theorie cablage rs485
VVV said:
Absolutely. Just use termination resistors (120Ω) at both ends of the line. That would be the right-most slaves on each floor.
Long 'stub' from unit to 'T'-connection is not recommendate - 2 meters seems little bit long.
Think this simular SCSI and 10-base-2 - ie, no long cable from 'T'-connection point to unit , or this can make 'strange' problems later.
newer make 'star' connection point (and stubs in different way), only daisy-chain connection between units whole way from terminator in end to terminator on other end. (your 'split' between floors is a 'starpoint' here, and can make signal reflexes and possible ruins datapackage) - again - use same rules as 'old' 10-base-2'
Use very slow datarate (< 20kbit) or very fast - If possible, not use range between 20 KHz to 100-200 KHz - all signal cable type (including CAT-5) have bit a trouble in this frequency range and needs complex source and complex terminate to working well - easiest to skip... (Why RS232 at 115200 Baud not working on long cable...)
100 - 120 Ohm terminate (depend of cable - CAT-5 uses 100 Ohm) is only valid for high speed datarate above 200 kHz (>= 400 kbit/s).
Make if possible communicate-protocol not level sensitive, only level change-sensitive (ie use 8-16 bit sync sequence before valid data in every datapackage) if you want break up RS485 line galvanic between zones via trafo.
- using optocoupler to transport logic level state all time is lot more expensive compare to using simple transformer.
test system with isolating trafo simular ethernet-trafo and long lines before setting up equipment on customer...
(all 10-base2 and 10-basT, 100-base-T use transformer between each segment to avoid later problem in time network growns and connected equipment coupled to differ power source and grounds.)