You could pull the IrDA interface - this goes to an IrDA encoder chip (due to bugs in the SA1100's internal irda encoder). You're then left with a 3.3v-level serial port that will go up to 230,400bps - ideal for bluetooth.

This sounds interesting. How difficulty you think it would be to grab the TxD and RxD lines from the IrDA interface?

http://www.blueunplugged.com has modules designed for embedding (see the "development" section) for UKP 85+VAT.

Nice. Would it simply work if the TxD and RxD lines from the IrDA interface would be connected to i.e. Ericsson ROK 101 008 module (which you referred)? The chip would seem to run on 3.3V logic, so it should be fine.

On software level, this would provide a HCI-level access directly from /ttyS2 which could then utilize i.e. Axis BT stack which provides layers upto RFCOMM which neatly connects to i.e. pppd directly.

This kind of hack would be easiest to do into a small PCB which resides inside empeg for easier connectivity. Otherwise, taking the "second" serial port out of the IrDA interface and carrying it outside of empeg is difficult since the docking connector does not do the job.

Real, external serial modules that are a bit more robust aren't a lot more, though.

Do you have some other links in mind? At least all of the Bluetooth devkits are ridiculously priced. They are usually more than 10 times expensive compared to the Bluetooth core on its own.

Hugo
(who's just got a T68 with bluetooth, replacing his 6210+bluetooth battery)

Drool... I've been looking at the T68 also, seems a way superior to any competitors. Even the upcoming 6310 loses the battle easily.

Kim