Not far enough to do listening tests. The goal is to do the absolute best sounding DAC on earth, but then, people have different preferences, or have access to tech from other planets....
About the FFT plots:
I did some googling and at some point stumbled over a company called Metrum Acoustic making a series of Audio DACs, apparently using an obscure industrial DAC chip, the DAC8580.
They have great reviews, one of the with FFT plots:
http://www.metrum-acoustics.nl/octavehificritic.pdf
Looks like my -1 db plot, except my DAC have about 20 db less harmonics....
Also, a review of the MSB Platinum with FFT plots, again like mine, just a little lower, probably like my 0.01% version will look like:
MSB Technology Platinum Data CD IV transport & Diamond DAC IV & D/A converter Measurements | Stereophile.com
Seems like those kind of harmonics by them self is not a problem
I did some googling and at some point stumbled over a company called Metrum Acoustic making a series of Audio DACs, apparently using an obscure industrial DAC chip, the DAC8580.
They have great reviews, one of the with FFT plots:
http://www.metrum-acoustics.nl/octavehificritic.pdf
Looks like my -1 db plot, except my DAC have about 20 db less harmonics....
Also, a review of the MSB Platinum with FFT plots, again like mine, just a little lower, probably like my 0.01% version will look like:
MSB Technology Platinum Data CD IV transport & Diamond DAC IV & D/A converter Measurements | Stereophile.com
Seems like those kind of harmonics by them self is not a problem
Hi,
The whole digital Frontend differs, the supplies differ, the PCB Layout differs, the output stage differs, etc. etc. ...... heck, even the taste of the Listener differs.
If You change a thousand parameters at the same how could You possibly tell that parameter 269 is the one that decides?
Its my impression that Sœren has put in much sensible thinking and has provided his DAC with good genes.
That alone will make it the best DAC for some people
@Sœren
You said You kill capapacitive glitches theat occur inside the R2R network with a single 470p cap at the output, as the single switches are not compensated individually.
That hides the inherent problem of very fast switching of (stray) capacitive loads, but doesn't cure it.
Could You measure the bandidth limit of the DAC yet?
Is 384kHz clock possible at full bandwidth?
Hows about higher clock rates?
jauu
Calvin
Sorry, but how do You think could it be possible to gain ain insight about the sonic quality of R2R Core vs. SigmaDelta Core, if almost everything that could possibly be different is different?
The whole digital Frontend differs, the supplies differ, the PCB Layout differs, the output stage differs, etc. etc. ...... heck, even the taste of the Listener differs.
If You change a thousand parameters at the same how could You possibly tell that parameter 269 is the one that decides?
Its my impression that Sœren has put in much sensible thinking and has provided his DAC with good genes.
That alone will make it the best DAC for some people
@Sœren
You said You kill capapacitive glitches theat occur inside the R2R network with a single 470p cap at the output, as the single switches are not compensated individually.
That hides the inherent problem of very fast switching of (stray) capacitive loads, but doesn't cure it.
Could You measure the bandidth limit of the DAC yet?
Is 384kHz clock possible at full bandwidth?
Hows about higher clock rates?
jauu
Calvin
For those who might wish to understand some basics and the challenges of a discrete R2R DAC, here are a couple of useful links :
Discrete DAC
Discrete R2R DAC
http://www.diyaudio.com/forums/digi...c-serial-data-demultiplexing.html#post2918473
The fundamentals are still more or less the same.
It is still using 74xx595 to switch, and uses the power supply of the 595 as output voltage reference.
So all the challenges remain.
http://liu.diva-portal.org/smash/get/diva2:559515/FULLTEXT01.pdf
NerdKits - R-2R ladder using random crap resistors (Basic Electronics)
In the end the measurement capability limits the performance, even if you are prepared to go to all extremes.
Patrick
Discrete DAC
Discrete R2R DAC
http://www.diyaudio.com/forums/digi...c-serial-data-demultiplexing.html#post2918473
The fundamentals are still more or less the same.
It is still using 74xx595 to switch, and uses the power supply of the 595 as output voltage reference.
So all the challenges remain.
http://liu.diva-portal.org/smash/get/diva2:559515/FULLTEXT01.pdf
NerdKits - R-2R ladder using random crap resistors (Basic Electronics)
In the end the measurement capability limits the performance, even if you are prepared to go to all extremes.
Patrick
How good your power circuit's are, will tell you the freq. spectrum in the 200Hz to 10Hz region...
Hp
Hi,
so Yes the listening will be a personal opinion.
Some will vote in favour of one device, some will favour a different device.
You could as well make a ABX on the DAC vers. some analog vinyl stuff or any other source. The outcome of the ABX would be similar.
In which way does Your ABX tell anything about which technology of the DAC core maybe superior -in what respect/parameter anyway?
All this is known even before a single note is played through the DAC and listened to.
In #56 I reported that our own tests -which very much more looked at the differences of the DAC cores than a ill defined ABX test- resulted in a small majority of listeners preferring the Sigma Delta Core, but admittedly not truely distinguishable but on the most capable stereo setups.
Sören has done a marvellous job so far and I'm sure that in the end the DAC will be good enough that any decision for better, more musical or whatever is solely a matter of personal taste or preference.
Sören, I assume You measured the waveforms without the 470pF Lowpass and can tell that no glitches or even faulty switching bits occur (i.e. that glitches may occasionally and by accident trigger the wrong Bits)?
jauu
Calvin
so Yes the listening will be a personal opinion.
Some will vote in favour of one device, some will favour a different device.
You could as well make a ABX on the DAC vers. some analog vinyl stuff or any other source. The outcome of the ABX would be similar.
