Digital, but not by the numbers - Page 33 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level

Digital Line Level DACs, Digital Crossovers, Equalizers, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 9th December 2012, 09:15 AM   #321
marce is offline marce  United Kingdom
diyAudio Member
 
Join Date: Jun 2007
Location: Blackburn, Lancs
Cadstar is one of the cheaper options, though Altium is good value at the moment.
Yes the higher end software is not cheap, and for a lot of designs overkill, but when you do a 12-24 layer design with say 10,000+ pins you appreciate it
Been thinking how I would do a digital design, well more an integrated design, but it would be by the numbers!!! First thing I would do is use a one box solution, then you only have one protective earth connection, thus avoiding in one go any earth loops..
More later, I'm suppose to be working!!
  Reply With Quote
Old 9th December 2012, 10:17 AM   #322
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
the lite+ version is 2700 British pounds...
  Reply With Quote
Old 9th December 2012, 12:06 PM   #323
jgazal is offline jgazal  Brazil
diyAudio Member
 
Join Date: Dec 2008
I have posted the following reasoning in this threard ("NOS DAC idea", started by Barrows and sponsored by Twisted Pear Audio):

Quote:
Originally Posted by jgazal View Post
Thread resurrection...

Suppose there is no need to stick to a certain media.

Then the audio chain starts with a multi-bit output ADC like Texas Instrument PCM4222. It has a 6-bit modulator output.

There is no fixed media, but there is still internet bandwidth. So the PCM4222 works at its minimum master clock input: 2.048 MHz.

What is the bandwidth of such file in Mbit per second?

So our code has 6 bit of quantization and 2.048 MHz of sampling rate. Not a novel idea as it seems like DSD-Wide.

That means quantization noise at higher frequencies then 1 bit DSD and Nyquist sampling images way after DXD (352.8 KHz) easing the analog filter requirements at both sides (ADC and DAC).

The low bit quantization (6-bit) and 2.048 MHz frequency also allows a discrete non-oversampling DAC. I suppose it is possible to use switches with low crosstalk at 2MHZ region and a lower number of Z-foil resistors in a logarithm R2R ladder.

This discrete DAC also allows a voltage output. In other words, no I/V converter or op-amps. Then we need a direct coupled buffer to keep the audio output impedance as low as possible.

Is this a good idea? Is it feasible? Is it affordable?

If the Z-Foil resistor ladder makes such discrete DAC too expensive, is it possible to use ES9018 with a lower 2.048MHz synchronous master clock (from TPA USB transport) and no oversampling filter?

Is it possible to transmit such 6-bit, 2.048MHz audio code through USB or Firewire?

Is it possible to implement a digital volume control using such 6-bit files. Does someone have any idea?

Last but not least, is there any computer software to convert DSD and DXD to a 6-bit, 2.048 MHz file while on-line providers do not adopt such format? If they ever adopt such format...

p.s.: I prefer a voltage DAC since I would like to use an electrostatic transducer... That means the whole chain works at voltage mode...
It seems that multi-bit raw output from sigma delta ADC's (usually 6 bits with samples as high as the master clock frequency allows) is usually processed by the IC itself, which delivers at its interface a 16-bit, 24-bit or DSD stream. TI PCM4222 is one exception.

Q.1 - What is the bottleneck in the audio chain preventing the use of such raw output as "distribution file"? Is it storage? Is it Internet bandwidth? Is it USB bandwidth?

Quote:
Originally Posted by abraxalito View Post
Charles has a valid point about resolution - why are S-D DACs marketed with the high resolutions that they claim for them? Since they all suffer from noise modulation - meaning the noise varies with the signal level - they cannot have a constant resolution. Resolution means the ability to resolve, tell the difference, between two things. But with an S-D this ability to resolve two different input codes depends on signal level. At lower signals, the resolution is higher than at higher levels.



The opamp non-linearity would be a different thing - it produces IMD products which are fairly well characterised in terms of their correlation with signal. S-D noise modulation is not, and the datasheets don't give this information in any form, apart from ESS who indicate how SNR changes with DC offset if I recall correctly. S-D incorporates feedback too - around a much more serious non-linearity than any I've seen in an opamp.
Quote:
Originally Posted by qusp View Post
not really, my comment simply related to just about every single piece of music will be sampled with SD already, so what can a non SD dac do about this problem of which you speak to make these distinct samples again? its a red herring

re the dac, I know that, but its still supposed to be R2R according to the claims. I guessed it at the 45 straight after it was released. not much can operate on such tiny voltage, tiny current and without large complexity thats impossible in a little portable.

anyway how about we leave it, nothing is going to change my mind here, my stance is the same as going in, just now its all been confirmed from the horses mouth.
Q.2 - Would a "discrete, 6-bit, R-2R ladder, with voltage output using foil resistors" also suffer with noise modulation? Would it also require such large negative feedbacks of Sigma Delta DAC IC's?

