DIY PC audio interface

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Rochey

Well-known member
White Market Member
Joined
Jul 2, 2004
Messages
2,697
Location
Dallas, TX - Formerly UK
Okay...

Let me get it out the way -- I have no idea where to begin on something like this, but i thought i'd start a thread to collect ideas and thoughts, and posisble even start a design with you folks.

So I'll start things off in a simple way... :D

We have 4 options:

Firewire (1394)
USB2 (forget usb1 - it sucks for more than stereo!)
ADAT Optical - easy to use, but could be difficult to get higher sampling frequencies.
PCI -- who feels brave?

Where would we begin as a DIY community?

Cheers 'n Beers

R :grin:
 
[quote author="Rochey"]USB2 (forget usb1 - it sucks for more than stereo!)[/quote]
Can you get any easy to use USB2 high speed chips? Most of the chips (including the tusb3410) support either USB 1.1 or USB2 full speed, which isn't any better...

PCI -- who feels brave?
I have designed ISA cards - I'm not sure I'm brave enough to design PCI cards yet... But I guess there may be some more-or-less plug and play PCI I/O chips available?

Best regards,

Mikkel C. Simonsen
 
Heya MCS,

your right about USB2 -- there are only a few players doing the high speed devices. TI doesn't, and I don't know the details for the others...

as for the ISA cards - i'm impressed... doing an ISA card alone should get you a gold medal.

From what I can see, the ADAT option is the easiest, as the interface card itself can be bought elsewhere.
But I'm not sure if you can double teh sampling frequency for 96KHz operation on ADAT. Is it possible? I know that ADAT devices are available from wavefront semiconductor.

Any thoughts?
 
great idea...
I tryed adat for some month now, some stuff works, some other don't...

For pci card, I think the easiest way is using a CPLD or FPGA chip, but need someone who knows both this and pci stuff...
Firewire maybe?

Well for adat, easy to make a simple 8 in and 8 outs, with chips from alesis (al1401 and al1402 for adat interface, and al1101 and al1201 for AD-DA), then a simple 48/44.1 clock and that's all... I made an al1101 adc already working with the al1401 and al1402. I tried to make dac too, I first though about ti's one, but to small, to hard for a first try, I thin i'm going to order al1201 in the next day... this gives you the same as the Beringher AD800 (without the mic pre) as far as I read today...

if you want more than 48khz with adat, you can use s/mux, but then you'll get only 4 channels instead of 8, and can't use the alesis adat encoding and decoding chip, btw I think it's simple to make adat encoding with CPLD.

My first idea was to make something modular... but I really can't design a such things, without wire solder everywhere on pcb.

Mcs : I though about using a pll chip from ti to, but I need the LRclock (and maybe the bitclock for other adc). I can't do anything without it, as adat chip needs one lr clock for 4 pcm signal. I can't see another way for doing this instead of using a 4040 or such divider to get the right stuff from the master clock. Do you really think I'm going to get a bad clock from this?

thank you!
 
Hey Guys,
Welcome to the Lab! :thumb: :guinness:

Good to see the arrival of more digital players, please go slow so we(I mean I) can sort of follow since all this stuff is REALLY over my head....... :oops:
However, I would love to observe and learn! :thumb:

still struggling with analog.........and primary 8051 machine code/interfacing using C..... :?
 
I'm still wondering if one way to go might be to buy cheaper converters (M-Audio Delta and the like) and mod them to get higher quality. Of course, I'm not too keen on modding surface-mount circuits...
 
Some folks are talking about imporving a MOTU 1224 over here.
http://www.groupdiy.com/index.php?topic=880&highlight=1224+motu
might be of interest to you.
 
Actually, that MOTU thread gave me the idea of DIYing a wordclock that you could use to clock external devices. I've heard just about every audio engineer in the world say that higher-quality clocking improves the sound of interfaces quite a bit. I don't know how true it might be, though.

Why not just set up a simple crystal oscillator circuit and plug that into a wordclock input? I know it wouldn't allow for variable clocking rates, but I'm not too interested in working in anything other than 96khz anyway.
 
Heya guys...
>>It would be great to have a AD > lightpipe converter solution for protools where you DIY the AD with very high converters and bring that into the Digi002 through lightpipe.

Hmmm, this sounds like a great idea... Hmmm... Here's some food for thought :D

PGA2500's providing mic inputs, into some PCM1804 (112dB) or PCM4204's (once they are released) into an ADAT interface.
Now here's the clever bit - use an MSP430 to recieve midi control info from the Digi002 (i think it can send midi info too, can't it?)
Use the midi control info to control the input gain on teh mic inputs...

