R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 2 [All]   Go Down

Author Topic: curious DSD question...  (Read 5003 times)

Adam Dowling

  • Newbie
  • *
  • Offline Offline
  • Posts: 13
curious DSD question...
« on: April 23, 2010, 12:39:07 AM »

While I believe I understand the basic principles of DSD (please correct me if I'm wrong  Smile ), i have one nagging question:

What is the minimum sample count required for a full wave cycle, or more simply, what is the highest Nyquist frequency possible?

As I understand it, a DC steady state signal is (almost) accomplished by a rapid toggling of 1 & 0 so the Nyquist can't relate directly to the sample rate in the same way a 24/96 system will.

How fast are those damn 1's and 0's?
Logged
- Adam

biigniick

  • Newbie
  • *
  • Offline Offline
  • Posts: 45
Re: curious DSD question...
« Reply #1 on: April 23, 2010, 02:31:53 AM »

Adam Dowling wrote on Thu, 22 April 2010 23:39

What is the minimum sample count required for a full wave cycle, or more simply, what is the highest Nyquist frequency possible


the nyquist criterion still applies. . .

Adam Dowling wrote on Thu, 22 April 2010 23:39

How fast are those damn 1's and 0's?


2.8224 MHz is the sample rate.

because the sample rate is so high, the anti-aliasing filters can be a simpler, more elegant design.  there isn't an exact number of samples that have to make a waveform, but i think there is something like a 70% duty cycle limit (someone that knows more about DSD could elaborate).  there is also a 'loudness wars' like trend with over-modulating the DSD waveform (also, i'm not exactly sure on the tech specs).

i'm sure someone else will chime in soon.

- nick

compasspnt

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 16266
Re: curious DSD question...
« Reply #2 on: April 23, 2010, 08:33:17 AM »

biigniick wrote on Fri, 23 April 2010 02:31

2.8224 MHz is the sample rate.



Standard, but KORG has their own (dff) version @ 5.6448 MHz.
Logged

Adam Dowling

  • Newbie
  • *
  • Offline Offline
  • Posts: 13
Re: curious DSD question...
« Reply #3 on: April 24, 2010, 01:02:28 AM »

Thanks Nick and Terry.

it is the Korg products that initially caused me to muse about frequency limit when I saw the different sample rate options.

If you consider the 2.8Mhz rate, that's a Nyquist freqency of 1.4Mhz.  Since the bits toggle to imitate DC and to manipulate amplitude, the toggling must not result in much movement at high rates.

I've seen mention of a 100Khz frequency range so I guess there are diminishing returns in amplitude after that? A kind of "slew rate" to the speed at which the bits rise and fall?  

Anyway, I appreciate the info.  I don't like using anything without some grasp of the principles.  It actually bothers me a little not knowing how the automatic transmission works in the car!
Logged
- Adam

compasspnt

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 16266
Re: curious DSD question...
« Reply #4 on: April 24, 2010, 11:17:03 AM »

Adam Dowling wrote on Sat, 24 April 2010 01:02

It actually bothers me a little not knowing how the automatic transmission works in the car!



Careful, you'll get Keith started...
Logged

Adam Dowling

  • Newbie
  • *
  • Offline Offline
  • Posts: 13
Re: curious DSD question...
« Reply #5 on: April 27, 2010, 09:34:18 AM »

compasspnt wrote on Sat, 24 April 2010 12:17

Adam Dowling wrote on Sat, 24 April 2010 01:02

It actually bothers me a little not knowing how the automatic transmission works in the car!



Careful, you'll get Keith started...



And I don't know if Brad wants to moderate the PSW Automotive forum  Laughing
Logged
- Adam

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: curious DSD question...
« Reply #6 on: April 28, 2010, 02:58:28 AM »

The sampling theorem still holds, so it is 1.4MHz in principle. However, that frequency range is already occupied by one big whopping cyclical feature called the dominant tone. Depending on the situation (signal history) the modulator spends most of its time going "1010101010" or "110011001100" at low signal levels and alternating between them if there is a small DC offset present. So you'll get a spike at fs/2 and sometimes another one at fs/4. As soon as there is a signal these tones will drop in value. The fs/4 basically disappears and the fs/2 becomes frequency modulated by the absolute value of the input voltage. Notwithstanding all that you could put in a tone at a very high frequency and if it's big enough to stand out from all the rest you'll see it. The bandwidth of DSD is indeed 1.4MHz, but there's not much sense in trying to put any actual signal there because of all the noise.

