We are running some led billboard displays using a multi-drop 2-wire RS485 network. The cable length is 50 feet; the baud rate is 2400 baud.
Are RS485 terminators really needed for this? I know they are needed for 'long' cables and 'fast' baud rates, but does 50 feet at 2400 baud qualify for this?
And yes the reason I am asking is there are some problems with some of the displays. (If it was easy to just plug in some terminators at an existing customer site with the problem I would; but it's not easy.)
Also, it may be that the cable being used it not twisted pair. Is that important for this setup?
It could work without teminating resistors. What is essential is the common more range of some
7 V or so. So you should always have the GND as reference together with the differential signal. If there is the chance to exceed the common more range, couplers should be inserted and held at the GND potential. These 7V may not be much and are quickly exceeded, and be it only at burst noise. It happened in one building with an inhouse-net, that communication failed when the lift was running even though the RS485 was held at EARTH potential. Yes, the lift made the EARTH jump by more than the common more range. (...)
I suggest you start with this article written for Circuit Cellar by Jan Axelson...
formatting link
Bias the signal leads per Jan's recommendations. Also note Rene's comment on common mode voltage and the fact that three wires (not two) are required (even if one of them is the DC ground wire).
You'll probably have to provide terminations on one of the signs that is giving you problems, just to see if it fixes the problem.
I designed the electronics hardware and wrote the software for the full-size and mini message boards shown here...
formatting link
The balanced data bus is RS-485 at 115,200 bps using randomly-dressed non-twisted 18-gauge wire. There have been no problems. I don't recommend such haphazard wiring without extensive testing.
Just to give an idea of how noisy some environments can be: At a shopping center with cable runs up to 500m, we have picked up noise spikes over over 200V that lasted in the ms range. This was using good quality twisted pair cable with a defined impedance. This was measured directly over a 120 ohm termination resistor.
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
I read in sci.electronics.design that Anton Erasmus wrote (in ) about 'Is termination needed for short and slow RS485?', on Sun, 19 Sep 2004:
It can happen, but you have to be extremely careful how you do the measurement. 'Balance' in balanced circuits means 'balanced impedances'; it has nothing to do with signal voltages. The line remains balanced (or not) when no signal is present.
So, you need to make sure that the sending-end impedances from each conductor to ground are closely equal and preferably low. At the receiving end, the impedances to ground must also be closely equal, but in the interests of signal transfer, especially at low frequencies, are preferably not equally low. This is because the four impedances form a balanced bridge and the error voltage due to the impedances not being exactly equal, which is differential and becomes an inextricable part of the signal if it's in-band, is smaller if the impedances at the sending and receiving end are very different.
'Closely equal' can mean really close. To get 80 dB common-mode rejection, the impedances have to match within 0.01%.
At higher frequencies, of course, the *differential* impedances at each end may well need to match the line impedance.
Measuring such 'highly balanced' circuits requires extreme care, otherwise the measuring instruments upset the balance and produce seriously pessimistic results.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
It is better to use 3-wire for differential comms so that the receivers have at least some idea of the common mode voltage. Also a low-pass filter on the input of each receiver can be useful in avoiding difficulties with line-ringings, (from a line that is being driven too hard), or stray interference.
2400 bits/sec is 417uS per bit.
2+2k*2.2nF is an 8.8uS RC time constant, which will get to 99.9% of where it is supposed to go in about 44uS, or about 10% of each bit-width.
The 100R resistor connecting each receiver 0v to the line 0V is there to limit the current in case of earth loops. In fact if there is any voltage across any 100R (when not transmitting) then you know that you probably have an earth loop problem.
At least you should have somewhere a pull-up and a pull-down resistor to drive the bus into the idle mark ("1") state when there is no active transmitters.
This is essential.
This might generate more problems than it solves.
If the equipment are ground referenced to the electric system neutral (directly or through PE connected to neutral), the reason for the equipment ground potential differences is the different voltage drops in the mains system neutral wiring carrying the whole mains neutral current. The drops can be quite with large single phase loads or even in a three phase system there can be quite considerable 3rd harmonic currents at 150 or 180 Hz if there are a lot of switching mode power supplies in each phase.
Connecting the signal ground in the signal cable between two equipment that also are referenced to the mains neutral will effectively connect two points in the mains neutral together. If these two points in the mains neutral network have a different potential (due to different voltage drops), some of the neutral wire load current will flow through the signal ground in the signal cable between the two neutral connections. This current could be several amperes and in the worst case even burn some PCB ground tracks. This current could be mainly at the fundamental 50/60 Hz frequency, but at least also in three phase electric distribution systems at 150/180 Hz and in addition all kinds of high frequency interference from switchers etc.
If the distance between the two equipment is the same measured on the serial cable and along the electric network, the ratio between the current flowing in the neutral wire to the current in the signal ground is directly proportional to the cross section area ratio between the mains neutral and signal ground.
A thin signal ground between the equipment does not carry a lot of current, so it does not reduce a lot the ground potential difference (and hence common mode voltage) between the equipment. If a thick serial cable shield (with cross section comparable to the mains neutral cross section area) is used, about half of the mains neutral current will flow through this shield and drop the voltage difference between the signal grounds to one half of the original value. However, if such system is used, it is essential, that the shield is connected directly to the equipment chassis and then to the mains PE and not let this large current flow through the PCB of the equipment.
If the worst case ground potential difference is non-zero but well within the common mode range, it should be sufficient to run the connection without a separate signal ground and rely on the mains grounding.
Galvanic isolation from the local mains grounds is required, if such large variations exists, since a thin signal ground would have no effect on the voltage difference in the mains neutral wiring.
IMO, a separate signal ground in the RS-422/485 cable should only be used, if all the equipment are floating relative to the mains grounding. In this situation, it simply is supplying the bias currents for the receiver input transistors.
With floating system, the signal ground wire can be eliminated, if the transistor bias current is supplied some other way.
On RS-422 connections, the "fail-safe" termination (a voltage divider between local Vcc, the signal Rx wires and the local signal ground) should supply the bias current. On multidrop RS-485 the A and B signal lines should be terminated at both ends of the bus as usually with resistors, however, high resistance (1-10 kohm) pull-up/down resistors from _local_ Vcc and ground to the signal line are required to supply the bias current for each transceiver. The internal power supplies within each station will be floated to approximately the same potential without any external ground connections.
You're perfectly right Paul. I was somewhat unclearly indicating that the GND should be designed into the system, into the connector and into the cable. Current flow through it should be avoided. Should at some point be decided that isolators were required, the GND could be used. The 100 Ohms in series with the GND could als be employed to detect current flow. Depending on the location, measuring GND differences are unpractical. I actually run +5V together with the GND too. This was isolators can be fully powered through the cable.
That 100 ohms could function as a fuse. Grounds can have massive differences, think start-up currents into elevator motors and the ilk. The fault there is the connection to 0v, assuming that is connected to local ground. Any long wire can have large voltages inductively induced by nearby equipment.
--
Chuck F (cbfalconer@yahoo.com) (cbfalconer@worldnet.att.net)
Available for consulting/temporary embedded and systems.
USE worldnet address!
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.