slippyr4 wrote:
> The machine that will ultimately be running the software for this
> application is an Asus Eeepc. This doesn't have any PCMCIA or even
> ExpressCard slots. It has to be USB.
>
> I am still waiting for a replacement adapter to test with. However, I am
> currently mostly suspecting poor drivers. When I start to get bad data, y=
ou
> can see from the NMEA packets that some of it *looks* alright. Well, furt=
her
> investigation reveals that fragments of it contain NMEA fragments from 30=
+
> seconds ago (some of the NMEA sentences have the time in them). This real=
ly
> smacks to me of some sort of buffer over or underrun in the driver code.
>
> Obviously there is the possibility of a similar buffer error in my PIC co=
de,
> but I have added some extra code to zero the buffers before each use. Als=
o,
> an error in the code on the PIC would manifest itself irrespective of wha=
t
> kind of serial port it was connected to.
>
> I have a couple of the FTDI chips in my drawer, I guess the optimal solut=
ion
> is to integrate the USB-Serial converter into my project. I hoped that th=
is
> would not be necessary because it's added complexity, but if that means
> reliablility then I shall have to do it.
>  =20
Note that FTDI sell their converters in a number of forms other than=20
bare chips including

Conventional USB to serial cables (with the USB-serial chips built into=20
the serial end afaict)
Cables with the USB to serial built into the USB plug and bare ends at=20
the other end of the cable. These are available with both RS-232 levels=20
and logic levels (both 5V and 3.3V).
Modules designed to replace a 9-way D connector (these are rather=20
expensive though)

So you can probablly come up with a soloution that forces the use of an=20
FTDI converter without having to integrate the chip on your own board.
> Thanks for your help so far everyone.
> Slip
>  =20

--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.