First page Memory aid for the newest readers updated! (usefull links towards the others threads, blogs, wiki to save time)
http://www.diyaudio.com/forums/digi...oekris-s-dac-implementations.html#post4184471
Thank you for sharing
http://www.diyaudio.com/forums/digi...oekris-s-dac-implementations.html#post4184471
Thank you for sharing
Absolutely.
That was one of the reasons I decided to make a start with looking at alternative filters. They were advertised as a user configurable option, and it wasn't reasonable to expect Søren to do the work - especially when no single filter configuration seems to satisfy everyone.
And that should have been J11 not J1.
Last edited:
I haven't forgotten anything, I have just been hampered by a danish winter flu that just don't want to go away, maybe because I don't give it a chance to go away....
Manual and new version of firmware / filter sets will come soon, but don't know when exactly. Sorry for the delays.
And the dam1021 threads keep multiplying, just so it's clear, please direct any questions to me here, that will increase the chance of a reply....
Manual and new version of firmware / filter sets will come soon, but don't know when exactly. Sorry for the delays.
And the dam1021 threads keep multiplying, just so it's clear, please direct any questions to me here, that will increase the chance of a reply....
Hi Soekris
What do you think of my idea for the pure DSD decoding + volume control method for dam1021 dac?
Would like to know your thought.
Thank you
Let me explain in more details.
If we can get the 2.844Mhz 1bit DSD signal to the DAM DAC.
Since the FPGA can already out put 2.844Mhz.
Then we can just turn the R2R DAC PCM all bits high for DSD signal 1,
Then turn the R2R DAC PCM all bits low for DSD signal 0.
In essense, I just treat DAM R2R 28bit DAC as one single bit DAC.
The DAC still functions in PCM mode, however, with 2.844Hmz 28bit PCM signal
Just the PCM signal only has two value, equavlalent to a single bit DAC. And this is essencial pure DSD.
I would expect this to be very low calculation load for the FPGA. And the whole process does not manupulate DSD signal at all. No Digital Filter whatsoever.
In addition, as R2R PCM ladder out put depends on the PCM signal value, we can set the PCM value to tune volume.
one example,
Let's say the DSD signal is 0,1,0,1,1,0
Then the corresponding 28bit PCM vlaue is 0,2^28, 0,2^28,2^28,0
If we want to tune the volume down a little bit, then we can set the PCM signal to be: 0, 2^20,0,2^20,2^20,0
If this is possible, then this DAM dac can be a pure R2R DAC and a Pure DSD DAC
What do you think?
Thank you
Sent from my iPhone using Tapatalk
What do you think of my idea for the pure DSD decoding + volume control method for dam1021 dac?
Would like to know your thought.
Thank you
Let me explain in more details.
If we can get the 2.844Mhz 1bit DSD signal to the DAM DAC.
Since the FPGA can already out put 2.844Mhz.
Then we can just turn the R2R DAC PCM all bits high for DSD signal 1,
Then turn the R2R DAC PCM all bits low for DSD signal 0.
In essense, I just treat DAM R2R 28bit DAC as one single bit DAC.
The DAC still functions in PCM mode, however, with 2.844Hmz 28bit PCM signal
Just the PCM signal only has two value, equavlalent to a single bit DAC. And this is essencial pure DSD.
I would expect this to be very low calculation load for the FPGA. And the whole process does not manupulate DSD signal at all. No Digital Filter whatsoever.
In addition, as R2R PCM ladder out put depends on the PCM signal value, we can set the PCM value to tune volume.
one example,
Let's say the DSD signal is 0,1,0,1,1,0
Then the corresponding 28bit PCM vlaue is 0,2^28, 0,2^28,2^28,0
If we want to tune the volume down a little bit, then we can set the PCM signal to be: 0, 2^20,0,2^20,2^20,0
If this is possible, then this DAM dac can be a pure R2R DAC and a Pure DSD DAC
What do you think?
