Filter brewing for the Soekris R2R - Page 23 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

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 27th February 2015, 08:02 AM   #221
TNT is offline TNT  Sweden
diyAudio Member
 
Join Date: Apr 2003
Location: Sweden
Well one can choose to make it 2kHz. In some sense the 22,05 is fixed but you could back down to say 18kHz if 20 isn't sacred to you

But I do agree - where its done is not so important.


//
  Reply With Quote
Old 27th February 2015, 08:16 AM   #222
diyAudio Member
 
Join Date: May 2005
Location: Melbourne
Blog Entries: 1
Quote:
Originally Posted by TNT View Post
Well one can choose to make it 2kHz. In some sense the 22,05 is fixed but you could back down to say 18kHz if 20 isn't sacred to you

But I do agree - where its done is not so important.


//
Point is that you still face the same issues of selecting which frequencies and type of filter is used to remove (or not) the images whether you filter on computer or in FPGA.

The other thing is that the recommendation to use PC-based applications to upsample on the fly on a computer assumes that everyone is using a computer as their only source, which is not always the case.

20kHz, 18kHz, 16kHz, 14kHz - pick your poison.

I've been reading Peter Cravens papers on Apodizing filters and the assumption is that frequency should be at worst less than .1dB down at 20kHz. That is one approach I guess.

Last edited by spzzzzkt; 27th February 2015 at 08:25 AM.
  Reply With Quote
Old 27th February 2015, 08:24 AM   #223
TNT is offline TNT  Sweden
diyAudio Member
 
Join Date: Apr 2003
Location: Sweden
He musts have very good hearing.

//
  Reply With Quote
Old 27th February 2015, 08:37 AM   #224
diyAudio Member
 
Join Date: May 2005
Location: Melbourne
Blog Entries: 1
Quote:
Originally Posted by TNT View Post
He musts have very good hearing.

//
Sorry ±0.2dB, but this was discussing higher sampling rates where it's easier to pull off.

This quote is from Craven's "Controlled pre-response antialias filters for use at 96kHz and 192kHz" AES paper.

Quote:
Traditional filter design has been either minimum phase or linear phase. A minimum phase filter has a monotonically increasing impulse response prior to the main peak, all downswings occurring afterwards. A linear phase filter has a symmetrical impulse response, so that in the maximally flat case there must be at least one downswing prior to the main peak. Thus, there is tension between the desire for a linear phase response in the audio frequency region, and a desire to avoid pre-responses in the time domain.

Does an antialias filter’s frequency response need to be strictly limited? We have assumed that the response at the Nyquist frequency should be down by 80dB or more, so that alias artefacts will be down by at least this amount. We do not comment on whether this is psychoacoustically necessary, indeed in [1] we see that some consider that it may be better to allow some aliassing in order to have a filter with a wider transition bandwidth and a ‘tighter’ time response.

How flat does the frequency response need to be in the range 0–20kHz ? Some contributors to [1] are prepared to contemplate a droop of 1–2dB at 16kHz. Here we will assume that a slow variation of ±0.2dB in the amplitude response over 0–20kHzwill be acceptable.
  Reply With Quote
Old 27th February 2015, 08:41 AM   #225
zfe is offline zfe
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by spzzzzkt View Post
Regardless of where it is done you have to deal with the imaged generated by resampling at some point.
Any kind of upsampling or oversampling of 44.1kHz still has a 2.05kHz transition band between the audio band and fs/2.

The DAM basically does in the FPGA what HQ Player does in software.
Essentially I agree, but I think there are two things that need to be expressed more carefully:

The images of the <20kHz spectrum are not "generated by resampling". They are a property of a signal sampled with 44.1kHz. Even with no resampling (so with a 44.1kHz DAC) you will have the images, but here they are at the analog output and can only be handled by analog filtering. The advantage of resampling with a higher frequency , is that the first images now also appear on the digital side and can be treated with digital filtering.

The first image comes mirrored in the frequency domain so at 44.1-20 kHz
(see e.g. http://www.dspguide.com/CH3.PDF), so you have a 4.1kHz transition band. This of cause only applies if, at time of recording, the signal was properly filtered so that there are no signal components above 20kHz.
If the recording engineer was a friend of slow filters you have a signal in the "transition band" which might call for a faster filter.

