Termination in RS485 network

[sumiii]

I have a half duplex RS485 communication between 1 slave and master but was wondering what is the effect of the termination resistor?

My length of CAT5 cable is 30 meter and the baud rate is 19200.

A cable is not seen as a transmission line if the signal rise time is more than four times the propagation delay of the cable

I have calculated the propagation time, by using two methods

Method 1 :
Assumption "Wires have an approximate propagation delay of 1 ns for every 6 inches (15 cm) of length"

Cable Length Propagation Delay
15cm -> (1/15) ns
1m -> [(100/15) ns
30m -> [(100/15)*30] ns
30m -> 200ns
so my propagation delay is 200ns.

Method 2 :

Propagation velocity = 0.66 x speed of light = 197863022.28 m/s.
For a 30m cable the propagation delay should be 455 ns assuming it damps out in 3 round trip.

Please tell me which method is correct of calculating Propagation velocity?

My baud rate is 9600,so bit width will be 104.17 us .

In both cases propagation time is short relative to the data bit duration, the effect on the signal quality is minimized, and I don't need any termination.
Is it right?

 

[dick_freebird]

You can "consider" it all you like, but consider also that
the risetime will -not- be "electrically long" and there will
be edge over/undershoot, or staircasing, which the
receiver may respond to with "chatter" if the far-end
edge (singular) is not clean. This could give you clock
or data timing uncertainty or periods right after edge
where the data can be clocked in, in the wrong state.

So terminate it. If you don't like the power draw, you
can use a dynamic termination (large C in series with
R). Then the 1/2*C*Vdiff^2*f shuttle charge is your
power loss, not a DC Vdiff/R. C just has to be big
enough to ride out a few times time-of-flight.

 

[sumiii]

Thanks man,


Propagation velocity = 0.66 x speed of light = 197863022.28 m/s.
For a 30m cable the propagation delay should be 152 ns but I took (155*3)455 ns "assuming it damps out in 3 round trip."

So I think 455ns is enough.
What you say?

 

[FvM]

For a 30m cable the propagation delay should be 152 ns but I took (155*3)455 ns "assuming it damps out in 3 round trip."

I understand that you are calculating a maximum baud rate for the cable system. but propagation delay is clear technical term. It doesn't depend on reflections and will be 150 ns in any case.

With usual 120 ohm termination, there won't be considerable reflections.

 

[sumiii]

Ok. I got your Point, Let me give you some detail of problem that I am facing,

I am using a RS-485 Network in my circuit with ST485 ECDR IC.
My Mater device is working fine but my Slave IC stop working,
Circuit diagram : https://obrazki.elektroda.pl/9951537300_1363460102.jpg

I think I should remove Pull-up and Pull down resistor and should add a transil.


Please give me your valuable suggestion

Regards,

 

[FvM]

PU/PD resistors are usually required in RS485 networks to set a defined idle state when both transmitters are disabled. They are specified in many RS-485 based IO-standards, e.g. PROFIBUS.

I don't know what "stopped working" exactly means.

I have operated RS-485 multi-drop networks at 19200 Baud with > 1000 m cable length, by the way.

 

[sumiii]

PU/PD resistors are usually required in RS485 networks to set a defined idle state when both transmitters are disabled. They are specified in many RS-485 based IO-standards, e.g. PROFIBUS.

Yes I know the purpose of PU/PD resistor, but what I was asking is that do I need to add PU/PD resistor on Slave node also? (as Mention in my circuit diagram)

I don't know what "stopped working" exactly means.

Stopped means that during working slave stop to reply the master,then when I replace the IC it start to replay again.
This behaviour is very random,sometime it occur with in a hour and sometime it occur after months.

I have operated RS-485 multi-drop networks at 19200 Baud with > 1000 m cable length, by the way.