Thank you
Sent from my iPhone using Tapatalk
I haven't forgotten anything, I have just been hampered by a danish winter flu that just don't want to go away, maybe because I don't give it a chance to go away....
Hi Søren, cure the flu and make a good restart - it's better to keep a level head
Finally finished reading this whole thread, thats pretty long omg.....I saw that sebastian reported problems in lack of bass , which is also mentioned by one of my friend who owns a 0.1 version. Is it caused by power supply or filter? Does someone know?
Now Im waiting for my soldering station, once it arrives i ll also buy a 0.1 version lol
Now Im waiting for my soldering station, once it arrives i ll also buy a 0.1 version lol
Randy, be sure to link to Robert's output stage thread on your wiki page.
Thanks to you both.
Hi Dimdim,
my friend is using the buffered output. But later he added that its just a different style of bass, not really lacking.
zhang
Thanks, Ray--very good explanation. I will start to research the Botic software.
Alan
I don't have a problem with other's helping, in fact I encourage it.
Algar made some changes already, and I'm grateful for the assistance.
I started it, but it is intended to help everyone, so anybody that can add useful information is welcome to do it.
But, I will add the link soon.
Randy
Hi Soren,
I've run into a puzzling problem whilst using SoX to test out raw filter coefficients on audio files.
When I use zero insertion x 8 upsampling to 352.8k, apply the FIR filter, then adjust gain x8 some filters cause clipping and others don't. The default 44.1 filter shows just shy of 55,000 clipped samples in a 3 second .wav containing a 0dbFS sine test tone when checked using this method.
I know the DAM1021 is not SoX, so I was hoping if you could expand on whether this will be an issue with the internal filter handling, and what we need to look out for with custom filters.
cheers
Paul
I've run into a puzzling problem whilst using SoX to test out raw filter coefficients on audio files.
When I use zero insertion x 8 upsampling to 352.8k, apply the FIR filter, then adjust gain x8 some filters cause clipping and others don't. The default 44.1 filter shows just shy of 55,000 clipped samples in a 3 second .wav containing a 0dbFS sine test tone when checked using this method.
I know the DAM1021 is not SoX, so I was hoping if you could expand on whether this will be an issue with the internal filter handling, and what we need to look out for with custom filters.
cheers
Paul
This might be due to the filter design tools itself. E.g. with the t-filter tool you get with the standard settings, as passband, 0dB + passband ripple. If the ripple is positve somewhere you have an possible explanation of your clipping.
For testing if you hear the clipping:
With the DAM you could set the digital volume controll to -6dB, then you should have shifted the processing one position down in the resistor ladder. You still have 24 bit precission (without any errors), added some headroom to avoid the clipping but halved the output voltage.
For testing if you hear the clipping:
With the DAM you could set the digital volume controll to -6dB, then you should have shifted the processing one position down in the resistor ladder. You still have 24 bit precission (without any errors), added some headroom to avoid the clipping but halved the output voltage.
С135/142 are part of the final low pass filter, they are not "in" the signal path, The DAC is DC cuppeled.
Could it be worse than that? The volume adjust what you get out of the filter. So if the filter algorithm clip, the volume doesn't matter. So maybe one of the parameters in the filter file need top be adjusted.
But I cant see which of them would do this?
-- Format:
-- signature, samplerate, interpolationrate, type, numbercoefficients, multiplier
This somehow correlate with my impressions - as the intensity in music goes up, a sligth sense a hardness/congestion signature appear - its not as free. This is also on low listening levels.
//
Last edited:
To judge if the filter induces clipping independently of the volume setting one would need to know the used data structure (e.g. if numbers greater than 1 are posible at critical locations) and the algorithem in the FPGA.
Logically the cliping is a flaw of the filterdesign (in my oppinion). The passband setteings schould be -x dB where x is the choosen limit for the passband ripple.
I would not alter anything in the DAC filter.txt
An other source of getting a gain >1 could be due to rounding errors in the arithmetic. I do not think this will have a major impact (with a little bit more savety margin one should be on the save side, but again, without implementation details nothing definitive could be said).