Last edited by zfe; 27th February 2015 at 08:49 AM.
  Reply With Quote
Old 27th February 2015, 09:25 AM   #226
diyAudio Member
 
Join Date: May 2005
Location: Melbourne
Blog Entries: 1
Quote:
Originally Posted by zfe View Post
The first image comes mirrored in the frequency domain so at 44.1-20 kHz
(see e.g. http://www.dspguide.com/CH3.PDF), so you have a 4.1kHz transition band. This of cause only applies if, at time of recording, the signal was properly filtered so that there are no signal components above 20kHz.
If the recording engineer was a friend of slow filters you have a signal in the "transition band" which might call for a faster filter.
I've checked a selection of my CD rips, and they ALL have content above 20kHz, most to at least 21kHz and some right through to 22kHz, so I suspect it's actually fairly widespread.

see attached screen shots from a track called Stopover Bombay by Alice Coltrane. Sorry about the different zoom levels.

It is pretty low level but it is there.
Attached Images
File Type: png Screen Shot 2015-02-27 at 21.15.35.png (74.1 KB, 519 views)
File Type: png Screen Shot 2015-02-27 at 21.23.40.png (50.4 KB, 436 views)
File Type: png Screen Shot 2015-02-27 at 21.40.34.png (76.5 KB, 408 views)

Last edited by spzzzzkt; 27th February 2015 at 09:41 AM.
  Reply With Quote
Old 27th February 2015, 09:38 AM   #227
diyAudio Member
 
Join Date: Feb 2009
Quote:
Originally Posted by zfe View Post
If the recording engineer was a friend of slow filters you have a signal in the "transition band" which might call for a faster filter.
No recording engineer I know does filtering up there by default, only if audible problems occur, so I would guess most recordings have some signal in the transition band.
But the AD converter with its filter should take care of that also.
  Reply With Quote
Old 27th February 2015, 12:10 PM   #228
zfe is offline zfe
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by spzzzzkt View Post
I've checked a selection of my CD rips, and they ALL have content above 20kHz, most to at least 21kHz and some right through to 22kHz, so I suspect it's actually fairly widespread.

see attached screen shots from a track called Stopover Bombay by Alice Coltrane. Sorry about the different zoom levels.

It is pretty low level but it is there.
Quote:
Originally Posted by hop.sing View Post
No recording engineer I know does filtering up there by default, only if audible problems occur, so I would guess most recordings have some signal in the transition band.
But the AD converter with its filter should take care of that also.
I see no real problems. All I wanted to say is,
if there is a x-dB signal at 22kHz, there will be a x-dB signal at 22.1kHz.
If you do not care about the 22kHz signal you design a filter, as before, with transition band of, say, 20-...kHz, and you are fine. If you are a believer of the importance of the 22kHz component and want to transmit it unatenuted ... you have to design with a much smaler transition band.

Before recording (or converting) to a 44.1kHz signal you have to ensure that there are no (or only very weak) components above the Nyqvist frequency 22.05kHz (or at least above 44.1kHz - f, where f is your desired later passband frequency) as they will result in aliases, which are later (with playback filtering) uncorrectable errors in the recorded signal.

So filtering (at recording) is only not needed if the signal does not contains such high frequency components anyhow.
Recording with higher frequencies, eases life as you need only to take care of much higher frequencies with (analog) filtering (usually the microphone bandwidth will already be the filter) and can later (if needed) do digital filtering before sample rate conversion (e.g. for CD pressing).

Last edited by zfe; 27th February 2015 at 12:12 PM.
  Reply With Quote
Old 27th February 2015, 12:30 PM   #229
derekr is online now derekr  Barbados
diyAudio Member
 
Join Date: Mar 2009
Quote:
Originally Posted by spzzzzkt View Post
Ok...

So I now understand why the NOS filter sounds harsh

This means that the NOS filter files need to be altered so there is no gain applied to any filter that has a bypass setting. The trade off is that level will be low and the DAC will clip if you set the gain any higher than V+00.
Paul, did you ever go back and alter the NOS file posted?
  Reply With Quote
Old 27th February 2015, 03:11 PM   #230
diyAudio Member
 
Join Date: Apr 2012
I am a little confused about FIR2, is it also bypassed/upsampling by the NOS file or is it still active? What is it doing if it is not bypassed?
  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
Soekris 's DAC implementations Eldam Digital Line Level 1038 20th April 2017 05:30 PM
Sold: Soekris DAM1021 R2R DAC Eldam Swap Meet 6 6th February 2015 07:23 AM
24-bit R2R DAC using miltiple 16/18/20-bit R2R chips Marek Digital Source 21 1st April 2011 09:05 PM
project brewing in my head, challenging, but seem like a good way to try new concepts lemans23 Multi-Way 26 11th October 2005 12:29 PM


New To Site? Need Help?

All times are GMT. The time now is 04:42 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2
Wiki