Thank you, guys!
So there is no way to realize NOS in the way it was meant.
This is a limitation, that will hinder us to resolve some questions.
I've posted this, because I've set up a real NOS DAC from G2Labs based on an
old TDA-1543 some time ago. It sounds very interesting- let's say- analogue. My intention was to bring the DAM to similar conditions for a comparation.
...in order to repeat myself: I've never achieved better sound results through upsampling in all of my experiments and tests with dac's till now. The only effect I've noticed was a little more spherical soundstage- but for the price of loosing accuracy and resolution (subjective).
Greetingz, Robert
So there is no way to realize NOS in the way it was meant.
This is a limitation, that will hinder us to resolve some questions.
I've posted this, because I've set up a real NOS DAC from G2Labs based on an
old TDA-1543 some time ago. It sounds very interesting- let's say- analogue. My intention was to bring the DAM to similar conditions for a comparation.
...in order to repeat myself: I've never achieved better sound results through upsampling in all of my experiments and tests with dac's till now. The only effect I've noticed was a little more spherical soundstage- but for the price of loosing accuracy and resolution (subjective).
Greetingz, Robert
ID like to hear an r2r done like that too
The fpga system might be too restictive
But if we can bypass one oversampling we should try it
Have you done oversampling/upsampling experiments with a true R2R DAC? The tda1543 is indeed a multibit DAC, but not an R2R DAC. Perhaps a pedantic comment, but there is a distinction.
FWIW: I'm by no means a DAC expert, as I've only dabbled with a few of the lower-cost off-the-shelf models; and I've really only messed with a grand total of two DACs on the DIY side of things.
That said, until I got the Soekris dam1021 going, my "go to" DAC was an NOS-one based on eight of the tda1387 DAC chips (essentially a cousin of the tda1543). Here's the thread for that DAC. If you go to post #146 of that thread you'll see a little write-up I did on my perception of its sound quality.Now, my dam1021 build isn't completely finished yet, so it's still not easy to quick switch between it and the tda1387x8 for direct comparison... but so far my gut impression is that the dam1021 is at least as good as the tda1387x8. I like what I'm hearing.
If you can source a couple PCM1704 chips, you can build such a thing with this kit: DIYINHK PCM1704 R2R NOS DAC.
But back to the dam1021: I wonder if it's at all possible to program the FPGA in some kind of "pass through" mode?
What please should be the difference at the output terminals between the DAM running, hypothetically, natively at 44.1kHz and physically restricted to 16 bit, and the real DAM (running at 64 x 44.1kHz output) with only the NOS filters (in FIR1 and FIR2), so holding the unaltered input sample for 64 cycles of the output frequency?
Posted by matt_garman:
"Have you done oversampling/upsampling experiments with a true R2R DAC? The tda1543 is indeed a multibit DAC, but not an R2R DAC. Perhaps a pedantic comment, but there is a distinction."
No, I never have made experiments like this with a r2r dac. But just because of the differences between the two technical philosophies I am really turned on to try the dam in a real NOS- configuration;-)
BTW: The two dac's- the G2Labs and the dam sound very different- and all in all the dam is the better dac- but I really want to know how it would interprete music as a real NOS- dac;-)
Greetingz
"Have you done oversampling/upsampling experiments with a true R2R DAC? The tda1543 is indeed a multibit DAC, but not an R2R DAC. Perhaps a pedantic comment, but there is a distinction."
No, I never have made experiments like this with a r2r dac. But just because of the differences between the two technical philosophies I am really turned on to try the dam in a real NOS- configuration;-)
BTW: The two dac's- the G2Labs and the dam sound very different- and all in all the dam is the better dac- but I really want to know how it would interprete music as a real NOS- dac;-)
Greetingz
Did you try what Soren recommended in post http://www.diyaudio.com/forums/vend...magnitude-24-bit-384-khz-159.html#post4219786
I think his main worry is, in case of non oversampling, how to compensate for the 3db loss at 20 khz. Which is a 'drawback' of all R2R dacs. This, however, can easily be compensated as part of the analogue output filtering.
@metalrob
Reading this and other posts and Soren's recommendations I understood that it is possible to run the dac at 16bit 44.1kHz without oversampling (see http://www.diyaudio.com/forums/vend...magnitude-24-bit-384-khz-159.html#post4219786).
