EIA 485 hub?

[Eriond]

Hi,

I've spent quite some time googling around for a schematic/design/example of a 485-bus hub. All I found was two commercial devices that allow for up to 8 buses to interconnect, and those were quite expensive imo.

The basic idea is to use n x MAX485 (or equivalent) and then wire the Rx (RO) from one unit to the Tx (DI) of all other units, and doing so for all the units. Somehow I get the feeling that it won't work... or maybe it will?

Before You start questioning why someone would need a hub for 485; well the cabing in my home has a star configuration, and I don't feel like redoing it over again...

Any ideas are appeciated!

Regards

 

[Sceadwian]

You'd have trouble with bus contentions, you'd need something to act as a traffic controller, a micro controller comes to mind.

 

[JimB]

Assuming that the devices on your network are truly RS485, ie only one master device, two wire, half duplex, and the cable lengths and the comms speeds are not too great, you could try connecting them together in a simple star configuration without the "hub".

The problem with such a setup would be where to put the terminating resistors? Which is why I wonder about cable length and comms speed.

Your original idea would be OK if you were using 4-wire RS422 with a transmit pair and a recieve pair to each device.
You could make the "hub" as a simple splitter, connect the master device to the common port and then radially connect out to the slave devices through the splitter.

JimB

 

[Eriond]

Cable lengths are at most 5-6 meter, standard 2-pair indoor telephony cable. One pair is for supply (12V), and the other would be half duplex 9600 bps traffic over EIA485. Maybe, I'll allow burst-mode with greater baud rates if the connection proves error-free.

Having some time to think, I realise that the master will reside where the "hub" is, so maybe I'll just use a bunch of controller pins to enable one 485 tranceiver at a time. In practice it will result in a number of point-to-point connections.

The drawback is that it will become a true master-controlled system, where no peripheral device can call for attention without being asked first.... Currently I'm looking at 7 or 8 individual "buses" for the master to talk to, so each "session" must be kept short, in order to poll them all so that near-realtime efficiency may be reached.

I'll make a prototype and see where it goes!

Eriond

PS. Edit: Just a quick schematic with two demuxes (138/238) and a Rd/Wr control pin. Only 4 of 8 tranceivers visible.

 

[saturn1bguy]

Hi,

I've spent quite some time googling around for a schematic/design/example of a 485-bus hub. All I found was two commercial devices that allow for up to 8 buses to interconnect, and those were quite expensive imo.

Before You start questioning why someone would need a hub for 485; well the cabing in my home has a star configuration, and I don't feel like redoing it over again...

Howdy,

Yes, turning a bus into a star is not necessarily easy. But you'd rather turn a funky circuit instead of rewiring?

Given that EIA-485 is a real multipoint bus, and the stubs from said bus are to be kept as short as possible, all that comes to my mind is collapsing the bus into a box, your hub, and use bus repeaters to each leg of the star.

Use a National DS96176CN (EIA-485 transceiver) at each remote node (with proper termination); and your hub would comprise the proper quantity of DS96177CN bus repeaters all interconnected. And with that short of a collapsed bus within the hub, you may not even need termination. However, each outgoing leg still requires terminators.

(Addendum: I just looked at the bus repeater again, and it's only unidirectional. So, never mind...

Just a thought,
Corey