Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
09/10/06 03:16
Read: times


 
#123980 - Best?
Responding to: ???'s previous message
Like most things in engineering, everything is a compromise. What is 'best' may not be available!

Most of my stuff uses 485 and 24VDC down some spare pairs with distances of 500-1000m. But you need to do your calculations!

Firstly - 150-200nodes is way beyond the RS485 spec. I think you'll have loading problems. For example: The Schneider TWIDO series of PLCs have 4k7 bias resistors in each unit and they tell you not to use more that 7 units per bus. Why? Calculate the parallel resistance of 7 4k7 resistors in parallel. Each RS485 transceiver chip represents a specific load. What is the way around this problem? I use a RS485 hub and distribute the slaves over a number of busses.


Sending volts down the spare pairs represents another problem. It all comes down to OHM's law. The cable itself has a specific resistance - this may dictate what cable you use. Standard CAT5 has a resistance of around 100ohms per 1000m (of the top of my head - check this!). So you need to calculate the expected voltage drop across your cable. This might preclude the use of CAT5 simply due to voltage drop. With my equipment, I use a Belden 4pair twisted cable with steel wire armour that has bigger conductors and I use two pairs for the DC volts thus I have less voltage drop. Other techniques I use for this style of installation are:

DC/DC converters - A LM7805 is a linear regulator - thus it wastes the excess voltage as heat. For 24V in and 5V out at say 100mA, you've got (24-5) * 0.1 = 1.9W of heat. This means your node is also pulling 100mA at 24VDC. Using a DC/DC converter using switching techniques, less wastage occurs. In this instance, the current draw for 5V at 100mA on the 24VDC side is: (5 * 0.1)/24 = 20mA (approx) and assuming a 75% efficiency is about 26mA.

Voltage drop across 0V - this can cause the system to not operate as you may exceed the input voltage range of the RS485 transcievers. Depending on the exact type of transceiver, the exact voltage may differ. I use isolated RS485 transceivers with their own 0V wire - voltage drop across the DC 0V does not have any effect.

Protection: most 485 tranceivers don't appreciate having 24VDC put into them. Running DC with your comms - there is a fair chance that due to a wiring error, that 24VDC will find its way there. I use SA5.0 transzorbs from 0V to each dataline and a 100mA polyswitch in series with each dataline. With this, you can have 24VDC on your comms all day with no smoke. There's also a fair chance that the power will also be wired backwards - so you need a series diode or some other means to ensure that no smoke happens.


It might be easier to use CAT5 cable for the comms and much thicker cable for the DC volts and run them in a conduit for protection. This may be cheaper and just as effective as finding a specific cable to suit all the requirements. As Kai mentions - you need the twisted pair for the comms.

Also worth mentioning - keep the RS485 cables away from AC wiring - otherwise you will have problems!


National Semiconductor have a document '10 ways to bulletproof RS485' - this is a good read. The points I mention above have been gleaned from years of experience with RS485 in hostile installations - reliability is the result.








List of 9 messages in thread
TopicAuthorDate
Best 485 Cable In Industrial Environment            01/01/70 00:00      
   Think about why twisting is provided at all!            01/01/70 00:00      
   Best?            01/01/70 00:00      
      no, Russell, it is not            01/01/70 00:00      
         Yes, indeed            01/01/70 00:00      
      Thanks Russell and Kai...            01/01/70 00:00      
         Additional hint...            01/01/70 00:00      
            the slew rate limit is nice, but even better            01/01/70 00:00      
   i think stubs and termination will be more of a            01/01/70 00:00      

Back to Subject List