass the filtering of a S-D chip there would be significant improvement?

Exactly. Its possible on some chips. That's how I know this.

But what are you implying regarding the relationship in the case of something like the TDA1541?

TDA1541 can sound really really good. No digital filtering. It NOS. Part of the problem with this chip is current drive on the outputs. If you parallel several, problem solved.

Edit: Are you agreeing, or disagreeing that the TDA1541 benefits with by-passing the filter chip?

What filter chip?

The version that is most popular is the TDA1543.

Steve N.

Link to comment
Share on other sites

I doubt it. I have not experienced this jump factor correlation. I get great jump factor out of all D/A chips. Jump factor has more to do with jitter and power subsystem of the DAC IME. Maybe it was just coincidence that these were related by the poster.

Steve N.

Here's what I experience & surmise. SD DAc chips in the short experience I have had with them have definite strengths in the bass with more slam & definition but when it comes to top end resolution it can reveal some thinness & sibilance. NOS DACs have superiority in the HF rendition & clarity but are not somewhat noticeably bass shy. Jump factor is something that was introduced here but I've no direct experience of it.

My premise is that the averaging of DS DACs works well when there are a large number of sample per musical cycle over which to average (in the LF area) & produce an apparent higher bit depth but are less capable at HF where the number of samples per music cycle is much less leading to a less focused sound. The transient attack of the leading edge on cymbals, & other instruments would be less well catered for in this averaging at HF & therefore the jump factor would also be a more noticeable characteristic of NOS DACS where no averaging noise shaping is happening

Edited by jkeny
Link to comment
Share on other sites

Here's what I experience & surmise. SD DAc chips in the short experience I have had with them have definite strengths in the bass with more slam & definition but when it comes to top end resolution it can reveal some thinness & sibilance. NOS DACs have superiority in the HF rendition & clarity but are not somewhat noticeably bass shy. Jump factor is something that was introduced here but I've no direct experience of it.

My premise is that the averaging of DS DACs works well when there are a large number of sample per musical cycle over which to average (in the LF area) & produce an apparent higher bit depth but are less capable at HF where the number of samples per music cycle is much less leading to a less focused sound. The transient attack of the leading edge on cymbals, & other instruments would be less well catered for in this averaging at HF & therefore the jump factor would also be a more noticeable characteristic of NOS DACS where no averaging noise shaping is happening

To be completely honest here, I think we are all wading around in the mud here trying to describe subjective artifacts that may or may not be produced by the DAC itself as opposed to the supporting circuitry.

For just about every argument here I can think of a technical counter argument. Just to use your above case here John (not picking on you), the HF performance of DS DACs is theoretically way superior to NOS multibit DAC's. If you

pass a 2 (or 3 or 4) tone 18k+19k tone through a DS DAC it will be almost perfect, do the same to a NOS 1541 and it will be obliterated. Measurement wise that is!

I think the main issue here generally is these DS DAC's are a complex amalgamation with quite a few different internal processes.

If we compare the 1541 to a Sabre, they are utterly diametrically opposed in just about every aspect.

- 1541 uses relatively slow / low noise current steering internal architecture // Sabre is a very high speed Cmos design

- 1541 has very easy PS requirements // Sabre will tax the PS, especially at high frequencies to the max

- 1541 is simple and devoid of all other internal functions such as digital filtering, ASRC etc.

- 1541's OP stage runs low sample rates and an easy +- 2mA // Sabre runs a high current, low impedance super fast switching OPS

The requirements on surrounding circuitry for Sabre are much more demanding - but as we know surrounding circuitry on 1541 is already -very- important.

There is also the issue that a not insignificant part of the typical 0xOS 1541 -> tube OP stage sound is -additive- ie; euphonic

A lot of factors to consider before we jump to conclusions about jump factors - there's a mouthful :)

Link to comment
Share on other sites

What filter chip?

Steve N.

TDA1541 (non A), SAA7220P/A

Edit:Not SAA7220P/B, so corrected.

Edited by datafone
Link to comment
Share on other sites



Guest
This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...
To Top