That's really good, then you have a very good idea about RS-485 networks

I have some concerns in my circuits
1. Transil Protection missing : I think I should use CDS0T23-SM712 TVS diode
2. Biasing resistor needs to be removed from slave side

3. The RS-485 specification recommends connecting a 100-W resistor of at least 0.5 W in series between each node’s signal ground and the network’s ground wire, as attached Figure shows. This way, if the ground potentials of two nodes vary, the resistors limit the
current in the ground wire.
https://obrazki.elektroda.pl/index.php

4- May be we can remove the termination resistor from circuit. I decided to remove the termination resistor after doing this calcultaion
Propagation velocity = 0.66 x speed of light = 197863022.28 m/s.
For a 30m cable the propagation delay should be 455 ns assuming it damps out in 3 round trip.
Baud rate is 9600,so bit width will be 104.17 us .
Since propagation time is short relative to the data bit duration, the effect on the signal quality is minimized, and I don't need any termination.

Kindly do comment on my calculation and assumption, and give your valuable advice
Thanks in Advance..

 

[FvM]

but what I was asking is that do I need to add PU/PD resistor on Slave node also?

No, you have been asking if you should remove the resistors, without referring to slave.
Having the resistors on one side only would be sufficient as long as the cable is connected. I don't see a reason to remove resistors on one side.

Stopped means that during working slave stop to reply the master,then when I replace the IC it start to replay again.

Saying the tranceivers was obviously damaged is much clearer, I think.

Massive overvoltage is the most likely reason. Without knowing the application details, I don't feel able to guess about the exact reason, but I would generally add some protection. TVS diodes are generally a good idea. I don't see a specific purpose of ground series resistors. They definitely don't reduce overvoltages at the bus terminals. Some designs have however current limiting resistors in the bus lines.

I agree, that you don't necessarily need termination for 30 m. Many people would install RS-485 termination though. But anyway, termination or not isn't related to damage by overvoltage.

 

[sumiii]

Thanks man, you are very helpful.

No, you have been asking if you should remove the resistors, without referring to slave.

Sorry my mistake

Having the resistors on one side only would be sufficient as long as the cable is connected.

I don't see a reason to remove resistors on one side.

Your both statements are contradicting

I don't see a specific purpose of ground series resistors. They definitely don't reduce over voltages at the bus terminals.

Actually when first time the problem occur in my slave device, I checked on oscilloscope and found that there was some noise from ground, so I separate the ground of Master and slave device and then it start working fine.
After some months this problem again occur and this time failure ratio was much greater, so I thought without a common ground, the circuit may work, but the energy from the imbalance has to go somewhere and may dissipate as electromagnetic radiation.
Thats why I thought to add a ground series resistor

 

[FvM]

Cutting the ground connection (or insertig a series resistor) doesn't remove the common mode voltage. If it increases above the common mode range of the receiver, you'll read wrong data. Up to a certain interference level, a low impedance ground connection between both circuits can help. To widthstand strong interferences, optical isolated RS-485 is required.

Your both statements are contradicting

No. It's usually wanted to put the receivers in a defined state also without bus connection.

 

[sumiii]

If it increases above the common mode range of the receiver, you'll read wrong data.

Is it possible that if common mode range of the receiver increases then transceiver become faulty?

Let me conclude that I need to improve my circuit by doing these modification,

1- Low impedance ground connection needed through a short fat wire instead of a longer thinner wire.
2- TVS diode needs to be add.
3- PU/PD resistor needs to be remove from slave side
4- We don’t necessarily need termination resistor for 30m cable, so I can remove it

Please tell if any other improvement required.

Thanks for your support. I really appreciate it.

Regards,

 

[FvM]

The datasheet tells you that maximum rated bus voltage is higher than maximum common mode voltage, but not much.

I noticed, that you are talking about "a transil diode". You surely should have two of it, limiting the voltage at each bus line against ground.

 

[sumiii]

I noticed, that you are talking about "a transil diode". You surely should have two of it, limiting the voltage at each bus line against ground.

Yes right, I had this thing in mind.

Thanks for your help.
It was a very nice talking to you.

Regards,