Pc -> Dac, How ?

Status
Not open for further replies.
Jump to Latest
#102
wakibaki wrote
Click to expand...

I like trolls. Sometimes I do troll myself. It's always fun to annoy other people, so I understand it turns you on...

Yet, we're on the internets. It is a place very different from Real Life. In Real Life, if someone annoys you, you can do stuff like talk loud, or smash them with a steel bar. All you can do here is write in bold. You'll understand therefore that I fail to be impressed by your might 😀

All you can do on the internets is eat your own frustration and end up with an ulcer. If that's your thing, then I'm glad to help.

Anyway. To everyone except the idiot, thanks for the posts, there is some good info here...

I'm not too interested in single board linux computers, because I'd like to try some DSP stuff like custom oversampling, and a FPGA is good at this. Will it yield a useful result ? I don't know... that's why I want to try.

Also, you have a chicken and egg problem : if a given single board computer is powerful enough to do interesting things, then it will be comparable to a PC wrt noise and complexity...

> IMHO the quality and accuracy of the digital filter has
> more impact than HOW you remove/de-correlate jitter
> spurs.

Which makes me even more interested in trying various impulse responses.

I'm thinking about the usual sinc response, with various lengths (lots of taps or a little), then maybe try a phase-shifted response so the pre-echos don't happen, then some slow rolloffs...

Pjotr, in your experience, is it the impulse response, the number of taps, the arithmetic precision, dither and truncation, that matter most ?... which filters did you like best ?

Some DACs like the new delta-sigmas support an external filter, so you can feed them 1.5 Msps 24-bit data direct to the modulator. I wonder how this will sound. If it sounds better, will it come from the better filtering, or from less noise in the chip ? So many questions...

> I guess any miniPCI-equipped router should be
> powerful enough for 8 channel 192/24 wav ethernet
> playback (netcat?)

NetJack... read about this... it's a really cool piece of software !

Project update : we're arguing about the ways to connect the modules together, it's important to have a system that leaves all the options open. We're getting closer.

> When we test for side bands we are testing the
> system, not the chip.

You're also testing the ADC clock... it's a real can of worms, I can't wait to open it. I'm impatient for the day this thing makes its first analog loopback test, with independent DAC and ADC clocks.
 
#107
Well sorry for not posting -- I'm busy having a life and on my free time, taking care of this project. It's advancing, a rough estimation of the first "playing music" stage would be about 1-1.5 months (cause making PCBs takes a bit of time). Basically we're building the basic module set : dac, adc, clock, power supplies, 1 or 2 IVs, that takes time... actually most of it is done, but it's always the last details that take most time ! Don't worry it's going to happen 😀
 
#109
peufeu said:
Well sorry for not posting -- I'm busy having a life and on my free time, taking care of this project. It's advancing, a rough estimation of the first "playing music" stage would be about 1-1.5 months (cause making PCBs takes a bit of time). Basically we're building the basic module set : dac, adc, clock, power supplies, 1 or 2 IVs, that takes time... actually most of it is done, but it's always the last details that take most time ! Don't worry it's going to happen 😀
Click to expand...

Hi peufeu
I apologize for writing this, I did not mean to be pushy or demanding, I just wanted to remind that there are people that are really amazed by your projects and the contribution from other people, I do hope this thread will evolve in something magnificent. You have become quite a famous man in the DIY community, you know it
🙂

I wish you great time with all your hobbies, we can wait patiently here hoping you get even more inspiration.


P.S. have you seen the movie "Heat" ? Al Pacino has a very memorable scene there, interrogating some guy - he just starting banging on the table shouting "GIMME AL YOU GOT!!! GIMME AL YOU GOT!!!" That's exactly how I feel 🙂 🙂 I am just a guy that enjoys reading these highly intellectual threads, hoping to get a good sounding dac someday, ignore my posts as spam if you like, I am not trying to be offensive or intruding in any way...
 
#115
Telstar said:


Yes, but it is far from ideal, its like a mini-pc with native i2s out. Way too basic and unpowered for normal tasks.
Click to expand...


Oh its plenty powerful for many tasks.

