A/D Converter Possible??

[tragikremix]

Hi, i'm new to this site and i have been checking things out alot. I'm interested in building a mic preamp, probably going to try the Green Pre out, as it doesnt look too crazy.

i have been thinking alot about this, i always see 8 channel DIY pre's, like the Green Pre, and the SCA Pre kits. Is it possible (worthwhile) to build an A/D Converter to say change it from having analog outputs, to having ADAT or S/PDIF?

that is one project i havent seen, and i wondered if anybody does that.

Thanks a bunch guys.

 

[gyraf]

We've discussed this before on several occations - try a search.

Conclusion is that it's possible, but probably not worthwile.

Jakob E.

 

[kafka]

I've looked into this also, and have come to the conclusion that it's not a cost-effective solution to the problem. There are a number of products on the market that will take several line-level inputs and put them onto an ADAT Lightpipe. These include the Alesis AI3, the Behringer ADA8000, the Mackie Onyx 800, and the RME ADI-8DS. The prices of these range from around $300 to around $1800.

In terms of raw parts, you certainly won't be able to assemble anything cheaper than the Alesis and Behringer products, and most likely will end up using the chipsets in one of them anyway. If you're not happy with the sound, you can always get out the exacto and the soldering iron and modify it.

Personally, if I were going to go through the effort of engineering something like this that wasn't already on the market, it wouldn't be as a DIY project.

 

[tragikremix]

i figured it would be something like that. i have the ADA8000, and i doubt anyone could build something for less than the cost of it.

i tried the search, it didnt come back with much.

anyways, thanks for the replies.

 

[Rochey]

I keep looking at the parts we have and wondering if I can make a board. However, I've now come to conclusion that as a first rev, I'd be better off taking one of our EVM's, racking it up and putting some control in it.

 

[Svart]

whoa, that would be perfect for an app that I am planning on designing shortly.. can you direct me to more info about it??

 

[Rochey]

*edit*

 

[ruairioflaherty]

Rochey,

I'd love to see you post on that. I think the digital stuff intimidates a lot of Diyers (certainly does me) - a little bit of hand holding from an experienced digital kid could get a worthwhile project rolling,

cheers,
Ruairi

 

[Svart]

I was going to do something with s/pdif outputs, can this IC directly output s/pdif? if so that would be so excellent it hurts..

 

[gyraf]

[quote author="Rochey"]If admin (or nobody else minds) I will post a small post about it once we have a datasheet published.

Can I get a green light first? I don't want to piss anyone off.

Cheers

R[/quote]

At least one of the Admin's wouldn't mind at all!!! - after all, isn't this exactly why we're here?

It would make completely sense if you posted that sort of info - after all, there would be few here that would suspect you of being an advertisment agent sent my a multinational electronics gigant.. :razz:

Jakob E.

 

[kvintus]

Green light! Green light! Post! Post! Post!!!!

/Anders

 

[Regis]

Go ahead, count me in as I have been looking for a spdif adc project.

:thumb:

 

[Rochey]

I'm still waiting to hear back from Ethan for a green light, and for a green light from the product group on publishing the info.

However, if what you're looking for is a nice ADC S/PDIF project, take a look at the PCM4202 EVM that TI has. There's a full schematic in the EVM users guide.

Cheers

R

 

[Klaus Mogensen]

Regis,

you can take a look at Uwe Beis' ADC.

http://www.beis.de/Elektronik/ADDA2496/AD2496.html

it is very transparent and fairly easy to do. He is selling pcb's and hard-to-get components as well.

Klaus

 

[lagoausente]

I was thinking about buying the Cirrus Logic CBD5381. I´m guessing about what quality would get if comparing with aclaimed comercial devices like Mytek 192k. I also though about sent the unit to Jim Williams to mod it.
Seems there are more options. What about the quality we would expect?

