I should have music though it in a couple of weeks, my references are not that fancy, an older Sony ES series CD player and a Cambridge DacMagic Plus, plus a Sennheiser HD-650. But don't expect me to be objective
.
.
If you lend me a beta-board, I'll be objective
Hi,
thanks for posting the Svenson links.
He seems to have made the same experience and drawn similar conclusions as I made and asked for linear upsampling in my first posts at #56 and #76.
IIR filters behave completely analog.
Steep analog filters always had their sonic issues, so have IIRs.
Even if FIR loose the correlation of phase versus amplitude I see no reason why the sonic issues of steep filters shouldn´t apply to FIR also.
In that -simple to calculate- linear upsampling reduces the requirement on filter steepness (digital as well as analog) it has all it needs for good sound.
jauu
Calvin
thanks for posting the Svenson links.
He seems to have made the same experience and drawn similar conclusions as I made and asked for linear upsampling in my first posts at #56 and #76.
IIR filters behave completely analog.
Steep analog filters always had their sonic issues, so have IIRs.
Even if FIR loose the correlation of phase versus amplitude I see no reason why the sonic issues of steep filters shouldn´t apply to FIR also.
In that -simple to calculate- linear upsampling reduces the requirement on filter steepness (digital as well as analog) it has all it needs for good sound.
jauu
Calvin
x2, IIR sounds much more analog.
An inverse attack and decay gives the illusion of more detail, it's tiring to listen to FIR for over 10 years, the human mind is just so adjusted to it that we keep deleting the inverse, so we must drink more sugar because the FIR filtering requires this mental energy, to listen, maybe around 100kcal per day.
Hey I just want to find out if resistors sound different in blind listening, that's all! But maybe you are right, I should try this resistor experiment in an amplifier instead of a DAC.
I have seen with my eyes that normal resistors are very magnetic for example, no questions asked there.
Esgigt I wasn't referring to 0.01%, Riedon have Silver and Platinum in them, precious metals, anyway sorry for the off topic and esoteric instead of technical.
The Danish DIY-er Duelund was esoteric, Silver and oil was king to him, but now I will cease these comments here.
Last edited:
The problem is the number of resistors... if it were a couple, one could try esoteric types. In these kind of applications that would be almost ridiculous.
But if one just won the lottery....
The price of a Duelund product is more than most of us can (or are willing) to afford.. besides, you'll need a lot of space due to the sizes..
Last edited:
If you lend me a beta-board, I'll be objective
Well, I'm in - if not for anything else then for the beauty of the board. It's really appealing. Well done layout Søren.
Great work...
soekris! You have a really cool project going here and clearly you have spent a lot of time and energy.
I am also glad you are considering a version with a discrete output stage, I generally much prefer the sound of a good discrete design vs. most IC opamp implementations.
I also am not so sure about your clocking scheme, it seems you are asynchronously re-sampling the data into a new clock domain if I understand correctly, is that so?
My (crazy audiophile belief?) is that asynchronous re-sampling is generally best avoided, and integer oversampling is a better sounding approach.
Does your DAC need to run at very high speed for best performance? Or can it run at say 352.8/384 kHz and do well? If so, I would suggest allowing the option of a NOS mode, where folks could experiment with oversampling/filtering in a computer: for most DIYers programming a FPGA may be impossible, but trying different oversampling filters via computer using playback programs like Audirvana Plus is quite simple.
I would love to see a NOS approach (not because I like NOS sound, but so we can experiment with our own OSF) with micro BNC connectors for the I2S lines (u.fl cables) and masterclock input as well...
soekris! You have a really cool project going here and clearly you have spent a lot of time and energy.
I am also glad you are considering a version with a discrete output stage, I generally much prefer the sound of a good discrete design vs. most IC opamp implementations.
I also am not so sure about your clocking scheme, it seems you are asynchronously re-sampling the data into a new clock domain if I understand correctly, is that so?
My (crazy audiophile belief?) is that asynchronous re-sampling is generally best avoided, and integer oversampling is a better sounding approach.
Does your DAC need to run at very high speed for best performance? Or can it run at say 352.8/384 kHz and do well? If so, I would suggest allowing the option of a NOS mode, where folks could experiment with oversampling/filtering in a computer: for most DIYers programming a FPGA may be impossible, but trying different oversampling filters via computer using playback programs like Audirvana Plus is quite simple.
I would love to see a NOS approach (not because I like NOS sound, but so we can experiment with our own OSF) with micro BNC connectors for the I2S lines (u.fl cables) and masterclock input as well...
Should have options to choose. Minimum-phase type is preferred, agree completely with John Atkinson.
No async resampling here, re-clocking with FIFO ala:
http://www.diyaudio.com/forums/digi...ifo-project-ultimate-weapon-fight-jitter.html
Oversampling will of course be even multiple, and you can load custom filters, including non oversampling and/or bypass ones....
So we can load secret rabbit code, SoX, linear-phase and minimum-phase filters?
Most modern DAC chips operate with 8x oversampling, is this one higher?
So we can load secret rabbit code, SoX, linear-phase and minimum-phase filters?
Most modern DAC chips operate with 8x oversampling, is this one higher?
Any filters you can create, limited by resources in FPGA, probably two or three FIR sections plus probably one IIR section for crossovers. Details not done yet.
The R-2R DAC itself will probably be limited to 3.072 Mhz, running 8x at 384 Khz input, can be higher at lower input rates.
It's not a problem, you'll never HEAR the difference, and bottom line, that is the reason for building the DAC!
Just to say that I'm really impressed with this project, the prototype pcb shown in the first post is a work of art in itself.
One question, could the buffered output be reasonable used with 'conventional' single-ended headphones. If I were to go with this DAC I would go with a tube buffer from J8 and only use the onboard buffers for headphones.
I await the first listening results with interest.
Ray
One question, could the buffered output be reasonable used with 'conventional' single-ended headphones. If I were to go with this DAC I would go with a tube buffer from J8 and only use the onboard buffers for headphones.
I await the first listening results with interest.
Ray
Hi Soekris,
I think this is going to be a great project for DIYs, I always like multi- bit DAC myself and believe your R2R design will enable us to experience a step further in multi-bit DAC without paying too much.
Great work and please keep it up.
Please also put me down for two sets of best resistors you can offer.
I think this is going to be a great project for DIYs, I always like multi- bit DAC myself and believe your R2R design will enable us to experience a step further in multi-bit DAC without paying too much.
Great work and please keep it up.
Please also put me down for two sets of best resistors you can offer.