Is anyone going to try to build a Powered By Tent boards 16 stacks. I have 16 of the old boards running and the performance is simply stunningly good. I did try a while back but had terrible distortion so had to give up on the project for various reasons (Divorce etc).
If anyone manages it let me know as it would be an amazing sounding Dac.
If anyone manages it let me know as it would be an amazing sounding Dac.
OK, I'm aware that this thread is very old but I must mention something about using PCM1794 in NOS mode.
The DAC is operated in DSP mode. The guys who designed it intended it to be run with a DSP that makes the oversampling, usually 4 or 8 times.
SO I actually found a quirk some time ago when I experimented with this DAC. In this mode, the SCK may never be higher than 64 x WCLK, that's why this project uses the BCK for SCK. If running with higher S/D rates it produces garbage.
64 times wclk doesn't give a proper resolution, unfortunately, and the 130dB dynamic range can never be reached. At least not instantly enough. The 130 db will be there but after an longer period of time ( since it's an averaging device).
My solution was to feed the DAC with four consecutive identical samples at four times fs. This improves resolution with a factor of four. And it's still a NOS DAC.
At fs of 44.1, I used a SCK of 256*fs.
Ok, I only want to mention this. Any thoughts about this?
The DAC is operated in DSP mode. The guys who designed it intended it to be run with a DSP that makes the oversampling, usually 4 or 8 times.
SO I actually found a quirk some time ago when I experimented with this DAC. In this mode, the SCK may never be higher than 64 x WCLK, that's why this project uses the BCK for SCK. If running with higher S/D rates it produces garbage.
64 times wclk doesn't give a proper resolution, unfortunately, and the 130dB dynamic range can never be reached. At least not instantly enough. The 130 db will be there but after an longer period of time ( since it's an averaging device).
My solution was to feed the DAC with four consecutive identical samples at four times fs. This improves resolution with a factor of four. And it's still a NOS DAC.
At fs of 44.1, I used a SCK of 256*fs.
Ok, I only want to mention this. Any thoughts about this?
Is there any further documentation of your work / experiments? I guess your results might be interesting for the DDDAC crowd!
This is not unknown of course and is described in the test section of the DDDAC web page.
In practice it is no issue. Unless you are connecting the DAC output without volume control to a high gain amplifier chain and use digital volume control at the media player (which is not recommended anyway) you can read this here:
Link to test page of the DDDAC web site
But still I would be interested in a simple solution in hardware to speed up the bit clock times 4 for red book. For the others it is not necessary. Maybe you can post what you did here?
PS: just to make no secret of my position, that no one needs 130dB dynamics. What we need is great analog sounding music
Could this be done in software? What if the playback software does some kind of very basic upsampling by just repeating each sample four times, but at a four times higher sampling rate?
I am using Pure Music 3.08 - I upsample to 176 or 352 most of the time. There are 2 choices for upsampling in Pure Music - Maximum Fidelity 64 bit and NOS type. I’m not sure what NOS type is. I haven’t tried it but I am wondering if it just inserts more identical samples like is mentioned above. In other words, no interpolation at all. I think I will send an email to Channel D to find out what NOS upsampling actually does.
I guess that is what they mean with NOS - Which means Non OverSampling, which is a nice paradox PUN
You can test this VERY easy.... lay your hands on some test signals and pick a squarewave signal and play it - looking at an oscilloscope (or via sound card). if it is still "square" it is "NOS" if you see ringing they use FIR up-sampling technics (Like almost every DAC and CD player)
OK, DDDDDAC, ( I'm stuttering ).
You are right, your DAC probably sounds good with SCK at 64*fs.
BUT.... since the device is a sigma delta DAC, it will always obtain full 130dB resolution after a couple of samples. But it will take longer time for the DAC to "home in" at the desired value. Thats a very well known disadvantage for any S/D device, even if a PCM1794 ( with it's 6 hardware levels ) is almost as good as a 16 bit (ideal)multibit DAC.
So, since this DDDAC is supposed to be a real hi end DAC we should squeeze every possible bit of resolution out of it.
My own solution is to feed the DAC from a DSP. I'm using a fairly cheap one from Microchip. This DSP also acts as a buffer and outputs its data, clocked from a master oscillator. It's simply reclocking the input stream. It also handles a radio controlled remote control to set the volume. And other tasks.
So, this DSP feeds the PCM1794 with identical copies of the samples, so the SCK can be run as high as fs*256. And it's still in NOS-mode, since the samples are identical.
Perhaps you are interested to make a DDDAC v2.00 from my project?
You are right, your DAC probably sounds good with SCK at 64*fs.
BUT.... since the device is a sigma delta DAC, it will always obtain full 130dB resolution after a couple of samples. But it will take longer time for the DAC to "home in" at the desired value. Thats a very well known disadvantage for any S/D device, even if a PCM1794 ( with it's 6 hardware levels ) is almost as good as a 16 bit (ideal)multibit DAC.
So, since this DDDAC is supposed to be a real hi end DAC we should squeeze every possible bit of resolution out of it.