 

[Rochey]

in all my talks with designers at AES, about our new ADC, the big point that came up over and over again was the point that the quality of an ADC product relies on a few different things:

The ADC itself
The Power supply
The Clock (jitter etc)
The Layout (crosstalk, digital noise etc)
The Front end circuity (line level to ADC levels)

However, copying an EVM that's already getting high performance shouldn't be too tough

 

[Klaus Mogensen]

We are of course hoping to make converters that are comparable to the best commercial ones if not better.
If you download and read this White paper from digidesign

Clocking, jitter and...

http://akmedia.digidesign.com/support/docs/192ClockJitter_30957.pdf

you will see, that a limitation of commercial converters is that the clock goes through a frequency synthesizer to generate the 256x oversampling clock from the word clock, which adds jitter. In the commercial converters it is necessary to be able to select different clock frequencies (else they can't sell the product, I guess???), therefore they use a frequency synthesizer to generate the oversampling clock from a fixed frequency from e.g. a crystal.
As I see it, the best is of course not to use a pll at all and connect the output from the crystal directly to the converter chips (non-pll XO clock). The disadvantage of this is that you can't select different clock frequencies.

I use a XO clock from Tentlabs (11.28 MHz for 44.1kHz word clock)

http://www.tentlabs.com/Products/cdupgrade/xo2xo3/index.html

which should have jitter < 3 ps.

 

[lagoausente]

[quote author="Klaus Mogensen"]We are of course hoping to make converters that are comparable to the best commercial ones if not better.
If you download and read this White paper from digidesign

Clocking, jitter and...

http://akmedia.digidesign.com/support/docs/192ClockJitter_30957.pdf

you will see, that a limitation of commercial converters is that the clock goes through a frequency synthesizer to generate the 256x oversampling clock from the word clock, which adds jitter. In the commercial converters it is necessary to be able to select different clock frequencies (else they can't sell the product, I guess???), therefore they use a frequency synthesizer to generate the oversampling clock from a fixed frequency from e.g. a crystal.
As I see it, the best is of course not to use a pll at all and connect the output from the crystal directly to the converter chips (non-pll XO clock). The disadvantage of this is that you can't select different clock frequencies.

I use a XO clock from Tentlabs (11.28 MHz for 44.1kHz word clock)

http://www.tentlabs.com/Products/cdupgrade/xo2xo3/index.html

which should have jitter < 3 ps.[/quote]
That´s very interesting. Really if can get the same result or even better shoud be awesome.
What about CBD5381 eval board, comparing to the one you are telling about?
What´s oversampling for? Just to can select sample rates?? I would like to understand this.
I also would like to can find a high end preamp to build, there are so many threads and META links that I´m afraid I´ll get old before I´ll be making some music with the perfect gear. :?
Anyone has a link of the theory of ADC , oversampling and all this issues?

 

[Klaus Mogensen]

Here are some picture and graphs

The adc

http://groupdiy.twin-x.com/displayimage.php?pos=-1759

Below are intermodulation distortion measured using the Right mark audioanalyzer.

Here the input is the diy adc and the ouput is RME fireface 400 dac. I used the clock from the diy adc and the RME is then clocked through the SPDIF input.

http://groupdiy.twin-x.com/displayimage.php?pos=-1760

Here's the RME fireface 400 by its own (internal clock)

http://groupdiy.twin-x.com/displayimage.php?pos=-1761

So using the diy adc doesn't really result in extra peaks. It is of course unfortunate that I have to use the RME dac in the first graph, because that used an oversampling clock generated from the input clock. At some point I will built and dac as well, to avoid a pll-clock al together.

Here is m-audio firewire 400:


http://groupdiy.twin-x.com/displayimage.php?pos=-1758

I don't know why the 60 Hz peak is so broad. Is it the Right mark program, because it is similar on all measurements??

 

[Klaus Mogensen]

Sigma delta converters and oversampling

http://www.beis.de/Elektronik/DeltaSigma/DeltaSigma.html

almost all converters are Sigma-delta converters that uses oversampling, e.g. 256 times. This is done to shift the images of the spectrum to higher frequencies so they more easily can be filtered out.

So if you have a word clock of e.g. 44.1 kHz. Then you need to feed the Sigma-delta converter chips with 256x44.1 kHz = 11.29 MHz. That is why I used a 11.29 Mhz clock from Tentlabs.

This also why I'm quite sceptical about statements like
'using 96 Khz is better than 44.1 kHz because the filter is better, since there is more room to filter...'

because isn't 256x already more than enough?? It probably makes sence for non-oversampling converters, which are not the converters they are listening to.