[tin-ear]

I thought that this was a Schiit thread in an audio forum. I just slogged through four pages looking for Schiit, or at least audio. I guess that I am in the wrong place.

J.P.

You never know. Sometimes Schiit will introduce a new product, a review of a product will be published/posted, someone will post something fun, scandalous, or just interesting, the International Space Station will go retrograde in Scorpio, whatever, and then this forum will have a flurry of replies, comments, thoughts, flame wars (rarely more than minor) all..., well, mostly, ...okay, at least somewhat about Schiit products, or the company, or Schitt projects--you get the idea.
In the meantime, be careful. Since hanging out here I have been inspired and bought two headphones and made 2 ventures into coffee making gear. The later at least gives me a mostly better cup of coffee than buying one, and cheaper. Although McDonald's and Jack 'n the Box compete on price with the senior discount. (I have other reasons for avoiding those venues and a cardiologist to prove it).

 

[artur9]

Anyone think Schiit will make an all-digital preamplifier with stereo subwoofer outputs?
I mean, digital input and digital output to feed into Schiit DACs.

They have analog preamps covered.

Another question: DAC before digitizing preamp or DAC after digitizing preamp? I've heard some say that DAC before a preamp that does A/D sounds better.

 

[Pietro Cozzi Tinin]

Another question: DAC before digitizing preamp or DAC after digitizing preamp? I've heard some say that DAC before a preamp that does A/D sounds better.

Just try and listen and pick your best.
I always put the DAC after sources and then the pre-amp picks up for the poweramp.

 

[bosiemoncrieff]

Okay the espresso machine is back up and running, but the beans have been sitting (sealed) since May. The waste of it breaks my heart, but I don’t think they’re ever going to make great espresso. I’ll futz with the dose and grind and see if I can get something halfway decent - at least with grapefruit juice and elderflower syrup, if not without.

 

[ev666il]

Question for @Baldr and/or the digitally savvy: Do "multi-bit" D/S ADCs and DACs always ditch all original samples? Or do they retain the original samples for the actual bits they have (usually 5-6 in modern chips, if memory serves) and only approximate the rest?

 

[US Blues]

Question for @Baldr and/or the digitally savvy: Do "multi-bit" D/S ADCs and DACs always ditch all original samples? Or do they retain the original samples for the actual bits they have (usually 5-6 in modern chips, if memory serves) and only approximate the rest?

Dearly Beloved, we are gathered here today to lay to rest the original digital information stream that was cruelly disposed of by a Delta Sigma DAC. This data stream was young, vibrant, and full of great music. Discarded in the prime of life and replaced by a mathematical model that bears only some small relationship to music. Our dear friend @Baldr is here to say a few words.

 

[yonson]

Question for @Baldr and/or the digitally savvy: Do "multi-bit" D/S ADCs and DACs always ditch all original samples? Or do they retain the original samples for the actual bits they have (usually 5-6 in modern chips, if memory serves) and only approximate the rest?

IIRC Schiit MB DAC's retain all original samples, however I'm no expert, and I could very likely be wrong. Hopefully @Baldr can set us straight.

 

[FLTWS]

I kinda remember it being mentioned on several occasions by @Baldr and other posters in this thread that that was the special part of multibit, it retains all original info.

 

[ev666il]

Schiit's MB DACs do indeed, and so did Mike's GAIN ADC. My question has to do with D/S DACs and ADCs that have 5-6 "true" bits (as opposed to the first D/S chips that were 1-bit) and whether they retain the original samples for those bits.

 

[earnmyturns]

Question for @Baldr and/or the digitally savvy: Do "multi-bit" D/S ADCs and DACs always ditch all original samples? Or do they retain the original samples for the actual bits they have (usually 5-6 in modern chips, if memory serves) and only approximate the rest?

