Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Multi-bit DACs (Read 37743 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Re: Multi-bit DACs

Reply #25 –
Quote from: apastuszak on
From what I am gathering, multi-bit is more than just the DAC chip itself.  There is talk of PCB design and resistors.
That's got nothing to do with multi-bit. You need a PCB and resistors in any case. And a power supply. and capacitors. And connectors. And an enclosure. Audiophiles can make a big deal out of everything.

Quote
There is discussion of R2R DACs, which are claimed to be multi-bit.  While Delta Signal DACs are not.  Then another discussion starts up around Delta SIgnal DACs that are supposedly multi-bit.
R2R DACs are multi-bit quasi by definition. Delta sigma (not signal, the name derives from two greek letters) can be either single-bit or multi-bit, depending on their internals.

Quote
The only article I was able to find was from someone who sells R2R DACs at ridiculous prices.  I will admit that I don't understand enough about DACs to read this stuff and come up with an informed decision.  And even if I did, there's no proof any of this superiority is actually audible.
Due to their problems with low-level linearity, R2R DACs were long regarded as inferior for audio. Sure, with enough trimming and cleverness, you can get the linearity problems reduced below audibility. That's an old art, people had to resort to this kind of design in the 70s and early 80s, before the advent of delta sigma converters, but because of this it is expensive. So in a certain sense, the R2R advocates are harking back to the early years of digital audio. It is, so to say, a "vintage" technology. For audiophiles, that means it must be good. :)

Quote
They make a whole line of more expensive stuff because there is money to be made from placebophiles.  If you can crank out a bunch of high end gear and sell it and make a profit, why wouldn't you?
I wouldn't mind the money, my problem would be the nonsense that I would have to produce to trigger their audiophile buttons. I don't like lying.

There is one aspect that is often overlooked and may account for audible differences. If you drive a DAC with loudness-maximized signals close to clipping, there may be overloads. If anything is being done on the digital side, either an oversampling filter or a delta sigma modulator, which both amount to digital processing, then those filters may be driven into an overload state. That obviously depends on their individual design, and can not be generalized beyond any particular chip. However, extreme cases have become known where digital filters took many milliseconds to recover from such a state, creating nasty distortions in the process. You would expect that more modern chips are a bit more resilient to this phenomenon.

DACs with no digital processing are inherently immune to such problems. But it would be extremely silly to discount everything else summarily, because the cure is simple: Reduce the level of the signal on the digital side by a few dB, and the problem is gone.

Re: Multi-bit DACs

Reply #26 –
Quote from: eagleray on
1. Hard to drive headphones in which case a DAC and a headphone amp are purchased, often as a single unit.  Most likely symptom is low volume.  I suspect this is the source of most happy purchasers.
The problem here isn't that the headphone is hard to drive. Higher impedance headphones are typically quite easy to drive, except that they need more voltage than what a typical PC soundchip produces. Headphones with 600 Ohm impedance are so easy to drive that they can work off a conventional professional line output, but they need professional voltage levels. Headphones suited for the low voltages of PC soundcards and portable players typically have much lower impedances, which means (in the usual sense of the word) that they are harder to drive.

Re: Multi-bit DACs

Reply #27 –
Quote from: apastuszak on
The latest post I just read said that the difference between an R2R DAC and a  Delta-Signal DAC is "night and day" and no blind testing is necessary because of that.  Soon as someone says that no blind testing is necessary, alarm bells go off in in my head.
Rightly so.

Quote
There is one review on YouTube where someone ABs a Schiit BiFrost Multi-bit and says he can't hear a difference between that and a regular 1-bit delta-signal DAC.  I believe the BiFrost is a multi-bit delta-signal DAC, and not R2R. (Wow, I almost sound like I know what I am talking about!)  But when someome ABs (not blind) the two DACs and says they sound the same, then I don't have a lot of hope of personally hearing a difference between the two with my non-placebophile ears.
From what I see on their website it is a 16-bit R2R type DAC with a digital filter (an oversampling filter, most likely) in front of it. This kind of architecture was typical for the second half of the 80s, and into the 90s, until delta-sigma took over.

Re: Multi-bit DACs

Reply #28 –
So, now the whole world gets to see how I carried "delta-signal" through the whole post.  I actually typed delta-sigma a few times and then went back and checked my previous post and "corrected" it to what I typed previously.  Sigh....

Anyway, I would love to hear it, to prove to myself that it doesn't make a difference.  The problem with high end gear is that a lot of stores don't carry it, and you're stuff with online reviews and online purchases.  And most retail "audiophile" stores don't contain any low end stuff to compare it to.

There is one "audiophile" store near me that's only open about 10 hours a week.  And he doesn't have anything under $100 in there.  Getting him to allow you to ABX anything is next to impossible.