Especially since the Sabre32 DAC does all the DSP work I need. 🙂

All I have to do is feed it coefficients and I have I fully functional digital crossover with 8 independent filters. This means I can filter the same signal 8 ways, or a stereo pair of signals 4 ways with one chip. 🙂

Very very nice.

All the beagle has to do is convert FLAC and spit out I2S. The clock on the Buffalo(Sabre32 board) will dictate the performance. Jitter on the bit clock will not really come into play.

It a nice inexpensive but very definitely hi-fi solution.

Now if I had to do the DSP before the DAC chip, then I would probably choose something else.

Some may think the bit clock jitter will come into play, I can confidently report, it sure does not sound that way to me. 🙂 It sounds superb at every sample rate I have tried. ESS did a wonderful job with their 100% digital PLL.

Now if you want you can also slave the bit clock to a low jitter source, but I think you will find not much return on that work/cost.

Just feed the ESS DAC a nice clean low phase noise clock and you will get an awesome project.

Try it, you'll like it. :yes:

Cheers!
Russ
 
#118
Russ White said:



I am pretty sure (I have not tried it yet) can do room correction on the DAC. 🙂

Its just another filter.

Of course I will use a PC to generate the coeffs.
Click to expand...

What length of filter is supported, and at what ultimate internal sampling rate? I highly doubt they're long enough to do what is commonly considered 'room correction'. 64k taps at 44.1 is pretty standard in DRC setups. Assuming the Sabre filters run at 192 or 384, that length would have to go up accordingly.
Still, doing the convolution on the PC and then the xovers on the Sabre would still be a pretty good setup.
 
#119
Re: i2s

Telstar said:
BTW, has anybody been able to get i2s from a (real) computer, besides the esi juli@?
I mean with a simple and cheap pci or pci-e (better) card.
Click to expand...


Although no longer in production, the older ESI rack-mount products use a small PCI card containing (IIRC - whichever the Juli@ uses...) a VIA ENVY 24HT 1704 controller chip. It sends I2S via transciever chips to the rack box via 25 pin connectors. I have the pins tentatively mapped, though I haven't needed to test them because I have successfully hacked the rack box to extract I2S - and I use the box's microphone inputs to record calibration scans. The PCI board sends 4 stereo pairs through the 25 pin cable. The ESI drivers are adequate.

Following this thread with interest,

Frank in Mpls.
 

Attachments

  • esipcibd.jpg
    esipcibd.jpg
    12.4 KB · Views: 657
#120
Re: Re: i2s

francolargo said:


Although no longer in production, the older ESI rack-mount products use a small PCI card containing (IIRC - whichever the Juli@ uses...) a VIA ENVY 24HT 1704 controller chip. It sends I2S via transciever chips to the rack box via 25 pin connectors.
Click to expand...

Very interesting.
I had a look at ESI products and there are a few pci and pci-express devices which can do what I'm looking for (with a simple mod).

It is not very clear what bitrate and sample rate you can get from such cards from the i2s header.

PCI-E has to be preferred because it is more stable, has dedicated bandwith. None of those boards are shielded unfortunately, but i dont think is very hard to put some panels on it.


I have the pins tentatively mapped, though I haven't needed to test them because I have successfully hacked the rack box to extract I2S - and I use the box's microphone inputs to record calibration scans. The PCI board sends 4 stereo pairs through the 25 pin cable. The ESI drivers are adequate.

Following this thread with interest,

Frank in Mpls.
Click to expand...

The latest ESI drivers (for julia, and some other cards) are quite good and support Vista. The ones for the older cards are a no-go for me for this same reason.

About linux world, the Envy24ht is well supported, so that's also a go.

Last but not least, there is the word clock issue. IIRC i2s contains a clock signal. I wonder if an external clock would be needed, like it is for spdif, or that can be skipped without compromising SQ too much.

These two newer products from ESI are in my radar now:
http://www.esi-audio.com/products/esp1010e/

http://www.esi-audio.com/products/maxio032e/
the second has a wordclock input (on the external interface)
 
Status
Not open for further replies.
Top Bottom