The question then turns to useful bandwidth. The quantisation noise is low up to 20kHz and then rises rapidly. If no lowpass filter is used the HF noise will slew-limit amps, mislead compressors and whatnot. How high you should cut off depends on what you plan to do with the signal down the line. 100kHz is definitely too high. Few power amps handle this without demodulating at least some of the HF noise. A realistic value is 50 to 60kHz.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

dcollins

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2815
Re: curious DSD question...
« Reply #7 on: April 28, 2010, 03:17:15 PM »

bruno putzeys wrote on Tue, 27 April 2010 23:58


The question then turns to useful bandwidth. The quantisation noise is low up to 20kHz and then rises rapidly. If no lowpass filter is used the HF noise will slew-limit amps, mislead compressors and whatnot. How high you should cut off depends on what you plan to do with the signal down the line. 100kHz is definitely too high. Few power amps handle this without demodulating at least some of the HF noise. A realistic value is 50 to 60kHz.


Sony frequently advertised a DR of 120dB with an FR of 20-100k......



DC

Adam Dowling

  • Newbie
  • *
  • Offline Offline
  • Posts: 13
Re: curious DSD question...
« Reply #8 on: April 29, 2010, 01:47:31 AM »

dcollins wrote on Wed, 28 April 2010 16:17

bruno putzeys wrote on Tue, 27 April 2010 23:58


The question then turns to useful bandwidth. The quantisation noise is low up to 20kHz and then rises rapidly. If no lowpass filter is used the HF noise will slew-limit amps, mislead compressors and whatnot. How high you should cut off depends on what you plan to do with the signal down the line. 100kHz is definitely too high. Few power amps handle this without demodulating at least some of the HF noise. A realistic value is 50 to 60kHz.


Sony frequently advertised a DR of 120dB with an FR of 20-100k......



DC


I am mostly interested in DSD as a digital mixdown medium from an analogue source (console, summing mixer, etc) and as a delivery medium for mastering.  High frequency, high resolution and a system that more closely resembles electrical current.

Bruno: Is it the general high frequency content approaching 100Khz that would confound amps, comps, etc or DSD ringing noise specifically?  A conventional 24/192Khz has its aliasing filter at 96Khz for example.

I am also led to believe (by Korg) that their software can perform SRC down to conventional rates/depths in a unique way, creating better files than typical high-bit SRC.  Any insight there as well?

Logged
- Adam

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: curious DSD question...
« Reply #9 on: April 29, 2010, 03:05:53 AM »

Sony's manner of claiming DR and BW is disingenuous to be polite. Yes, you can have a frequency response up to 100kHz if the DAC post filter cuts off at that frequency. However, when measured from DC to 100kHz, SNR will then work out as about 30-40dB on account of the noise content above 20kHz. If your replay chain will handle it without causing noise fold-back it won't sound like 40dB because you only hear the bit up to about 20kHz and measured from DC-20kHz, an SNR of 120dB is indeed practical. I cannot help feeling uneasy about the fact that on the one hand they feel the need to claim a bandwidth up to 100kHz as if that were to be audible, while relying on the non-audibility of the noise beyond 20kHz in order to claim 120dB.

So far my feeling remains the same: 50kHz suffices. I'm not alone in this, nobody makes DSD gear with a 100kHz bandwidth. As far as the impact on amplifiers etc goes, it is this HF noise I'm speaking of. The ringing on PCM converters does not increase slew rate and causes no issues.

The situation as it stands now is this: if you are unsure about what format (PCM or DSD) to record or mix down in and your only interest is in sound quality, do not choose the format first. Under no circumstance should conceptual arguments come into the discussion, like how much it seems to resemble analogue etc. This is immaterial. A circuit that "has a certain philosophical rightness to it"* and that performs inadequately is, after all is said and done, still a circuit that performs inadequately. The only thing that counts is how close the output signal sounds to the input signal, taking into account other requirements like cost, portability, work-flow etc, to the extent that they are important for you.

