9 data bits?

[atk_nut]

I have an older obscure peice of hardware that has a serial protocol with 9 data bits. The protocol is as follows:

Baud: 4800

1 start bit + 9 data bits (described below) + 1 stop bit

No parity bits

Byte 1 - channel id: 1nnnnnnnn

Byte 2 - channel level: 0xxxxxxxx

I've tried faking it out in various ways using Mark and space parity, but it doesn't work.

Has anyone out there done this before?

Thanks.

[Barrie]

I have an older obscure peice of hardware that has a serial protocol with 9 data bits.  The protocol is as follows:

Baud: 4800

1 start bit + 9 data bits (described below) + 1 stop bit

No parity bits

Byte 1 - channel id: 1nnnnnnnn

Byte 2 - channel level: 0xxxxxxxx

I've tried faking it out in various ways using Mark and space parity, but it doesn't work.

Has anyone out there done this before?

Thanks.

4966[/snapback]

I haven't found any USART in a PC that supports 9 data bits, but I implemented a serial interface using one of the bits in the parallel port and then doing the framing and synchronization in software. If you need bi-directional communication, then you will need to use two bits and write a transmission driver as well.

You can access the parallel port using the Port I/O primitives in the Advanced Palette.

A few things to watch out for:

-Depending on the PC, the parallel port can be found at one of two base addresses. One is "standard" and the second, less frequent is often used in laptops.

-The chipset controlling the port most often emulates the original chip, an 8255. It is programmable so you may need to access various registers at Base+0, Base+1 and so on, to change the modes available.

- You will need to give the serial routines a high priority because there will be no hardware buffering. If the routine isn't running, you will miss data. I used a subroutine priority for a small part of the code which just look for state changes and then buffered the data.

Sorry I don't have any specifics right at hand but connector pin-outs and register information is easily found on the 'net.

It helps if you have some hardware and register-level programming experience, but it's not as tough as it may sound. I was pleasantly surprised when I found out how fast the final product was (~120Kbps), even though it was implemented in S/W, so 4800 Baud should be a piece of cake, particularly on a contemporary machine.

This should give you a head start if you chose to take this path.

Good luck!

Barrie