Re: Multi-bit DACs

Reply #29 –
Quote from: apastuszak on
Getting him to allow you to ABX anything is next to impossible.
Chances are that he wouldn't even have a setup for proper level matching, let alone the switching gear.

Re: Multi-bit DACs

Reply #30 –
And here is what all the big stink was about:

http://schiit.com/news/news/introducing-modi-multibit

Schiit has a new cheaper multibit DAC.

From the email chain I had with Schiit about this, I'm pretty sure they made this because they know it will sell, not because they actually believe it's superior.

I think it's funny that "audiophiles" won't take a company like SMSL seriously, or even FiiO.  But a soon as a company offers a high priced product, the company becomes worthy of audiophile notice.

The pre-hydrogenaudio placebophile in me in screaming to get one of these things to try it out, when the level-headed me knows very well that I'm not going to hear a difference.

Re: Multi-bit DACs

Reply #31 –
Quote from: apastuszak on
The pre-hydrogenaudio placebophile in me in screaming to get one of these things to try it out, when the level-headed me knows very well that I'm not going to hear a difference.
Worse: With some of these expensive audiophile products, you might indeed hear a difference. It wouldn't be a good sign, however. There's a real chance that it is actually worse in quality than what you get from a "normal" product.

Re: Multi-bit DACs

Reply #32 –
Quote from: pelmazo on
Quote from: apastuszak on
The pre-hydrogenaudio placebophile in me in screaming to get one of these things to try it out, when the level-headed me knows very well that I'm not going to hear a difference.
Worse: With some of these expensive audiophile products, you might indeed hear a difference. It wouldn't be a good sign, however. There's a real chance that it is actually worse in quality than what you get from a "normal" product.

But it costs more, so it must be better, right?

Is the whole R2R comeback thing kinda like the resurgence of vinyl?

Re: Multi-bit DACs

Reply #33 –
Quote from: apastuszak on
Is the whole R2R comeback thing kinda like the resurgence of vinyl?
There certainly is a vintage aspect to it. They don't say at Schiit, however. Perhaps because they don't use true vintage parts, but rather their more contemporary equivalents. Hardwired oversampling filter chips are largely extinct, so they use a freely programmable DSP. The old R2R converter chips are long out of production, too, so they pick what they can find now, even though these chips are made for different applications.

At least, I have no reason to assume that their design is neglecting the basic sampling theory, like in the case of many NOS DACs. Without having looked at it in any detail, it does seem that they use a more or less straightforward oversampling setup similar to what was en vogue some 25 years ago. Nothing particularly wrong with that, except that we can have similar performance for a lot less money these days with delta sigma.

Re: Multi-bit DACs

Reply #34 –
The need for R2R may come from the claimed superiority of dsd.
Golden ears hear Delta-Sigma DACs only sound good these days when you convert your PCM to dsd with high-end software players because it is the claimed native format for these chips. You need dsd recordings to get out the best, PCM doesn't cut it.
R2R brings life back into PCM and makes it competetive again.
In reality both systems exceed existing recordings by far.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!
 

Re: Multi-bit DACs

Reply #36 –
Quote from: pelmazo on
Quote from: eagleray on
1. Hard to drive headphones in which case a DAC and a headphone amp are purchased, often as a single unit.  Most likely symptom is low volume.  I suspect this is the source of most happy purchasers.
The problem here isn't that the headphone is hard to drive. Higher impedance headphones are typically quite easy to drive, except that they need more voltage than what a typical PC soundchip produces. Headphones with 600 Ohm impedance are so easy to drive that they can work off a conventional professional line output, but they need professional voltage levels. Headphones suited for the low voltages of PC soundcards and portable players typically have much lower impedances, which means (in the usual sense of the word) that they are harder to drive.

That's an entirely semantic argument.  The fact is high impedance phones don't produce enough volume off the audio sections of many motherboards and notebooks.  You can call that hard or easy, but the result isn't satisfying.  The professional voltage levels you refer to are not available in many consumer products.  The result is the same, one goes off in search of an alternative solution.  At any rate, it isn't the DAC, it is the compatibility of the headphone amp.

Re: Multi-bit DACs

Reply #37 –
Quote from: greynol on
Well hey, people eagerly ran out and bought NOS DACs during their resurgence.  Why not some other throwback design?

I wonder if it will sound better if laid upside down?

I don't know about turning the DAC upside down, but I am heading straight to my den and will try to listen to music while I am upside down.

Re: Multi-bit DACs

Reply #38 –
If a DAC works better upside down then it's a piece of garbage with either a very dodgey solder job or has serious thermal problems due to very bad design that causes things to overheat to the point of malfunction (hopefully not a fire).

It should work the same if it's right side up or upside down.

Re: Multi-bit DACs

Reply #41 –
To put a few things into perspective:

1. Classic R2R multibit DACs actually were used in CD players well into the '90s. This was because 1-bit DACs at the time still had some hurdles to overcome.