--why?? because you can :twisted:


As for generating an LRclock --- couldn't you generate it using a counter? For example, if your output bitclock it X bits per sample, as long as it's a factor of 2, you should be able to use some logic counters to generate the LRclock every X Bitclocks.

Maybe i haven't considered something here, but that's the way it fits in my head :green: -- I have no doubt i'll be corrected any minute... but it's worth putting it out there all the same :cool:
 
I think I posted something to this effect on old TT. If you get the adat card ($100) for the SM Pro audio Pr-8 Mark 2 (MK2) you could easily mod it to work with just about any pre-amp. Still don't know how its converters compare though. It's basically a self contained card (with clock selector, and wordclock bnc, and adat lightpipe all on the card) that has a 8 pin ribbon cable to bring in the 8 channels of audio and a ribbon to bring in power. I'm thinking of building a stand alone converter with it and use xlr's for inputs so I can use any pre amp I want to (and then bring it in to the 002). I think I could do the whole project for around a $120 or so (maybe in a half rack?). I have also seen mention that Sm Pro audio had plans for a firewire upgrade for this very same unit (Pr-8 mk2) so it would stand to reason that if they ever release it it will also basically be standalone which would make a really easy interface to mod (8 channel firewire with drivers already written :grin: ) Hope this is of interest to someone.
 
> Where would we begin

Define goals. What are we building???

For 2-input audio, I just buy the first "cheap or good" USB interface I see. I do a lot of work with a $23 USB sound "card"; it is reasonably flaw-free (unlike the botched onboard sound in the PC that I use it on). You can go up from there to $150, $500, and up. More bucks may buy more quality; mostly it buys more features (and sometimes more bugs).

What are we going to build with hundreds of hours of personal time that isn't being better done by the mass-market $23-$500 interfaces? Or by minor bodging of commercial boxes (replacement caps or external analog paths tacked into the ADC)?

USB1 is universal and ample for 2-channel at reasonable rates. USB2 is fully fast enough for much more, and universal on modern PCs. Firewire will always be a niche market, ADAT is a specialty, and PCI only works for desktops and maybe not for much longer (I bet it goes the way of ISA slots).
 
PRR...

there is a hive of activity these days around firewire. Whilst almost all pc's come with USB1 ports (and more and more with USB2), I've been told that there are inherant flaws within USB2 which doesn't lend itself easy to maintaining steady data flow into a PC.

Firewire's protocol has been designed around streaming media, and as such many of the companies I know of are developing Firewire solutions. The main reason we've seen a slow take up of firewire solutions is that they are so damn hard to make!

Let me draw you a picture of where we are at the moment with audio-interface silicon. This is what I remember from some reading I did a few months ago... I think it's 90% correct. (I'm just covering my ass, in case i'm wrong!)

There are plenty of USB1 stereo devices on the market, which are literally plug and play. TI has the PCM29xx family of USB devices, that you squirt raw analogue into, or S/PDIF, and windows&mac automatically detect it as a stereo i/o port. Nice... Lovely in fact. In such an example, the main drawback is sampling rate.

In terms of USB2, I can't speak for others in the market, but TI only has a USB-ATAPI bridge at this point. Realistically, I don't there's enough demand for a USB2 plug and play type solution for multichannel audio. (Sure, we're demanding... very demanding! but the market sizr is too small to justify that kind of investment).

Moving on to 1394... I've seen grown men shudder at the thought of 1394... It's tough - very, very, very tough, but the rewards are very obvious. High speed, low latency, stable clocked audio and video down the same piece of copper *bliss*.

There are 2 main competitors in the 1394 game (that I see on a regular basis). One is a swiss company called Bridgeco, and the other is TI. Bridgco's solution is aimed at audio companies. They have a 2 chip solution that uses their own ARM based processor, and a TI PHY (Physical Layer) chip. (PHY Layer is used as the final 'electronic' link to the 1394 bus).

TI's solution is a single chip beast. In terms of potential, it wipes the floor with the bridgco solution. BUT (and i'm not ashamed of using the word but when i discuss TI product). It's a beast to program, and support for anything under the big Japanese guys is hard to come by.

Before xmas, I had a little dig and prod into the commercial aspects of the solutions. The TI silicon itself is very cheap, but, we charge $$$$ for the software (well, a 3rd party does in fact). Bridgeco's solution is more like a pay as you go... you more per chip, but if you're building in low volume, it tends to work in your favour.

Anyway, that's where things were 6 months ago, maybe things have changed since then. Does anyone have any new news on these things?

Thanks again

R
 
> I don't there's enough demand for a USB2 plug and play type solution for multichannel audio.