At issue is not the DAC itself but the upsampling method. For D/S DACs, oversampling happens when the PCM samples are converted to a D/S encoding (eg. 1 bit DSD) at the typically much higher sample rate used by the DAC chip to reconstruct the analog signal. Pure multibit DACs like Schiit's also use oversampling to do the conversion at a higher sample rate. In both cases, the higher sample rate places reconstruction artifacts much further up in frequency, making it easier to filter them out with an analog filter without interfering with the audible range. So, what's this oversampling? It's basically "inventing" new samples in between the ones in the original digital stream that "smoothly" interpolate between the original samples. Now, this "smoothly" hides a lot of subtlety, which is why many such oversampling filters have been designed. The majority of oversampling filters used in practice achieve their "smoothness" and efficient computation at the expense of not keeping the original samples (similar to drawing a trendline through a cloud of datapoints that does not go through any of the actual datapoints). Schiit's proprietary filters manage to preserve the original samples, and just add new ones in between. Whether that's the crux of the matter, or just a byproduct of the design, the fact is that they sound wonderfully more convincing than most of the others I've heard. (I have a thing for non-oversampling (NOS) DACs in some situations, which bypass this whole rigmarole, but that's for another time.)

 

[ev666il]

Thank you, @earnmyturns .

So when in possession of a Schiit MB DAC, there are two possible scenarios:

• If the ADC also maintained the original samples (that would essentially be Mike’s GAIN system; I’m unaware of any others), you get all the original samples straight. This is a best case scenario and should give the most natural sound.
• If the ADC discarded the original samples, the digital stream is an approximation. Schiit’s MB DAC retains the approximated samples instead of approximating them again

Do I get this right?

On a full D/S chain, you get an approximation of the approximation, which is a worst case scenario.

 

[porchwizard]

Thank you, @earnmyturns .

So when in possession of a Schiit MB DAC, there are two possible scenarios:

• If the ADC also maintained the original samples (that would essentially be Mike’s GAIN system; I’m unaware of any others), you get all the original samples straight. This is a best case scenario and should give the most natural sound.
• If the ADC discarded the original samples, the digital stream is an approximation. Schiit’s MB DAC retains the approximated samples instead of approximating them again

Do I get this right?

On a full D/S chain, you get an approximation of the approximation, which is a worst case scenario.

An ADC (Analog to Digital Converter) is creating original samples from an analog input, unless I've misunderstood something. Nothing to preserve.
The Schiit multi-bit DACs (Digital to Analog Converters) operate at 192/196k or 384/392k and if being fed lower rate samples (like 44.1k from a CD) the stream is converted to the higher rate. It is in this conversion (the digital filter) that the original, incoming, samples are preserved and additional samples are created in between to create the higher rate stream. {Most|Many|All} other multi-bit DACs compute all new samples.

 

[ev666il]

An ADC (Analog to Digital Converter) is creating original samples from an analog input, unless I've misunderstood something. Nothing to preserve.

You make a fair point.

However this leave me a little confused as to the benefits of Mike's GAIN ADC compared to a D/S ADC. The GAIN system was basically a reverse True Multibit converter, but if the point of True Multibit is to preserve the samples and ADCs aren't discarding samples to begin with, what does the reverse True Multibit do? I'm genuinely curious.

 

[US Blues]

You make a fair point.

However this leave me a little confused as to the benefits of Mike's GAIN ADC compared to a D/S ADC. The GAIN system was basically a reverse True Multibit converter, but if the point of True Multibit is to preserve the samples and ADCs aren't discarding samples to begin with, what does the reverse True Multibit do? I'm genuinely curious.

I don't know that the term "true multibit" applies to the GAIN ADC in same manner it applies to a multibit DAC. @Baldr explained the GAIN system several years ago in this thread. He designed it with Nelson Pass, so the whole package was special and not just the ADC portion of the system.

 

[ev666il]

@Baldr explained the GAIN system several years ago in this thread. He designed it with Nelson Pass, so the whole package was special and not just the ADC portion of the system.

I agree; I was merely curious about this particular facet of the GAIN system as mentioned by @Baldr in this older post:

At the time, it was a really expensive multibit 17 bit A/D converter which converted at 8x either 44.1 or 48KHz 352.8 or 384KHz – double the MQA rate. It also featured an inverse megaburrito filter of the time. Which gave us the sound.

Isn't the point of the megaburrito filter that it retains the original samples?