You don't say what kind of "power" is in the lines. AC 120v? DC 5 v? In any case, you will have a hard time unless you abandon the baseband and use some sort of modulation. The impedance of any type of power line is going to be very low at mid frequencies. Bigger caps would just ask more of your "RS485" drivers - probably more than they can deliver. But if you can get all your signaling up above 10kHz, then you can separate power from signaling using caps and inductors.
-Robert Scott Ypsilanti, Michigan (Reply through newsgroups, not by direct e-mail, as automatic reply address is fake.)
I'm trying to communicate between multiple devices and I only want to run 2 wires between devices. I want the 2 wires to be used for both power and communications. I've tried capacitive coupling the output from the RS485 chip onto the lines. I have inductors between the power supply and where I'm tying the caps into. When I have the power disconnected, I get a fairly normal looking signal on the lines (a square wave with an initial spike). When I apply power to the lines, I only see an initial spike and then an almost immediate return to 0 (I'm only looking at the AC signal). Would different size caps or inductors help?
I've seen a simular system work that used bi-directional RS485 over power and the polarity of the wires didn't matter (I didn't get to spend a lot of time looking at the system, only enough to verify that they are using RS485).
Just goes to show that there's lots of clever engineers out there (:
I honestly don't know how this could be made to work unless there was some kind of carrier. That the polarity didn't matter is really confusing.
Think about a current-based system instead of voltage. I once did a system that would power and communicate bidirectionally over a twisted pair for something like 4 miles. What I did was force
100 ma into the pair from one end. At the other end, the pair connected to VCC and ground of the board in parallel with a
5 volt zener. The board drew about 50ma and the zener burned the rest. Now at each end I had 600 ohm transformers in series with the loop and I used the secondaries to send Bell 103 modem signals for bidirectional communications at 300 baud. There was other trickery involved, but the end result worked very well for the application.
If you look carefully, you probably notice that the signal is a positive going spike that coincides with the leading edge of the data bit square wave, and a negative going spike that coincides with the trailing edge. Manchester encoders and decoders are typically used in this type of application since manchester code is concerned with transition logic rather than logic levels.
We used to run upto several megabit/sec data rates over high voltage AC power circuits for 10's of kilometers.
This really doesn't seem that hard with a totempole driver for each side of the pair so you can drive some current onto the line.
At the receiver end, you rectify the 'data' and use a cap for filtering (of course with some regulation)
So long as you don't draw more current than your driver can supply you should be able to happily read the data using a standard RS485 Transceiver.
If you need By directional communications, you clearly need enough power stored at the slave to drive the bus for the whole transmission, however the slave only needs to drive voltage so can be low power.
That's true, but there are other options too, depending on your interpretation.
We run comms over power in our new multi-drop industrial products, and each drop can be connected either way around. That works because the comms are coupled AFTER the bridge rectifier, so the polarity gets re-aligned to the power polarity.
Also, while Manchester and many other coding systems fulfil the need for zero DC, it's also possible to do it on the cheap using a simple UART as long as you send only characters with equal 1s and 0s, and use a look-up table (sacrifice a couple of bits per byte, eg send 4 characters for every 3 bytes - more efficient and cheaper than Manchester).
Regards, Tony (remove "_" from email address to reply)
Sorry to be pedantic, but you're talking about serial comms over power - which is not RS-485, which is a physical hardware spec. While you may well be able to use RS-485 drivers/receivers in some way, it won't conform to the RS-485 spec.
I'm pointing this out in the hope that it helps you to think outside the envelope, i.e. forget about RS-485 and just think in terms of data-over-power.
There's no requirement for modulating a carrier. You can do Manchester encoding and isolate the drivers/receivers appropriately just as you would isolate the modem. Poor isolation will result in a blown-up modem just as it would a blown-up line driver/receiver.
Grant Edwards grante Yow! I'm into SOFTWARE!
This is on a DC power rail, not the AC mains. But regardless, an error correcting protocol should be used even if it is just as simple as adding a CRC to each message and retransmitting ones that are corrupted.
Rick "rickman" Collins
Looking at stuff like ProfiNet, DeviceNet, and Foudation Fieldbus would be a lot more applicable. They all do power/comm over twisted pair for microprocessor based devices in industrial environments (baud rates in the 32K range).
Some of them are even intrinsically safe.
Grant Edwards grante Yow! Yow! Are we in the
at perfect mood?
In article , email@example.com mentioned...
There's no requirement for receiving uncorrupted data, either. By putting the data onto a carrier, the receiver can then bandpass filter the incoming signal and eliminate a whole lot of the garbage and 60 Hz that's already on the AC line.
Another helpful procedure is to detect the AC zero crossing and transmit during that time. That also saves a lot of corrupted data.
You just have to realize that the AC line is a very hostile, noisy medium, and not using some of these methods opens up your data to total, unrelenting corruption when the first vacuum cleaner is turned on and those nasty brushes and commutators send a ton of hash and noise down the line. So you don't use these methods, and you wait for the data to start to pass again after m'lady is finished vacuuming in a half hour. Your choice. :-)
And this doesn't include all the times when you use the electric drill, the blender, the portable saw, the weed whacker, the light dimmers and even the newer fluorescent lights that have the electronic ballasts that put out a lot of interference at 20 to 50 kHz.
Or you can just skip trying in vain to use the AC line as an error- free transmission medium and instead use RF at 900 MHz or 2450 MHz, like cordless phones and such do. It will probably result in a lot less complaints from users about not being able to receive any data while their significant other is vacuuming!
Oh, did I mention the arc welder? But then, that probably doesn't happen very often. Unless, of course, one of your hobbies is welding...
###Got a Question about ELECTRONICS? Check HERE First:###
I worked for an old man in Ohio that serviced Standard Electric, Edwards, and several other brands of school clocks and fire alarms. There were odds and ends in the back room he had salvaged when an old school building was torn down to keep some real antique systems in service, in cities that still used controlled loop alarm systems.