a little help needed about RS422 communications

[KTP]

Hi guys,

I need a way for the Atmel Xmega32 chips to communicate with each other on the RGB led matrix boards I am building.  There could be up to 100 of these boards, and I would like them all to be wired point to point in series with just a single transmitter for data.  The boards only need to listen to the transmitter, no ack or handshaking.  I think RS-422 only supports 10 devices hanging off a bus and RS-485 supports up to 32 devices (assuming this has to do with the capacity of the differential drivers).  Since I only need receive on each board, I was toying with the idea of using a transceiver chip like National DS8921 to act as a repeater by connecting the data out of the differiential receiver to the data in of the differiential transmitter, and then also running this data line to the Xmega32 UART.  I would have a termination resistor on each board.  Do you think this sounds crazy or would it work fairly well?  The chip specifies 12ns prop delay but that wouldn't really matter as I don't care if the 100th board gets the updated data 2uS after the first board.  Most of the RGB matrix data is going to be stored on a serial 8 pin flash chip on each board anyway.  Here is a link to the 8 pin transceiver I was considering:

http://www.national.com/ds/DS/DS8921.pdf


Thanks for any suggestions/help.

[arcom]

There are RS485 driver ICs from Maxim which have 1/2, 1/4 and even 1/8 unit load which means that you can connect up to 64, 128 or 256 devices (respectively) on a single RS485 bus.

Some examples are: MAX487 (1/4 load, 250Kbps), MAX1483 (1/8 load, 250Kbps) MAX1487 (1/4 load, 2.5Mbps), MAX3061 (1/8 load, 500Kbps, fail-safe, hot-swappable), MAX13485 (1/4 load, 500Kbps, fail-safe, hot-swappable).


Your solution will also work but I would go with the RS-485 driver ICs unless you have long distances between RGB matrix boards.
I'm actually working on a similar system for garden lighting -> one master controller and a bunch of slave units communicating over an RS-485 network using MAX3061 ICs.

[KTP]

Thanks.  I didn't know a 256 drop chip existed.  It is a bit more expensive than the other chip, but would give the option of communication back to the host (although I can't quite see why the rgb boards would need to)

[Psi]

Could you perhaps raise the voltage of the transmitter above the typical -6V / +6V and then use two simple voltage dividers to get the voltage back within RS422 spec on each board.

Might require a bit of testing but it might allow you more devices.
As i understand it, too many devices load the line too much and drop the voltage too low. So a higher transmit voltage should make the voltage drop less critical.

Might end up being a bit to messy though.

[arcom]

KTP> You could get temperature, current, voltage readings from the RGB boards (I know, overkill...just the way I like it ) as well as their settings/memory data and then display faulty nodes on an LCD screen. On the other hand, you don't have to implement bidirectional communication...just leave it unidirectional.

[RayJones]

Thanks.  I didn't know a 256 drop chip existed.  It is a bit more expensive than the other chip, but would give the option of communication back to the host (although I can't quite see why the rgb boards would need to)

Actually the way I read the reply is the *receiver* chip would impart less load on the bus for the same driver source....

You could break the system down and use repeaters, but that exposes you to problems with broken links taking out large parts of the system.
This has been discussed to death in another thread recently.

[Jimmy]

I did the something simalar I used one Master to drive 8 slaves over wireless. But wired would work the same. Do you want the slaves to do exactly the same thing or do they need to be addressable