Does there exist a simple ADC guitar interface project?

[edvard]

I searched the forum quite diligently, and couldn't find anything, so maybe you folks know...

Generic soundcards such as most normal folks can afford (still scraping pennies together for a M-Audio delta 44) just won't do for guitar-level signals and impedance, so the idea of a simple guitar-to-computer interface like the Behringer UCG102 is quite attractive.
So, that got me thinking; is there a project for a simple USB Analog-to-Digital interface like the Behringer?

I know, I know... "those things are only like 30 bucks, what's the hang-up?"
Well... many folks on this board will readily admit to having spent hours of time and sometimes insane amounts of money building something that often costs no more than $50 of the shelf (yours truly included).
Just sayin'...

So, how about it?
Simple DIY USB ADC anyone?

[zeptin]

Well, it probably won't be as simple as you'd like, but a device like the PCM2902B (http://www.ti.com/product/pcm2902b) can be used to create a USB soundcard without too much trouble, assuming you can create an appropriate circuit board. You could add a simple buffer and preamp to the input to make the guitar signal more friendly to the codec. The only slight drawback is that the codec only supports 16 bit samples, which is a little lower than most audio interfaces.

There are plenty of USB soundcard projects online if you look around for them, I can't vouch for any particular one though.

[artifus]

The only slight drawback is that the codec only supports 16 bit samples

plenty for guitar

*edit*

twenty years ago 12 bit was da bomb

[nexekho]

The only slight drawback is that the codec only supports 16 bit samples

plenty for guitar

*edit*

twenty years ago 12 bit was da bomb

Agreed on this - sample rate seems to matter more.

[markseel]

I have an XMOS based PCB design that contains a AK4556 audio codec.
The board is arranged as a 24pin 600 mil wide DIP.
You could hook up a few parts for the analog path (like an opamp for a hi impedance guitar input).
Then connect one the the board's GPIO's to the RX pin of an off the shelf USB to serial board.
With that system you could use the PC's standard COM port to stream in the audio data.
Just yesterday I was reading RF data (at work) with this arrangement at 2.89 Mbits/sec.
I was using the board to 'sniff' RF activity using two TI radio IC's.
One 24-bit audio channel at 48 kHz is 1.152 Mbits/sec.
Latency from guitar to com port might be really small too, not sure though.
I have code for the XMOS to do this if you want it.
Boards could be made but I'd need some time to get stuff ordered and made and tested.

[Gurner]

IMHO, life's to short to be climbing mountains just "becuase they're there", so I'd tackle this with a $0.30c opamp to change the impedance/level so it suited my existing soundcard!

About an hour's work from blank paper design to 'operational' ....job done.

[markseel]

Good point.  If you're goal is to cross the mountain then drive or get a taxi.  Some people like the experience and so going hiking, although it's work and takes longer, might actually be an enjoyable experience.

I've put together countless analog effects and created the SimpleDSP and XMOS boards (postings on this forum) that resulted in some good experience.  I'm a firmware developer for Wireless systems.  Doing HW design, PCB design and analog/digital signal processing as a hobby has cost me money and time but helped me in my career.  And it's been fun!

I totally agree with you, a simple high-impedance buffer, proper level matching and a sound card will get you there.  But, you'd have to deal with audio system architectures for Windows, Linux and Darwin that impose latency.  Latency has been discussed on lots of forums so it can be mitigated.  You'd probably be able to have something up and running in no time.

I guess DIY means different things to different people.  But thanks for the opinion Gurner - it's good to have different perspectives and put them out there - that's what forums are for!

[edvard]

Thanks for the replies all...

I have considered the opamp-buffer-to-soundcard route, but I'd need to power it with yet another wall wart.
Then I thought since all I'd use it for is a buffer to the computer, I could just steal power from there, which led me to the question of how easy it would be to build a USB interface.
I could make it high-impedance, shape the tone, it would be self-powered; seemed like a sweet deal all around, and I'd learn a thing or two about hardware-software interfaces.
I'm running a 64-bit Linux system with Realtime Pre-emption in the kernel, so latency would be pretty low, and there are existing USB audio drivers I could hack.

So, yeah I was just wondering if the long way around would be scenic and educational or just a pain in the neck.
I'll check out the TI chip and Mark's XMOS project and see where I can go with it.
In the meantime, I'll just steal power from an open USB port and use a low-power CMOS op-amp or FET for the head.
5v is plenty for straight guitar signals.

[Gurner]

Just be aware that 5V sourced from a PC USB port can be very noisy ....the 'ground' is normally polluted to hell (it caught Terratec out...I bought one of their guitar to midi converters an AX50 ...a very well regarded unit, but it was virtually unusable to me, as the noisy ground from the PC, permeated it's way into the analogue signal...resulting in whine in the guitar signal - check out the number of posts complaining http://www.google.co.uk/search?hl=en&qscrl=1&nord=1&rlz=1T4GGLL_enGB401GB401&biw=1603&bih=833&site=webhp&q=+site:axon-technologies.net+axon+ax+50+noisy+ground+forum ) ....if I was after sourcing 5V, then a PC would be the last place I'd look. Also if I was constructing a preamp/buffer for a guitar from scratch I'd likely roll with a 9V supply.

[markseel]

Here's links to the xmos board layout and a schematic for hooking it up.
The guitar signal can be pre-processed in the xmos if desired.
The adc, dac, dsp and uart functions on the xmos all run concurrently.

https://picasaweb.google.com/m/viewer?dc=gorganic&source=mobileproducts#photo/mark.seel/5468306090311368737/5668985290634797330

https://picasaweb.google.com/m/viewer?dc=gorganic&source=mobileproducts#photo/mark.seel/5468306090311368737/5669015900919691106

[edvard]

I read through the specs for the PCM 2902B, and it supports up to 48kHz (fairly typical) but I read on the diyaudio forum that it can be clocked to 96k if I don't use the SPDIF, but that it's effective bandwidth is internally restricted to 24kHz.
The poster expressed some misgivings about that, but due to Nyquist aliasing and all that, I thought all computer audio equipment had pretty much the same limit, which is fine with me.

@Gurner: 
Thanks for the heads up.
I should have figured as much, computers are not audio equipment by nature.
Is it not something a well-planned capacitor network can't fix?

@Mark:
Thanks for the links.
I've got some more reading to do. 

[Gurner]

@Gurner: 
Thanks for the heads up.
I should have figured as much, computers are not audio equipment by nature.
Is it not something a well-planned capacitor network can't fix?

No, If your ground is noisy...by & large you're screwed - you need a good clean solid ground to shunt all the junk too....but if the ground is the source of the junk, it's gonna inject a little bit of whine/noise into your signal at every opportunity.