First of all, all kinds of DACs have a penchant for outputting ultrasonic garbage that may upset the following stages (read: cause intermodulation distortion). R2R DACs are famous for their glitches, but 1-bit delta-sigma with heavy noise shaping is especially bad. Extensive filtering and/or fast-slewing opamps are required to cope with this. If you were ever wondering why sample circuits liked to use OPA2604s as I/V at the time (despite them not being very good performers on typical +/-12 V), these were a fast-slewing FET-input type, read quite good RF immunity.

In addition, pure 1-bit converters also tend to be terribly sensitive to clock jitter, and only the introduction of synchronous switched-capacitor filtering in the early-mid '90s (and subsequent advances in the technology) made them robust to varying degrees. Jitter could otherwise fold back into the audio band and degrade SNR among other things.

2. Multi-multibit delta-sigma converters were pretty much the last "next big thing" in audio converters. These employ technology that switches between multiple few-bit converters (maybe 4 bit, generally imperfect) in such a way that turns their quantization noise into random noise and linearizes the whole shebang. Their main advantage next to stability is higher inherent converter SNR, so you need less noise shaping to achieve the same output SNR, making them either more output-stage-friendly or achieving higher SNR than possible otherwise. The entire success of Wolfson Micro was built on converters like that.

As you might gather from the above, discussing DAC technology in isolation is of limited use. Whether a DAC performs well ultimately comes down to implementation - the ODAC shows that you can wring good performance from a part that's only moderately high-strung, while any number of China eBay DACs with fancy chips but barely cracking the 100 dB DR demonstrate the opposite.

BTW, the D03K is quite borderline. The LMV358 low-power opamp in the output stage is pretty meh (1 V/µs, 1 MHz GBW, modest current capability) and would need a lot of a following wind, and I don't think the little Cirrus DAC is a miracle in jitter rejection. It's probably good enough, but I'd generally prefer your average Realtek onboard audio unless afflicted by ground loop noise...

Re: Multi-bit DACs

Reply #43 –
Quote from: greynol on
Well hey, people eagerly ran out and bought NOS DACs during their resurgence.  Why not some other throwback design?

I wonder if it will sound better if laid upside down?

What's wrong with an NOS DAC?  New Old Stock products are a great deal.  You get a manufacturer's warranty at a discounted price because something has been discontinued.  I love buying NOS.\

:-)

Re: Multi-bit DACs

Reply #44 –
Quote
What's wrong with an NOS DAC?  New Old Stock products are a great deal.  You get a manufacturer's warranty at a discounted price because something has been discontinued.  I love buying NOS.\
As a consumer or hobbyist, it's fine.   But, as a manufacturer you generally don't want to design-in obsolete parts...    You may end-up with a shut-down assembly line or you may end-up paying a high price for parts that are in short supply.  And you can get into a situation where the products in the field are not repairable

Of course there can be exceptions...   If you know you are only going to build 1000 or 10000 units and then re-design, and you can buy (or reserve) all of the parts you need in advance, you may be OK.  

But as a rule, design engineers hate using obsolete parts and production planners and production managers hate it too.

Re: Multi-bit DACs

Reply #45 –
Quote from: apastuszak on
What's wrong with an NOS DAC?
Nothing if they are designed competently (far from a given in the boutique market).  Then there are the nutjobs who want one without a reconstruction filter.

Re: Multi-bit DACs

Reply #46 –
I guess everyone missed my pun there.  In the retail world NOS means "New Old Stock."  So, as a retailer I have a brand new in box 1985 Rolex.  It would normally sell for $5,000 if it was a 2016 Rolex, but since I just want to recoup my loss, I sell it for $3,000 as 'New Old Stock.'  Or selling a discontinued model of something that's never been opened.

Your NOS obviously means "non oversampling" DAC, something I know nothing about.  If you could provide me some links, I would like to research them.

Re: Multi-bit DACs

Reply #47 –
A NOS DAC simply outputs a stairstep waveform following the input sample values, zero-order hold. There is no extra interpolation as in calculating in-between sample values. That's why you get droop, i.e. significant FR roll off towards Fs/2. Audiophiles call this roll-off "silky, non-digital highs" when playing CDs (44.1 kHz). :P
Without an additional (analog) anti-imaging filter, you will also get high frequency images potentially harming or degrading the performance of downstream equipment.


As for multi-bit, stephan_g explained it pretty well.  Multi-bit sigma-delta is where it's at. 1-bit/DSD have serious problems.
"I hear it when I see it."

Re: Multi-bit DACs

Reply #48 –
One "bad" solution for those mid-80s releases where the treble has been exaggerated, or perhaps a possible reason as to why those releases got that way?

The true cure of course was to reissue them with different EQ and then crush their dynamics as an added bonus.

I guess they plan on saving the least processed for hi-re$.  Sorry that the best available source tapes now have audible degredation, ignoring that they were never that hi-fi to begin with.