Q.3 - Converting and down sampling a 1-bit, 2.8224 MHz, DSD file to a 6 bit, 2.048 MHz file requires any mathematical estimation? Or is just a matter of transposition?

I see bit depth and sample frequency as axis of the same component (resolution or mbytes/s).

If I were allowed to use a gross analogy, it would be the relativeness of time and space. Time and space are axis of the same component, which is the speed of light. Time and space are relative, but you cannot change the speed of light.

So you can reduce bit depth (space) and increase sample frequency (time) and vice-versa, and you will always obtain the same "resolution", "dynamic range" or "mbytes/s" (the speed of light...) .

But the "resolution", the "dynamic range" or "mbytes/s" of your master recording file must remain the same of your distribution file.

Otherwise, you will need digital filters. Nothing wrong using them to convert your raw ADC output to lower resolution files. But digital filters seems to be more complicated when used at the playback side, which usually a has distribution file with lower resolution then the digital filter output. That means such digital filters must rely on estimation algorithms.

Even if master recording files and distribution files have the same resolution, but different bit depths and sampling rates, I suppose there are side effects of translating bit depth and sampling rate.

A raw output from a 1 bit sigma delta that is sampled with a high jitter clock may result in a 6-bit file with wrong voltage levels since the sample was made in the wrong moment in time, when the wave energy was higher or lower. The same thing happens with a multi sigma-delta raw output that is also sampled with a high jitter clock. Such raw output file will be translated into a 24 bit file that also has wrong quantization levels. But at least the 2^6 different voltage levels as the raw output reduces the magnitude of such quantization error.

So I believe that not only the resolution of your master recording file must remain the same of your resolution file, but also its bit depth and its sample rate.

It is just a matter of choosing a viable standard for the benefit of the whole industry and consumers.

I understand that reducing the bit depth of you master recording has its virtues (storage, bandwidth) and compromises (quantization noise). The same goes when increasing sample rate of your master recording, which has virtues (easing analog filters) and compromises (storage, bandwidth).

IMHO, and is just an opinion, since I have no background to support the idea, the audio chain that I described in that post is a good compromise.

If a master recording file with enough resolution (mbytes/s) is also used as distribution file, there is no need for estimation oversampling, in other words, digital filters.

Sure it would not be "the best" playback hardware for the vast majority of distribution files available nowadays (16 bit or 24 bit, with moderate sampling rate). Converting such files to the 6 bit, 2.048 MHz, supported by such discrete DAC, would still require a digital filter with an estimation algorithm (at least a computer software can convert the files asynchronously with all the CPU computation power, using in several iterations) and they would also had the quantization inaccuracies related to the jitter at the raw ADC output.

But it will be the best for everybody if the distribution file stick to 6 bits and ~= 2.048 MHz:
Quote:
Originally Posted by qusp View Post
what will be interesting is when multibit DSD becomes more available; as in available anywhere apart from a couple of insane hobbiests programming FPGAs
So there is a compromise between analog filters, digital filters (with estimation algorithms, at least in the DAC side) and the imbalance with the master recording distribution file in terms of resolution, bit depth and sample rate.

Some people give up of the digital filters, which hardens the burden of analog filters.

Some people give up of analog filters, but then they need digital filters with estimation algorithms or they leave the playback chain under the influence of high frequency energy (images from 16 bit or 24 bit files or quantization noise from DSD).

Q.4 - Why not to reduce the resolution of the raw output of a widely disseminated sigma delta ADC, using it as a master recording and distribution file without any estimation filter?

The resolution of the raw output must be low enough to allow its use as distribution file and high enough to avoid playback digital filters, easing the work of playback analog filters.

I think the 6 bit, 2.048 MHz file would be a good compromise to be adopted as standard for all the audio chain (raw ADC output, master recording files and distribution files).

Someone would argue instead: "so why not using 16 bit, 24 bit files as standard and then just increasing the sample rate?"

It seems expensive to manufacture IC ADC's and DAC's with large R-2R ladders. 6 bit is definitely the best compromise at both sides (ADC and DAC). It allows mass produced IC's, but also discrete designs with Z-foils.

I had no response at that thread.

Q.5 - Did I say something really wrong?

Forgive me the long post.
  Reply With Quote
Old 9th December 2012, 08:19 PM   #324
fas42 is online now fas42  Australia
diyAudio Member
 
fas42's Avatar
 
Join Date: Jun 2012
Location: NSW, Australia
Blog Entries: 11
Digestion time is probably needed ...

The fundamental idea of having access to the raw data sounds good to me; after all, digital cameras have gone strongly down this road, and it's become a big thing in that field ...