In which way does Your ABX tell anything about which technology of the DAC core maybe superior -in what respect/parameter anyway?
All this is known even before a single note is played through the DAC and listened to.
In #56 I reported that our own tests -which very much more looked at the differences of the DAC cores than a ill defined ABX test- resulted in a small majority of listeners preferring the Sigma Delta Core, but admittedly not truely distinguishable but on the most capable stereo setups.
Sören has done a marvellous job so far and I'm sure that in the end the DAC will be good enough that any decision for better, more musical or whatever is solely a matter of personal taste or preference.
Sören, I assume You measured the waveforms without the 470pF Lowpass and can tell that no glitches or even faulty switching bits occur (i.e. that glitches may occasionally and by accident trigger the wrong Bits)?
jauu
Calvin
Calvin,
ABX testing can cause a sensory null effect, where our perception tells us two devices are the same during the ABX, when they are different when listened to alone, this is due to pre-attentive processing.
See this paper for some testing in that direction ----- http://www.partsconnexion.com/t/claritycap/Reseach_Summary.pdf
See you
ABX testing can cause a sensory null effect, where our perception tells us two devices are the same during the ABX, when they are different when listened to alone, this is due to pre-attentive processing.
See this paper for some testing in that direction ----- http://www.partsconnexion.com/t/claritycap/Reseach_Summary.pdf
See you
Sign magnitude only addresses the major carry. Find one of the carry glitches that are still there and do your -60dB tone with that point as the center. Or surprise your self with the noise floor of a mutitone FFT.
Good stuff the industry went through a lot of this 25yr. ago. I think even all integrated on one small chip the jitter from the logic was still measurable in those days. I won't talk about the abortive attempt to put a deglitcher on one.
Søren,
Could you take some time to detail inputs available on your future pcb ? (outputs are clear I think now - XLR buffered, RCA buffered taken from XLR, RCA unbuffered)
In the first post you said:
There seems to be lots of possibilities with inputs ( many J** connectors on the pcb pics),
I got one friend who is particulary concerned to know if he will be able to plug both a PC and a CDP without adding 200$+ of modules and power supplies if there only one I2S input (SPDIF input board + Amaneo + digital i2s switch + PSUs).
Please tame some time to detail J** connectors, inputs, and input switching possibilities and, and anything else that will matter to diyers
Many thanks.
Could you take some time to detail inputs available on your future pcb ? (outputs are clear I think now - XLR buffered, RCA buffered taken from XLR, RCA unbuffered)
In the first post you said:
There seems to be lots of possibilities with inputs ( many J** connectors on the pcb pics),
I got one friend who is particulary concerned to know if he will be able to plug both a PC and a CDP without adding 200$+ of modules and power supplies if there only one I2S input (SPDIF input board + Amaneo + digital i2s switch + PSUs).
Please tame some time to detail J** connectors, inputs, and input switching possibilities and, and anything else that will matter to diyers
Many thanks.
Søren,
Could you take some time to detail inputs available on your future pcb ? (outputs are clear I think now - XLR buffered, RCA buffered taken from XLR, RCA unbuffered)
In the first post you said:
There seems to be lots of possibilities with inputs ( many J** connectors on the pcb pics),
I got one friend who is particulary concerned to know if he will be able to plug both a PC and a CDP without adding 200$+ of modules and power supplies if there only one I2S input (SPDIF input board + Amaneo + digital i2s switch + PSUs).
Please tame some time to detail J** connectors, inputs, and input switching possibilities and, and anything else that will matter to diyers
Many thanks.
I production board will actually have a little more inputs than the prototype as I have decided to connect more pins from the FPGA to a connector, will also connect some pins so you can have some LED indicators.
So there will be one isolated I2S port, to be used with your preferred USB interface. Then three SPDIF/TOSLINK/AES/EBU ports, using a balanced input directly to the FPGA, to be terminated off board with transformers and resistor to match port type, don't really require any boards for that....
The Power Supply is just a simple toroid transformer, and then a little wiring inside whatever case you use.
There are three speed limits:
Input Sample Rate: Planning to stop at 384 Khz, but you could go higher if you want to, although I don't see any reason to do that.
DAC Sample Rate: I have measured 70 nS rise and fall times with a square wave, so I expect that 6 Msps will be ok, although I will probably stop at 3 Msps.
Audio Bandwidth: Only limited by digital filters and the 200 Khz -3db 2nd order bessel filter on output. Which you can remove if you feel like to.
To verify speed in details, I will probably buy an Picoscope 4262 for testing, only reasonable priced way I have found to do 16 bit / 5 Mhz bandwidth FFT's.... Unless somebody have a better suggestion.
Hi,
I also think that 384kHz is more than enough, 192kHz is imho absolutely sufficient.
I liked if the input clocks were upsampled linearly.
44.1k/88.2k and 176.4 kHz to 176.4kHz and 48K/96k/192K to 192kHz (or to 352.8k and 384k if one wished for).
The DAC may clock at 1.5MHz or 3MHz.
Higher clock rates generate imho more problems than they solve.
At the supposed clock rates analog post-filtering may be still ´soft´ enough to avoid sonic issues of steep filters.
jauu
Calvin
I also think that 384kHz is more than enough, 192kHz is imho absolutely sufficient.
I liked if the input clocks were upsampled linearly.
44.1k/88.2k and 176.4 kHz to 176.4kHz and 48K/96k/192K to 192kHz (or to 352.8k and 384k if one wished for).
The DAC may clock at 1.5MHz or 3MHz.
Higher clock rates generate imho more problems than they solve.
At the supposed clock rates analog post-filtering may be still ´soft´ enough to avoid sonic issues of steep filters.
jauu
Calvin