1
\$\begingroup\$

I'm working on a home automation/security system (Only in my head at this point) and stumbled upon RS-485 transceivers. using RS-485 with PoE seems to be like a great idea to keep cords to a minimum.

In a Cat5 cable, I'll have 4 twisted pairs, I'll need two conductors for the RS-485 bus, but I was hoping to use the remaining 3 pairs for a modified Power over Ethernet setup.

Can I run a long Cat5 and use 1 pair for a RS-485 bus, and the remaining 3 for a PoE bus? So I'll inject 48V DC into the other 3 pairs, so at each node I'll have 2 data lines, and combine 3 grounds and 3 VCCs.

I'm going to use Arduino's that hopefully will be drawing very little current (under 10ma each is my estimation) but can you bus PoE the same way that RS-485 lets you?

\$\endgroup\$
  • 1
    \$\begingroup\$ Does not seem like a bad idea, but you would need an RJ-45 socket without magnetics. Or just use a CAT 5 with screw terminals or some such. \$\endgroup\$ – Jan Dorniak Jan 19 '16 at 14:10
1
\$\begingroup\$

You can certainly do this within limitations. The main thing you have to be concerned with is the voltage drop that will occur along the wires. One reason that high voltages such as 48V are used is that the drop along the cables is a smaller percentage of the total voltage than say if 9V or 12V were passed down the wire. Put another way, you can pass more power down the cable at 48V than you can at 9V or 12V when living with a given cable voltage drop.

Now with RS485, even though it is a differential bus, the receiver parts used will have a maximum common mode voltage range specification that must be obeyed in order to ensure reliable operation without sacrificing silicon health. The voltage drop in the cables due to the power delivery scheme have a direct impact on the common mode voltage that the receiver will see with respect to its RS485-A and RS485-B differential pair pins.

Also be aware that RS485 modes are not particularly low power when you consider that the driver must be able to drive the terminations on the bus. Typically there are required terminations of 120 ohms between the differential pair at each end of the bus. This means that the drivers are sourcing and sinking current across a net 60 ohms when the bus is active. If your driver is pushing out 5V signal levels this corresponds to a current of ~83mA which is probably 5 to 10 times more than the typical low power microcontroller node current consumption.

\$\endgroup\$
  • \$\begingroup\$ Thank you, for some reason I didn't read deep enough into the chip I was thinking of using(MAX1483)'s data sheet. I now see that it will pull 240ma. I'm thinking though with a few large capacitors I should be able to smooth out any drop, only 1 slave will ever be transmitting at once so with lossless conversions I should be able to pull this off with only 1 pair for the PoE. \$\endgroup\$ – HilarieAK Jan 19 '16 at 17:12
  • 1
    \$\begingroup\$ Large capacitors will only help to the extent that you are able to isolate the local power circuit from the current flowing in the power source in the cable. You should setup some tests to evaluate this over the longest lengths of cable runs that you want to support. It can be very helpful to reference a multi-channel scope to one cable end GND and then monitor the other end of the cable's ground with a scope probe to see how the ground shifts happen whilst the RS485 is transmitting in both directions. \$\endgroup\$ – Michael Karas Jan 19 '16 at 18:18
  • 1
    \$\begingroup\$ No. I am not talking about the voltage delivery wires coupling noise to the RS485 lines - although that could happen I suppose if the voltage wire had current variation that was out of sync with the changes of the RS485 signal itself. My main point is that there will be voltage drop along the wires between the power source entity and the destination power consumption entity. This voltage drop can get significant when small gauge wires are at play over long distances. The voltage drop in the GND wire of the power delivery system will shift the common mode voltage at the receiver (continued) \$\endgroup\$ – Michael Karas Jan 20 '16 at 12:39
  • 1
    \$\begingroup\$ To put this into perspective lets say you have good quality 100% copper wire. Cat 5 would be 24 AWG at about 2.5 ohms per 100 feet. If you were running a distance of 200 feet and distributing 1A of current on the 48V supply the net voltage drop in the GND wire would be about 5V (i.e. 5ohms * 1A). (The V+ wire would have the same drop so the net voltage delivered to the destination would be ~38V). As you can see this 5V GND drop is a significant percentage of the RS485 common mode voltage range. Also consider the typical 3V common mode range produced by a typical RS485 driver. \$\endgroup\$ – Michael Karas Jan 20 '16 at 13:04
  • 1
    \$\begingroup\$ My points regarding the common mode voltage at the receiver due to shifts in the GND line are separate issues than the ones you start talking about due to the I R drop of the cable data lines them selves having to conduct the termination current. \$\endgroup\$ – Michael Karas Jan 20 '16 at 19:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.