So. If your workflow permits DSD, this only means that you have a handful more product offerings to consider compared to your PCM-only peers. Other than that, ignore the format and choose the converter or recorder based only on sonic transparency. The Korg box makes an excellent excuse for using DSD (usually to convert to PCM as soon as you can though) because regardless of format it is an unusually good sounding location recorder for the price. As non-portable items are concerned I can think of only one converter worth the hassle of recording in DSD but modesty prevents me from naming it (now there's a shamelessly immodest statement if I ever saw one). To use the Korg as a mixdown recorder strikes me as odd because there we're not interested in how good it sounds as an affordable portable recorder, but how good it sounds compared to anything else you'd normally use in this application.

In principle going from DSD to PCM or going from one sampling rate to another is a perfectible process. If Korg claims its DSD to PCM process is better than sample rate conversion they probably mean compared to hardware SRC's which have all sorts of practical limitations. But a good off-line converter like Izotope's SRC doesn't have those limitations and if Korg's is as good, well that's as far as it goes either way.


________________________________
*The person at whom this snipe is aimed knows who he is. Should you be reading this: get over it, man. I'm keeping a couple of beers chilled in the fridge for you to talk things over.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

Gold

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1453
Re: curious DSD question...
« Reply #10 on: April 29, 2010, 07:51:38 AM »

Someone is bringing in a Sonoma rig next week to cut from. I'm really nervous about the HF noise. Has anyone familiar with the systems noticed anything strange like random bursts of out of band noise? That will kill the cutter head. My LPF is only 18dB/oct.
Logged
Paul Gold
www.saltmastering.com

On the silk road, looking for uranium.

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: curious DSD question...
« Reply #11 on: April 29, 2010, 08:12:28 AM »

The HF noise is steady and even if the DAC is badly done it is still going to be more than 40dB below the signal, and that is not including your own LPF. I shouldn't worry.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

bblackwood

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 7036
Re: curious DSD question...
« Reply #12 on: April 29, 2010, 08:35:13 AM »

bruno putzeys wrote on Thu, 29 April 2010 02:05

In principle going from DSD to PCM or going from one sampling rate to another is a perfectible process. If Korg claims its DSD to PCM process is better than sample rate conversion they probably mean compared to hardware SRC's which have all sorts of practical limitations. But a good off-line converter like Izotope's SRC doesn't have those limitations and if Korg's is as good, well that's as far as it goes either way.

I've never used my Korg for anything other than playing back client's files at native rates so I'm not certain, but I'm pretty sure the Korg unit doesn't do any conversion in the box - you use some 'Audiogate' software to do it in non-real time on the host computer.

And as I've said before - I'm certain the Korg converters can be beaten by the likes of your own Grimm design, but even with that caveat it's still the best sounding delivery method I've ever had the pleasure of using. I wish they included an easy method of using outboard converters...
Logged
Brad Blackwood
euphonic masters

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: curious DSD question...
« Reply #13 on: April 29, 2010, 09:17:10 AM »

My point was that philosophical arguments pro/contra PCM or DSD are basically irrelevant. What matters is if the thing delivers sonically. If the Korg sounds that good, that is a sensible reason to use it. The fact that it uses DSD to store audio is not. Just my private crusade against "under the hood" sales pitches Smile
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

Adam Dowling

  • Newbie
  • *
  • Offline Offline
  • Posts: 13
Re: curious DSD question...
« Reply #14 on: April 29, 2010, 12:18:19 PM »

bruno putzeys wrote on Thu, 29 April 2010 10:17

My point was that philosophical arguments pro/contra PCM or DSD are basically irrelevant. What matters is if the thing delivers sonically. If the Korg sounds that good, that is a sensible reason to use it. The fact that it uses DSD to store audio is not. Just my private crusade against "under the hood" sales pitches Smile


The crusade is a valiant one! One always has to take manufacturers' claims as just that.

I admit that I was smitten with the actual concept of DSD. Bruno, your posts are excellent and give me a much better understanding of the process - Many thanks  Thumbs Up

I'm looking forward to trying one of the Korg units to put through the paces.  The field units are now discontinued but I believe they've dropped the price of the rack unit as a result.

The proof shall be in the pudding...
Logged
- Adam

Adam Dowling

  • Newbie
  • *
  • Offline Offline
  • Posts: 13
Re: curious DSD question...
« Reply #15 on: April 29, 2010, 12:24:56 PM »

bblackwood wrote on Thu, 29 April 2010 09:35

bruno putzeys wrote on Thu, 29 April 2010 02:05

In principle going from DSD to PCM or going from one sampling rate to another is a perfectible process. If Korg claims its DSD to PCM process is better than sample rate conversion they probably mean compared to hardware SRC's which have all sorts of practical limitations. But a good off-line converter like Izotope's SRC doesn't have those limitations and if Korg's is as good, well that's as far as it goes either way.