My own solution is to feed the DAC from a DSP. I'm using a fairly cheap one from Microchip. This DSP also acts as a buffer and outputs its data, clocked from a master oscillator. It's simply reclocking the input stream. It also handles a radio controlled remote control to set the volume. And other tasks.
So, this DSP feeds the PCM1794 with identical copies of the samples, so the SCK can be run as high as fs*256. And it's still in NOS-mode, since the samples are identical.
Perhaps you are interested to make a DDDAC v2.00 from my project?
Hello there,
I've been rebuilding and upgrading my DDdac for a while now and very very happy with the results i must say.
From wave-io to streaming: Rpi 3b with Volumio, ALLO Isolator with ALLO Kali.
This setup has the advantage that you don't need an extra "audioquality" PSU.
The Rpi uses it's own small crappy stock powersupply which also feeds the "unclean"
part of the Isolator just before Galvanic separation, the "clean" part of the Isolator is fed with clean audioquality power from the Kali which has the 5V from the stock PSU for the ditched wave-io.
The DDdac is upgraded from 2 to 4 decks.
I upgraded the stock 2 PSU's with one very large (500VA) transformer.
The 2 secondairy windings are manually adjusted to give the proper voltage to each PSU.
I made one upgrade to the PSU circuit, the half wave rectifier was ditched and replaced with a proper full wave bridge rectifier.
And finally i intergraded a TVC in this setup, not a Sowter but the Silk.
The endresult is much more than i wished for but it took a while to get there.
It noticed that it takes a long time, up to 250 hours of actual burn-in time, to let the new dac decks sing, these are the TENT versions.
I don't use anything special for burn-in, just music day and night, 24/7 for 2 weeks straight.
My experience was that after the 200 hours passed, the soundimage went from 4:3 to 16:9 and then slowly crept to 21:9.
The depth image did the same.
Well, as i said, it sounds breathtaking, it is deadsilent, no backroundnoise whatsoever just music!
A very happy DDdac user
I've been rebuilding and upgrading my DDdac for a while now and very very happy with the results i must say.
From wave-io to streaming: Rpi 3b with Volumio, ALLO Isolator with ALLO Kali.
This setup has the advantage that you don't need an extra "audioquality" PSU.
The Rpi uses it's own small crappy stock powersupply which also feeds the "unclean"
part of the Isolator just before Galvanic separation, the "clean" part of the Isolator is fed with clean audioquality power from the Kali which has the 5V from the stock PSU for the ditched wave-io.
The DDdac is upgraded from 2 to 4 decks.
I upgraded the stock 2 PSU's with one very large (500VA) transformer.
The 2 secondairy windings are manually adjusted to give the proper voltage to each PSU.
I made one upgrade to the PSU circuit, the half wave rectifier was ditched and replaced with a proper full wave bridge rectifier.
And finally i intergraded a TVC in this setup, not a Sowter but the Silk.
The endresult is much more than i wished for but it took a while to get there.
It noticed that it takes a long time, up to 250 hours of actual burn-in time, to let the new dac decks sing, these are the TENT versions.
I don't use anything special for burn-in, just music day and night, 24/7 for 2 weeks straight.
My experience was that after the 200 hours passed, the soundimage went from 4:3 to 16:9 and then slowly crept to 21:9.
The depth image did the same.
Well, as i said, it sounds breathtaking, it is deadsilent, no backroundnoise whatsoever just music!
A very happy DDdac user
This is very interesting! I tried (almost) the same, and I am getting some serious noise/hiss with the Kali (see here and followups). Did you ever observe anything similar with your setup? Do you have experience with running the setup without the Isolator?
Lower resistors would indeed give lower voltage swing. However, the DAC chips would be biased at a higher current, which would increase distortion (maybe dramatically). It would be better to use a different method to reduce the output voltage. You could use two smaller series resistors in place of the single I-V resistors and tap the output signal between those resistors. Or you could use a different output transformer that has a voltage step down.
However, the signal voltage swing of the DDDAC already is a bit on the low side (that's why Doede used a step up transformer in his design). I would therefore look for excessive gain elsewhere in your system.
Hello mbrennwa, sorry to hear about your problems, i do not have experience playing without the Isolator.
I've had some problems but not the one you discribe, i made one mistake with the I2S connections.
I connected GPIO 29 aka MCLK who we don't need, as soon as i disconnected this one everything was souding as it should be.
Therefore, you only need to connect GPIO 40 (DATA), 35 (LRCLK) and 12 (BCLK) for grounding i connected 39 also.
At the moment these connections aren't even shielded so i'm busy right now with creating some "air" silver shielded FMJ interconnects for the I2S connections.
Let me know if this works out for you.
Cheers, Paulus
I didn't connect MCLK. And I have already tried shielded cables. See here (sorry for jumping between threads, I am trying to pick the brains of the great people in both both threads). The main conclusion so far is that my noise/hiss problem goes away if I connect the RPi to the DDDAC directly (i.e. without the Kali).