Kuei Yang Wang said:
Thing ive always wanted to know is... how do you know its realistic??
Were you at the original live recording of the music and remeber what it sounds like?
Did you watch the sound engineer mix and master the CD so you know that the sound was not tampered with?
Sorry, but "realistic" isnt a description i can believe.
If you said "it sounds the best to my ears", then sure...
Re: Re: Re: Re: Re: Re: On topic
It isn't any kind of interpolation, not according to the meaning of the word interpolation.
You are quite right. I don't have much time for opinions. Everyone and his dog has an opinion, give me facts any day.
To a large extent whether or not I understand it is irrelevant as I am not the one offering a process to all and sundry. I happen to think that if one is offering something one ought to know what it does and have some idea of how it works. Admittedly, this is only audio so offering up every weird idea under the sun has insignificant consequences.
As it happens I have pretty good of what you are doing and have never dismissed it. I merely pointed out that it was not a form of oversampling.
That is just an opinion.
Originally posted by guido
no lin interpolation, sounds great to me! Must be an even better form of interpolation then
It isn't any kind of interpolation, not according to the meaning of the word interpolation.
You are quite right. I don't have much time for opinions. Everyone and his dog has an opinion, give me facts any day.
To a large extent whether or not I understand it is irrelevant as I am not the one offering a process to all and sundry. I happen to think that if one is offering something one ought to know what it does and have some idea of how it works. Admittedly, this is only audio so offering up every weird idea under the sun has insignificant consequences.
As it happens I have pretty good of what you are doing and have never dismissed it. I merely pointed out that it was not a form of oversampling.
That is just an opinion.
You had a different opinion some time ago...?
http://www.diyaudio.com/forums/showthread.php?postid=335737#post335737
But i agree with you i should have a look at the maths.
Seems werewolf made a start there. See if i'm still able to do this kind of calculations
http://www.diyaudio.com/forums/showthread.php?postid=335737#post335737
But i agree with you i should have a look at the maths.
Seems werewolf made a start there. See if i'm still able to do this kind of calculations
guido said:You had a different opinion some time ago...?
http://www.diyaudio.com/forums/showthread.php?postid=335737#post335737
But i agree with you i should have a look at the maths.
Seems werewolf made a start there. See if i'm still able to do this kind of calculations
Its all in the details. For linear interpolation its half the sample time for two dacs a quarter for 4 dacs etc. In the case of the CS8412 that would be 32 and 16 sclk cycles respectively. You referred to a delay of 64 cycles.
I also found that a single TDA1541 with passive (resistor) I/V conversion sounds boring and extremely polite. The soundstage is far behind the speakers you are just not close enough. While keeping the passive I/V conversion things get much better if you parallel two chips.
Going the active route (Pass D1 in my case) is a totally different game though. Everything is just about right without fiddling around with parallel DACs and different I/V resistor values. A clear winner (imho)...
rfbrw said:
1/2 = 32 for 2 DACs
1/4 = 16 for 4 DACs
1/8 = 8 for 8 DACs
1/16 = 4 for 16 DACs
1/32 = 2 for 32 DACs
1/64 = 1 for 64 DACs
So, feed the first DAC directly, the second with 1 cycle delay, the third with 2 cycles delay, ...
...the 64th with 63 cycles delay ?
And this is 64x oversampling ?
So we need only 64 x TDA1541 or 128 x PCM56 for that ?
Will the HF be 44,1kHz x 64 = 2,8 MHz and very easy to filter ?
Bernhard said:
1/2 = 32 for 2 DACs
1/4 = 16 for 4 DACs
1/8 = 8 for 8 DACs
1/16 = 4 for 16 DACs
1/32 = 2 for 32 DACs
1/64 = 1 for 64 DACs
So, feed the first DAC directly, the second with 1 cycle delay, the third with 2 cycles delay, ...
...the 64th with 63 cycles delay ?
And this is 64x oversampling ?
So we need only 64 x TDA1541 or 128 x PCM56 for that ?
Will the HF be 44,1kHz x 64 = 2,8 MHz and very easy to filter ?
All the designs I am aware of precede the shift registers by oversampling digital filters, so in the most extreme you are looking at 256x or more. One of the designs in MJ had the SAA7220P and 64 PCM56 dacs.
Musen To Jiken is a japanese audio magazine that is only available in japanese but nevertheless it has quite some subscribers in Europe. It is often abbreviated as MJ magazine and yes, most europeans only can read the schematics. I have seen quite bizarre designs in this magazine and a lot of them are unconventional.
For instance the non os designs around TDA1543 by Kusonoki were published in this magazine.
Maybe a good excuse to leave that analyzer off for a while ?
Subscription page ( in japanese ) :
http://www.seibundo-net.co.jp/CGI/search/list.cgi?key=z_sinkan&c_bunrui=Z04
You can find more info on this site as well if you use the Search function on : Musen.
For instance the non os designs around TDA1543 by Kusonoki were published in this magazine.
Maybe a good excuse to leave that analyzer off for a while ?
Subscription page ( in japanese ) :
http://www.seibundo-net.co.jp/CGI/search/list.cgi?key=z_sinkan&c_bunrui=Z04
You can find more info on this site as well if you use the Search function on : Musen.
You do not need a schematic. It is just a lot of 74HCT164's, a few buffers, an inverter or two, a digital filter and LOTS of dacs. I've seen a few of these designs in MJ and they have started to merge into one but the basic principle remains the same. Here is one,from a previous thread on the topic, for 4 dacs demonstrating the principle.
http://www.diyaudio.com/forums/attachment.php?s=&postid=450998&stamp=1091510412
Bernhard said:
1/2 = 32 for 2 DACs
1/4 = 16 for 4 DACs
1/8 = 8 for 8 DACs
1/16 = 4 for 16 DACs
1/32 = 2 for 32 DACs
1/64 = 1 for 64 DACs
So, feed the first DAC directly, the second with 1 cycle delay, the third with 2 cycles delay, ...
...the 64th with 63 cycles delay ?
And this is 64x oversampling ?
So we need only 64 x TDA1541 or 128 x PCM56 for that ?
Will the HF be 44,1kHz x 64 = 2,8 MHz and very easy to filter ?
I am not sure why I am reading this thread in the first place but...
I just want to point out that oversampling this way and summing all the DAC outputs, you get the equivalent of a FIR filter with a triangular impulse response.
-3dB at approx 14-15kHz and -8dB at 22050. Not much difference between going from 4 DACs to 64. And no, the HF will not be at 2.8MHz, this filter is not that good.
I hope you are all aware of the fact that this could be implemented equally well in the digital domain in a DSP?
Also if you are going to sum up the output of many delayed DACs, you could get much more useful filter shapes if weighting the DACs by using different value I/V resistor from each. A gaussian filter for example.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.