Frank

Last edited by fas42; 9th December 2012 at 08:22 PM.
  Reply With Quote
Old 11th December 2012, 12:33 AM   #325
jgazal is offline jgazal  Brazil
diyAudio Member
 
Join Date: Dec 2008
On a second thought, storage may still be the bottleneck...

Click the image to open in full size.
  Reply With Quote
Old 11th December 2012, 05:56 AM   #326
Julf is offline Julf  Europe
diyAudio Member
 
Join Date: Oct 2011
Location: Amsterdam, The Netherlands
Quote:
Originally Posted by jgazal View Post
On a second thought, storage may still be the bottleneck.
On the other hand, PCM formats compress pretty well using lossless compression - have no idea about the compression properties of DSD.
  Reply With Quote
Old 11th December 2012, 09:19 AM   #327
erin is offline erin  Australia
diyAudio Member
 
Join Date: Nov 2006
Location: http://www.makeitpossible.com/
Quote:
Originally Posted by fas42 View Post
Well, this might sound bizarre but the tiny Cowon music player of a friend on mine, using a Wolfson DAC, running on battery power easily outdid a highly tweaked low cost TT, and a reasonably tweaked Quad CD player.

Frank
Thank you for your comment. I think the iphone and some ipods also use a Wolfson DAC. I suspect there are a few factors at play here.
Battery power, reading the data from flash memory, and short track lengths, I think are contributing factors for good sound.

Quote:
Originally Posted by DF96 View Post
erin:
In response to your surprise at my comment I was going by what had been said in this thread. I had no other information to go on. The information you now draw my attention to would not have changed anything.
The original poster is an engineer. To what level of expertise, no one here will know, but the proof would be in the listening.

I appreciate his observation that there are many people who find even the best engineered delta sigma designs to have a sound quality which annoys or offends.

I unfortunately fall into this category of people, and this is why I was interested in this thread.

A person who has never had any problems with digital sound might have difficulty understanding the problem that "digititis" causes some people.

But if we can trust that this is an issue for some people, and then find out exactly what the cause of it is, and correct the issue, then I think this would make a vast improvement to digital audio for everybody.

Quote:
Originally Posted by TNT View Post
Thanks for the suggestion. It looks pretty. I doubt I will ever get to hear one.
Is there anything more affordable that comes even close?
__________________
Listening is best done using a pair of ears.
  Reply With Quote
Old 11th December 2012, 09:55 AM   #328
fas42 is online now fas42  Australia
diyAudio Member
 
fas42's Avatar
 
Join Date: Jun 2012
Location: NSW, Australia
Blog Entries: 11
Quote:
Originally Posted by erin View Post
But if we can trust that this is an issue for some people, and then find out exactly what the cause of it is, and correct the issue, then I think this would make a vast improvement to digital audio for everybody.
At a simplistic level the problems of delta sigma appear to lie in 2 key areas: they need long conditioning times, of the order of hours before properly coming on song, when started from cold; and, they are very susceptible to interference.

This has been my experience, with units some years old. The latest offerings may be better, perhaps ...

Frank
  Reply With Quote
Old 11th December 2012, 10:33 AM   #329
diyAudio Member
 
Join Date: Jun 2009
Quote:
Originally Posted by fas42 View Post
At a simplistic level the problems of delta sigma appear to lie in 2 key areas: they need long conditioning times, of the order of hours before properly coming on song, when started from cold; and, they are very susceptible to interference.

This has been my experience, with units some years old. The latest offerings may be better, perhaps ...

Frank
and your explanation for that is what ??
  Reply With Quote
Old 11th December 2012, 10:44 AM   #330
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
Quote:
Originally Posted by Trevor White View Post
and your explanation for that is what ??
Trevor, at least in the case of ES9018/2 this is completely legit, particularly if running a very low DPLL bandwidth setting, they take some time to warm up...weird I know but its true. It happens across commercial and DIY dacs. some will set the DPLL bandwidth to 'best' to avoid this (this is the default) and in this setting it will simply seek the lowest bandwidth needed to get a stable lock.

using a lower bandwidth than 'best' is preferred and during this 'warm up' time it seems effected by ripple on ground, fridges starting up etc. and will drop lock, pause etc. I normally just let it warm up before use if i'm running in asynchronous mode, I havent tested to see if there are other more subtle audible issues. running synchronously its not an issue.

Last edited by qusp; 11th December 2012 at 10:47 AM.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Tubes Numbers kmscouts Tubes / Valves 3 30th September 2012 04:19 AM
Confused by the numbers alexmoose Tubes / Valves 52 7th August 2006 03:06 AM
Still can't get reasonable numbers mashaffer Multi-Way 13 10th July 2006 04:55 PM
Where did these EL84 numbers come from? Sherman Tubes / Valves 2 13th June 2005 06:53 PM


New To Site? Need Help?

All times are GMT. The time now is 12:14 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2