Re: [Gumstix-users] Audio output?

[Zach W. <zw...@su...>]

Timothy M. Shead wrote:
> Zach Welch wrote:
> 
>> Sound is a good question; it's something that I'm personally
>> interested in myself, but I have to admit that the Gumstix (as
>> currently designed) does not expose the ASSP portion of the PXA.
>> That makes it a rather "difficult" (read: nearly impossible?)
>> mission for now.
> 
> I assume that this is not as simple as soldering a couple of leads
> from the board to an amplifier?

Correct. The pins in question are likely tied down on the board - under
the uBGA part. Good luck getting at that. :/

> Yeah, I was afraid of this - I hope the Gumstix folks are listening
> and will reconsider audio capabilities.  My market is a hobbyist /
> niche market where I'd be looking at cottage-industry, "one-off"
> production to get started in the short-term.  This is why Gumstix is
> so darned attractive - those fabulously reasonable prices make the
> low-volume stuff look doable to me.  I've been looking at other SBCs
> that run GNU/Linux, but I haven't seen anything that comes close to
> Gumstix. Most of the developer kits are $1K and up, which is totally
> out of the ballpark for me.

They are listening, but this is just another example of the Wonderful
World of Business Pragmatics. Again, I would not be surprised to see
what you want show up eventually, but Gordon made choices that I think
are equally valid even though they serve different potential audiences.

Since he designed it, built it, and is now producing them, he gets to
make those decisions. Certainly, I'll be harping to see different
flavors emerge, but I recognize that there is far too much to sort out
with the current model before the notion of pursuing such derivatives
would be anything but operational insanity. Fundamentals first,
differentiated derivatives later.

> So how about it, guys?  What's a "collar-top" computer without audio
> - "HONK, HONK, HONK ... STEP AWAY FROM THE SHIRT!  STEP AWAY FROM THE
>  SHIRT! ... "

As I mentioned earlier, there is nothing preventing you from building a
proof of concept design with the existing board. For example, I want to
hack a simple piezo speaker to the HWUART (the second serial port) and
write a little bit-banging sound driver (since its pins can be retasked
for general purpose I/O). Sure, it will be *terrible* fidelity and
fairly high resource consumption; but I bet it could accomplish the
effect you described quite nicely with the proper tweaking.

The complexity of interfacing a "real" audio chip also makes this
approach more feasible from the perspective that there will be *far* let
skill required to get from here to there. Heck, I consider myself a
software guy (despite my rather extensive and growing knowledge about
building the hardware systems I can bring up) and I dreamed up this
simple idea because it's within my scope of skills.

Honestly, now that I've written this. I realized that it would also be
possible to drop a high gate count FPGA on a daughtercard, load it with
something like an I2C core, and accomplish some really decent sound with
the current board. In effect, you'd be designing your own sound chip;
now *that's* the software guy in me talking.... mmmm... VHDL. :)

So to clarify my earlier message: what you will *not* get from the
gumstix is the onboard AC97 interface. To do sound with the current
board, you would have to resort to another interface and possibly
assemble some cores into a logic device, though I suppose it might be
possible to find an I2C sound chip exists off-the-shelf. This stinks of
horrible hacking, though -- you really just want the AC97 signals off
the PXA. That said, these ideas are really only suitable to develop a
proof-of-concept while awaiting something better to come along.

Really, I think there's very little that can't be done with these boards
and some creative external hardware. Some solutions (e.g. audio and
video) may be possible but non-optimal today, but later flavors *could*
expose these fully and optimally. I personally think we can hold high
hopes for such emerging over time, as these are really nifty boards;
however, I would not encourage speculation on any timeframe for such,
nor firmly expect them to ever arrive - more pragmatics of business.

With all that said, who's gunna prototype my simple speaker design? :)
(I can handle the software aspects, but someone with some real hardware
skills would produce far better results than I could. I'll probably whip
together a simple hack here, but I don't grok analog electronics nearly
as well as I do digital.)

Cheers,

Zach