hack #1?

Hi all. I've begun setting up the serial comm setup for the back-end. As expected, I've come across a few hangups while designing it.First the good news: the CON1 header provides me with a nice interface to the various GPIO pins. A section is connected to the LEDs, and another is connected to the switches. There are also 4 unconnected pins connected to solder pads on the reverse side.

Unfortunately, the DSP36761 does NOT have a UART, which would require me to bit-bang the RS232 interface. This is achievable, but not as elegant as I would like. Once the real DSP stuff gets running on the chip, I can foresee some communication problems (missed bytes, corrupt data, etc.) Therefore, I opted to use the SHI peripherals that the DSP36761 provides: SPI and I2C. As mentioned earlier, the CON1 header has connections to the GPIO pins, but NOT to the SHI pins... but remember what I said about the open solder pads on the reverse side? In the manual, it suggests that these pads be used as endpoints for jumpers to the I2C pins that are underneath one of the switches.

Now that that's done, I still need to somehow convert the I2C protocol stuff to RS232. Well I thought about it for a bit, looked around my shoebox, and realized I could use my leftover dsPIC to handle the comm with the front-end (implementing the protocol state machine) and convert the I2C bytes to RS-232 bytes. The MAX232 would be connected to the dsPIC's UART pins for some level-translating action.

edit: I just finished whipping up the schematic and layout for the I2C <-> RS-232 board. I'll put in an order to BatchPCB.com and get this thing spun in a jiffy.

edit 2: I just realized that since the I2C bus is essentially visible through the RS-232 serial port, it'll be possible for the EEPROM (where the code is stored for the DSP) to be updated through the serial port. Field updatability perhaps?
- Marc

Schweet! The DSP board came in!

Hey ya'll. Our DSP dev kit came in recently. As you can see from the picture, it's from Freescale semi'. It's called the "Symphony SoundBite" and it basically is a DSP platform designed for audio signal processing. It features the DSP56371, a codec, and some GPIO pins. Another cool feature is the built-in debugger with a USB interface.

We've already whipped up some code, based on some example stuff from Freescale, that passes-through a signal directly from the inputs to the outputs. Unfortunately for us, the GNU C compiler that targets this DSP is crap. The IDE, which is based on Eclipse, is nice, but also still very immature. From what I've seen on the Freescale forums, they don't seem to be supporting this board as much (GCC 1.37.1 anyone?) as their more expensive offerings, which is kind of understandable considering the board's target audience: hobbyists and cash-strapped university students. There's another tool chain, Altium TASKING for DSP53xxx, that provides nice DSP intrinsics for maximizing the special features of the chip, and a mature optimizing C compiler, but the 15-day trial and it's $500 price tag is much too prohibitive. Therefore, for us to get performance out of the chip, it looks like we'll be coding the computationally intensive stuff in ASM. On the upside, learning how to program this thing at the machine code level will definitely be a fulfilling experience because we'll be exposed to inherent architectural differences a DSP has between a regular general-purpose CPU. Besides, no one said the project was gonna be easy.

Aside from our previous little pet-peeve, we think this platform is a great enabler for what we plan to do, and it'll definitely provide us with some valuable real-world experience.

As for the front-end, Andy ordered some 40-pin DIP micro-controller samples from Microchip, as well as a programmer/debugger. We already came up with the command protocol for the application layer of the serial I/O. The preliminary command set can be found on our wiki.

We think we're off to a pretty good start. After speaking with Professor Jenks about our project, we now have to start making progress with our spec document. Anyway, we look forward to a very fun year and hopefully we'll be able to show something interesting for demo day!

First post

Welcome to the spkrtwkr blog!

The following is the location for the team wiki, which should be publicly viewable shortly...

http://hyperintense.com/wiki/index.php?title=Main_Page