Digital Drive Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it.

---

[ Follow Ups ] [ Post Followup ] Thread: [ Display  All  Email ] [ Digital Drive ]

---

[ Alert Moderator ]

---

RE: Bits of Resolution!

Posted by Thorsten (M) on October 19, 2015 at 08:27:15

In Reply to: RE: Bits of Resolution! posted by J. Phelan on October 18, 2015 at 09:47:48:

Hi,

> Factor in noise, it looks like no DAC is even 16 bits in resolution.

DAC's that offered 20 Bit real performance did exist at some point in time... 16 Bit, even a TDA1541A (which is 25 years old) does that up to -3dBFS or better if you get double crown.

> But DSD is not 1-bit - this myth still persists in audio.
> It is a marketing term by Sony to describe anything from 4 to 16 bits.

You are incorrect. DSD as a format has a clear and unambigous definition of using 1 Bit Data sampled at 2.822MHz using a specifc noise shaper algorithm.

Since then we have seen commercially a push to "superspeed DSD", aka DSD128/DSD256 and even DSD512 (good idea if you ask me)

Some DSD editing systems (by Philips and NPC/Sony) used internally a conversion from single bit DSD to 8 Bit at the sample rate for applying effects and edits, after which the signal was re-noiseshaped back to 1 Bit/2.822MHz.

But there has however never been any standardised and available quality audio format (not even for internal use by a large company) different from classic PCM (16-32 Bit at 32kHz-768kHz) and DSD and Superspeed DSD (1 Bit at 2.822MHz to 24.576MHz).

So (say) 8 Bit at 5.6MHz does not exist, even if it might be an excellent idea and easy to realise with modern high speed commodity 8-Bit Pipeline ADC's and Multibit DAC's. Heck, pushing the sample rate to 50MHz at 8 Bit would still not break the bank either.

FWIW, sampling at 8 Bit (256 Level) and 50MHz(ish) (2048 FS) is good enough for 18 Bit ENOB with 88/96kHz sample rate using no noise shaping, just limiting measurement bandwidth to 40kHz or so.

For "High End" we could make the 8-Bit resistor ladders from Vishay Bulk Foil (or carbon composition) resistors - to taste.

> Which is what actual "DSD" recorders worked at. Starting with 4, then 8,
> now twice that.

You are incorrect.

Commercial DSD recorders (few as there are and ever where) all operate at 1 Bit. Some offer sample rates of as high as DSD256, or you might say quad speed dsd, but they still record just 1 Bit.

> 1-bit DACs are nothing new - we had these in the 1980s (w/ noise shaping)
> called "Bitstream".

Correct. Bitstream ADC's and DAC's where introduced to reduce cost. The earliest worked at 1 Bit and 2.822MHz modulation with 5th order or lower modulators.

When both the bitstream ADC's and DAC's became widespread it was found that the conversion from 1 Bit / 2.822MHz in commercial ADC's to 16/44.1kHz impacted sound quality negatively and the reverse process in the DAC (back from 16 Bit PCM to DS) made things worse.

BTW, Bitstream was actually the Philips Trademark for their Delta Sigma (or Sigma Delta) system - I always insisted it should be pronounced "Bitscream". Others had other names. Cirrus Logic stuck with "Delta Sigma" (pronounce "Delta Stigma"), but we were also forced to listen to such delightful concoctions as Matsushita's "MASH" (pronounce "MUSH").

There were more but thankfully I forgot about them over time. With hindsight it was probably more a question of inadequate hardware based filters and modulatoers that made the sound so bad, but boy it was really awful.

Anyway, with Delta Stigma/Bitscream/MUSH ADC's some bright spark had the idea to just bypass the PCM conversion stage and to goose up the modulator order a bit more for more Dynamic Range. In an DS ADC the modulator is primarily analogue hardware and pretty decent performance may be achieved by just throwing money at parts. Then the bitstream recording can lie on ice until DSP improves... Think Cryonics for terminally ill.

It was originally meant to simply record the "raw" signal from the ADC so that future improvements in DSP could access the raw ADC data, not one messed up by poor quality DSP.

When Sony & Philips faced a major cut in revenue streams due to the expiration of CD related patents and licences, they figured they could repackage this tech into a new "SUper Audio CD" and do another of the CD cashcow gig. We know how that one went.

> And I could be wrong, but I don't think PS Audio is re-formatting to PWM.
> Making it a very different idea.

You could be wrong indeed.

Just as you are not correct in claiming DSD can be anything at any wordlength, DSD is strictly 1-Bit.

Maybe worth reading what the manufacturer has to say about their product, I am sure you can find their pages... It is quite explicit about what happens inside the box.

Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

This post is made possible by the generous support of people like you and our sponsors:

Atma-Sphere Music Systems, Inc.

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2025, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---