You are wrong. Once designed, USB2 costs no more than USB1. You can sell USB2 to people who don't really need it. And with modern 5- and 6-channel playback, we are at or over the limits of USB1, even before you get to the weirdos like us who want 8 or 16 inputs.

If TI does not get a good solid many-channel USB2 chip to market fast, someone else will and steal your lunch.

FireWire and 1394 are too tied up with vested interests. Strong niche market, but never mass-market.

An emerging contender is plain old 100-speed Ethernet. The port is cheap, the speed can be ample, and many-many homes, offices, and probably studios are ethernetted. It also shares better than USB ever will.
 
PRR --

agreed... TI does need to get it's USB2 solution out. I'm not too sure of the status at the moment. I know nothing regarding it, so I won't comment.

As for the ethernet idea... it's already in practice as far as I know, there's COBRA, Sony's Supermac and a few others all vying for the ethernet audio crown.

We will have to see who wins on that one.

to move back to the original post.. I think what we really need is our own device (be it a cpld or a small micro) that can interface with a PC and squirt I2S straight into windows.

That way, even in the future, when better devices come out, we can continue to use the same interface into the PC over and over again. I think Pilo had the best idea of using a cpld or micro to make a PCI interface... although I have no knowledge whatsoever of such things. :oops:

Do we have any pc architecture specialists here?
 
heya guys...

seeing as there's talk all over the behringer thread about the firewire interface, I thought it better to put it in my original discussion in this thread.

Question to all: Does anyone have any data on the Wavefront Semi Dice II?

Cheers

R
 
okokok, I've never heard of this before... audio over ethernet?!?

There are existing standards for this??

If this is possible, it may be the golden solution :D Network cards are dirt cheap commodity... if we can make DIY converters that spit out ethernet, then software such as Nuendo can see an ethernet card as a multichannel audio device... man, that would be fantastic!

Could a gigabit network card carry a couple dozen channels of hi-res?
 
The problem with all these things is not data rate but priority.

Unlike normal computer operation files being copied etc ... the audio needs to be in proper relative time and needs to happen in real time.
This is not as that easy.
There is a huge difference with audio/video in a Post sense where sync of the output is all that is required.

When recording audio as an overdub, all things need to be in time and with latencies as low as possible.

Sure our computers go fast and even though these things are serial good stuff is possible.

As PRR said earlier,
what will you build ?
but
I'd like to add,
why will you build ?

1394 and USB do provide a future to your unit.
What if PCI were to be replaced soon ... ?


Personally I'd love to see a method to get audio and video around the house on Ethernet.
MVP's are out there
Multi Video Payer ... I think
These have nothing to do with recording audio as we do BUT I do think about using this type of technology for send multiple cues around the studio so each musician can tailor their mix.
 
Regarding Audio over Ethernet:

There are already 2 standards:
Cobranet - developed by Cirrus Logic - cobranet is designed to feed audio over already installed networks. You don't need specialised switches/hubs etc.

Downside is, that there is as much as 40mS latency.

The other standard is Sony Supermac, which at the moment uses cat5 as a medium, but at the moment, does not work over IP networks. At the AES, they mentioned Hypermac - their next gen supermac (1Gbit per second!)


Regarding the what and why (for PRR and Kev) this is the first rev of my answer:

What: PC/Mac Audio interface capable of transmitting and recieving I2S streams up to 192KHz. - How many channels will depend on the technology used.

Why: Because having digital outputs for many of the pre's etc designed here to put straight into a PC would make a fantastic addition to their functionality - as well as opening projects for multiple input channels (e.g. things like focusrite octopre etc).
I'm also interested in seeing how well this community can handle digital based projects. There are a LOT of skills here in the analogue domain. I'm wondering whats possible if we marry that with some digital know-how.
 
I like the ADAT interface because of it's simplicity, it's readily available and pre-programmed interfacing chipsets, and the possibility to buy inexpensive computer-end-interfaces with already-developed software drivers..

But if higher sampling rate than the offered 56K/s is needed, all ADAT's advantages goes away - because you'd need to build your own converter/audio interface/computer interface/software driver in order to acheive this. Which leaves you with only the lightpipe itself in common with ADAT.

And then there's no reason at all to use the lightpipe.

I tend to like the idea of audio over LAN - TC electronics uses this in e.g. the Dynaudio "Air" speaker system.

I don't know in-depth how standard LAN-chipsets are organized, and if there's any possibility of using them as real-time devices with appropiate firm/software - but if indeed possible, this option makes most sense because of the easy availability and high degree of standardisation of this technology.

But we still have the unavoidable problem of having to develop a software driver that will interface hardware to contemporary OS's - win, mac and linux.

Jakob E.
 

Latest posts

Back
Top