I've never used my Korg for anything other than playing back client's files at native rates so I'm not certain, but I'm pretty sure the Korg unit doesn't do any conversion in the box - you use some 'Audiogate' software to do it in non-real time on the host computer.

And as I've said before - I'm certain the Korg converters can be beaten by the likes of your own Grimm design, but even with that caveat it's still the best sounding delivery method I've ever had the pleasure of using. I wish they included an easy method of using outboard converters...


Brad, I now realize that it was your mention of the Korg in another post that caused me to look into it in the first place.  Thanks for that too.

Now how about that automotive forum? ... Razz
Logged
- Adam

TotalSonic

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3728
Re: curious DSD question...
« Reply #16 on: April 29, 2010, 12:29:32 PM »

Gold wrote on Thu, 29 April 2010 07:51

Someone is bringing in a Sonoma rig next week to cut from. I'm really nervous about the HF noise. Has anyone familiar with the systems noticed anything strange like random bursts of out of band noise? That will kill the cutter head. My LPF is only 18dB/oct.


Don Grossinger cut the ABKCO Rolling Stones reissues at Europadisk from a Sonoma workstation that was brought in just for the release.  If he kept it completely DSD to analog (which since this was coming from Bob Ludwig I'm pretty sure he would have heeded what I would likely imagine would be strict requests not to capture back to PCM) then the only LPF he would have had available would have been the HT75's in the Neumann desk.  It's even possible he left this off for the longer sides and just used the acceleration limiters in the SAL84 rack to deal with cutter head protection.  Maybe you could contact him directly to get advice on this as he's the only one I know of who has cut a lot of sides from DSD.

Best regards,
Steve Berson

Gold

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1453
Re: curious DSD question...
« Reply #17 on: April 29, 2010, 12:34:50 PM »

I have the MDS2 before the LPF. If it starts going crazy I'll know there is a problem. I won't drop the head until I'm sure...I won't have to operate the Sonoma rig.
Logged
Paul Gold
www.saltmastering.com

On the silk road, looking for uranium.

Sonovo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 267
Re: curious DSD question...
« Reply #18 on: April 29, 2010, 02:36:58 PM »

The Korg unit sounds outstanding, as I recently found out myself.

Using external (Meitner/Mytek) DSD converters isn't possible, but the built in ones seem plenty fine to my ears.

For external converter use, there is the TASCAM unit (DV-RA1000HD).

The Korg DSD-PCM conversion is software based, and actually sounds quite good. Haven't compared it to something like SaRaCon, but it sounds like Korg definitely put some effort into this one.

As pointed out, the portable Korg unit is now discontinued and heavily discounted. It's tempting to pick one up and see if it's as good as the rack mount model...

Cheers,
Thor
Logged

crna59

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 277
Re: curious DSD question...
« Reply #19 on: April 29, 2010, 05:44:36 PM »

Sonovo wrote on Thu, 29 April 2010 13:36

The Korg unit sounds outstanding, as I recently found out myself.

The Korg DSD-PCM conversion is software based, and actually sounds quite good. Haven't compared it to something like SaRaCon, but it sounds like Korg definitely put some effort into this one.

As pointed out, the portable Korg unit is now discontinued and heavily discounted. It's tempting to pick one up and see if it's as good as the rack mount model...
Cheers,
Thor


We have the MR2000s with AudioGate and Saracon DSD. We actually had the Saracon before the Korg. I'd have to say the AudioGate is very comparable, but Saracon has alot more filter options and such. We also have the Tascam unit that's used with the Meitner converters.

Korg has now come out with a portable replacement called the MR2.


Regards,
index.php/fa/14753/0/
Logged
Puget Sound Studios
Bruce A. Brown
Mastering & Post Production
Seattle, WA

It is easier to ask for forgiveness than to ask for permission! Buy! Buy buy....  

Pages: 1 2 [All]   Go Up
 

Site Hosted By Ashdown Technologies, Inc.

Page created in 0.117 seconds with 22 queries.