Hi all,

The design of Balloon 3 is coming along well, and has reached the point 
where there are some niggles to resolve.

The first of these is the pinout of the backplane (Samtec) connector. 
Many of the pins will retain just the same functions as they have in 
Balloon 2 (the memory controls, data and address buses, PCMCIA, LCD and 
to some extent power) but others are going to have to change. The ones 
which change fall into a few areas:

1. Audio
The backplane on loon 2 has the I2S signals for the audio codec on it. 
Balloon 3 will use an AC97 codec, and therefore these signals become 
meaningless. There are a couple of choices here:
- present the AC97 audio signals on some other pins and leave the 
original I2S pins unconnected. We'd have to find 5 spare pins for this, 
which isn't too hard.
- wire the AC97 signals to the old I2S pins in such as way so as to 
allow I2S to work as before when the PXA270 is put into I2S mode (which 
it can do) and find one more pin for the remaining AC97 signal.
The latter strikes me as the most efficient option, allowing off-board 
codecs of both AC97 and I2S varieties without wasting pins. Thoughts?

The backplane on loon 2 also includes a microphone input. Loon 3 won't 
have a microphone input, only a line-level one, to ease noise problems. 
Should we present this on the same pins, or different ones?

2. GPIO pins
The loon 2 backplane presents GPIOs 0, 1 and 10-27 from the SA1110, most 
of which have well-defined functions (power switching, I2C, various 
interrupts, and so on). Not surprisingly, the GPIO allocation on loon 3 
is entirely different due to the PXA270's enormous selection of on-chip 
peripherals. Here there are a couple of options, I think:
- slavishly wire up the same GPIOs 0, 1 and 10-27 from the PXA270. This 
would keep the labels the same but render the pins almost entirely 
useless. Some of them would even be duplicates of signals present 
elsewhere. I vote we don't do this.
- find nearest equivalent signals for each pin as far as possible, and 
try to put general-purpose I/Os on the remaining ones. This should make 
it possible for backplane devices to work based on the signals they 
assume are present, but will present some software compatibility issues 
if anyone is doing direct GPIO twiddling, which they really shouldn't be 
;-) Inevitably there will be a few pins with no equivalent (things like 
the onboard ID signal and RTC interrupt) but we can use them as general 
purpose signals.

3. Serial ports
As luck would have it, PXA270 has 3 serial ports just like SA1110. I 
propose the following arbitrary mapping of one to the other:
SA1110 COM1 (pinko): PXA270 BTUART (doubles as Bluetooth)
SA1110 COM2 (bottom serial port): PXA270 FFUART (full-featured UART)
SA1110 COM3 (console, top serial port): PXA270 STDUART (doubles as IrDA)

Any complaints?

4. Other peripherals
Because the PXA has a bucketload of other peripherals on board, we could 
bring some of them out to the backplane connector. I can see 30 spare 
pins on the Samtec connector. Peripherals not currently exposed on the 
Samtec include:
- AC97 audio (but see above)
- extra LCD signals (L_CS and L_A0)
- FFUART serial handshake lines
- MMC/SD
- MSL (high speed parallel-ish bus)
- Camera interface
- Keypad (but these are mostly getting used as power control lines etc)
- Samosa bus
- USB OTG stuff (eeeuuurgghh)

Any votes? My top choices would be the extra LCD and serial signals, for 
completeness, plus MMC and Camera in that order. MSL is mostly intended 
to go to the onboard FPGA.

If there's anyone you know is not on the balloonboard list but will be 
affected by these decisions, please let me know and I'll copy them on 
the mail. All feedback most welcome, otherwise I'll make arbitrary 
decisions and not let you blame me for them...

Chris
-- 
Chris Jones - chris@martin-jones.com
Martin-Jones Technology Ltd
148 Catharine Street, Cambridge, CB1 3AR, UK