Soren also implies that by changing the settings you could set different oversampling rates (quote: the 8 is for 8 times oversampling, use 4 and 2 for other oversampling rates,
just like the 352/384 bypass filters using 1). As Soren said somewhere else, I believe these settings have to be specified for each individual sample rate.
Nevertheless, what I am still confused about is the Soren's statement that 'the dam1021 will always have the 2.822M/3.072M final sample rate, but t.ex. repeating the same 44.1K samples 64 times is like no oversampling....'
(see http://www.diyaudio.com/forums/vend...magnitude-24-bit-384-khz-169.html#post4221962 )
And I guess going from 352/384k to 2.822M/3.072M is defined as upsampling.
Reading this and other posts and Soren's recommendations I understood that it is possible to run the dac at 16bit 44.1kHz without oversampling (see http://www.diyaudio.com/forums/vend...magnitude-24-bit-384-khz-159.html#post4219786).
Soren also implies that by changing the settings you could set different oversampling rates (quote: the 8 is for 8 times oversampling, use 4 and 2 for other oversampling rates,
just like the 352/384 bypass filters using 1). As Soren said somewhere else, I believe these settings have to be specified for each individual sample rate.
Nevertheless, what I am still confused about is the Soren's statement that 'the dam1021 will always have the 2.822M/3.072M final sample rate, but t.ex. repeating the same 44.1K samples 64 times is like no oversampling....'
(see http://www.diyaudio.com/forums/vend...magnitude-24-bit-384-khz-169.html#post4221962 )
And I guess going from 352/384k to 2.822M/3.072M is defined as upsampling.
Last edited:
Hi TNT!
That all started with the audible empressions I've got from the BerryNOS1543 dac. Not that it sounds perfect concerning the last bit of resolution or 3D- imaging. Even not in dynamics or bass response.
But the subjective feeling enjoying music through my implementation of this device reminds me of authentic music reproduction, seems to be more 'organic'.
So I am excited to discover, how far I can get with our 'discrete' DAM that way.
In the last consequence I always trust my ears more than measurements. Having that said, I hope not to escalate a well known 'religious' struggle between tecnocrats and 'esoterics' (believing in technically rather unexplored effects- through experience);-)))
I think that both ways have their existential right and combining them without prejudice will bring synergy.
Greetingz
That all started with the audible empressions I've got from the BerryNOS1543 dac. Not that it sounds perfect concerning the last bit of resolution or 3D- imaging. Even not in dynamics or bass response.
But the subjective feeling enjoying music through my implementation of this device reminds me of authentic music reproduction, seems to be more 'organic'.
So I am excited to discover, how far I can get with our 'discrete' DAM that way.
In the last consequence I always trust my ears more than measurements. Having that said, I hope not to escalate a well known 'religious' struggle between tecnocrats and 'esoterics' (believing in technically rather unexplored effects- through experience);-)))
I think that both ways have their existential right and combining them without prejudice will bring synergy.
Greetingz
Because most likely it hurts the SQ in this project and there are better, from SQ point of view, solitions out there. This is an old dilemma, however, of purist audiofile vs. convenience/features approach. I guess we are trying to find out how the things are set up, limitations and possible solutions. And, needles to say, all is a matter of personal choices and tastes.
You see, that's the issue I have as well - my main front end is still analogue. But it helps as a reference point for upgrading the digital.
Good news is that there are examples of how non oversampling/upsampling R2R solutions can be implemented.
Last edited:
The dam1021 upsamples and filters in to steps, first from input sample rate to 352/384 Khz, then to 2.8/3.1 Mhz. As it has programmable filters you can make it in an non oversampling DAC by using "bypass" filters, basically filters with just "1" in the filter coefficients, and just enough coefficients to match the oversampling rate, t.ex. 8 from 44K to 352K, then 8 from 352K to 2.8M, this will work EXACTLY like a non oversampling DAC, including the issues like 3 db loss at 20 Khz and high frequency aliasing noise, which could ruin other parts of your system.... You could compensate for the 3 db loss by adding a filter in the IIR filter sections. But I don't recommend any of that, so doing the filters is up to somebody else.
I'll recommend the "soft" filter setting, as the name implies it's a pretty soft minimum phase filter.
I'll recommend the "soft" filter setting, as the name implies it's a pretty soft minimum phase filter.
upsampling is a necessary part of the oversampling process, so it is part of the software structure of the dam and as you probably understood from Sorens answer, he is not a fan of NOS and so he is probably not writing a new software for something he does not support, especially since you can do NOS with the board, the upsampling does not matter in this case.