Modbus RS-485 Network: Stub Length Help

Join Date
Aug 2019
Location
Krong
Posts
27
Good day all. I have an RS-485 topology question. I am hoping for some guidance so that I can make a confident calculation/decision that this will work.
Here is what I understand.

1. Star networks are bad practice.
2. If you do need to use a star, keep stubs as short as possible.
3. Data transmission rates are going to suffer the longer the trunk is.
4. Data will degrade when adding longer stubs because of reflections.
5. Add terminating resistors, selection will vary depending on choice/length of cable.

I have read other forum posts on this matter and I have read a couple Texas Instruments papers.

The RS-485 Design Guide
Page 3 I do not understand the "driver's output rise time", do I need to contact the sensor manufacturer and ask them for the specific RS-485 chip within their equipment?

AN-1057 Ten Ways to Bulletproof RS-485 Interfaces

Here are some questions that I have.

1. What does "as short as possible" mean? Is this a couple inches, or a couple meters? The TI literature talks 7-38 ft
2. Does the consistency of stub length matter?
3. Is it normal to ask sensor manufacturers about the specific RS-485 chips within their devices?
4. Are stubs not more advantageous than daisy chaining (separate in and out terms on a device), because when a device fails, you can still communicate to other devices on the network?

Some context I plan on having 24 gas monitoring devices all with the same stub length of 12 inches. I plan on connecting the trunk from JB to JB on a terminal strip. From the strip, I would make the stubs through jumpered connections on said terminals. The trunk leaves on the strip as well. It is okay if we do not receive alarms from the device for 5 seconds, if the transmission rate needs to be that slow.

I initially made the design with stubs because one of the sensor manufacturers defined the connection that way. Looking at the TI design guide, they are constantly showing stubs as opposed to true daisy chains. It also seems more attractive with stubs. And I like saying stub. Stub, stub, stub, stub.

I can provide the trunk cable length if immensely relevant, I just need to do a little digging. It is roughly less than 350 feet.

Please let me know where I need to clarify things if I have described inadequately.

Thanks for your time and for your help!
Have a great weekend.
 
Last edited:
I don't have any answers, but someone once told me that "multidrop" is a more accurate phrase for RS-485/RS-422 than "daisy chain."
 
Are the stubs already connected to the sensors when you take them out of the box?

If not why wouldn't you run two pieces of cable from the junction boxes to each sensor and make the connections at the sensor?

Question 1: My experience with the topology is with GE's Genius remote I/O network which was not RS485, but still multidrop at up to 156.3 kbaud. Their recommendation for twisted pair cable was that it be compliant with RS422. The following is their guidance for stub lengths.
The maximum length of all stubs on a bus should be 100 feet or less. Within each 20% section of the actual bus length the total maximum stub length is 20 feet.

Question 2: No.

Question 3: Perhaps not the chips they use, but certainly the serial protocol they use.

Question 4: If the device comes with a terminal strip for making connections, that's the place to land the wires. Assuming both pieces of the trunk cable land on the same terminals, the device can fail and you still have trunk cable continuity.
 
Thanks for your replies.

@DRBITBOY

I'm sure that there are many different terms for trunks and stubs.
I was using terminology in the design guide I referenced above (on the first page at the bottom).
"The RS-485 standards suggests that its nodes be networked in a daisy-chain, also known as party line or bus
topology (see Figure 3-1. In this topology, the participating drivers, receivers, and transceivers connect to a main
cable trunk via short network stubs."

@Steve

The sensors themselves screw into a main module, a PCB.
The PCB has tiny little screw terminals to land the RS-485 on. Not big enough that I would like to land two wires under the same screw.
In certain areas, we are going to have multiple main modules very close together. So I was hoping for some phoenix quattro terminals, put trunk in, two stubs and trunk out. Either that or jumper smaller terminals to the same effect. With that being said, I plan on getting both main modules ordered, and having a shop fab the JB with some pre-fab stubs at very short length.
Not sure if I fully understood your first two paragraphs.

1) This is encouraging, I am hoping for a max stub length of 1.5 feet.
4) Looking for multiple connections was the first thing I looked for. Unfortunately as described above, there is not. So maybe I am not thinking about the most correct solution here, but it makes sense at this moment of time.
 

Similar Topics

Hello Automation Experts I am using "EASUN's MJT 314 Relay" in RS485 Modbus network OUT of 26 Relay 21 are...
Replies
0
Views
4,402
Hi, I have a system in which I have to put together a network of 27 PLCs with an HMI. Each PLC will have an encoder connected to it and be...
Replies
2
Views
7,949
I'm trying to control a device via MODBUS RTU and the ModbusRtuMasterV2_PcCOM in Twincat 3. I've configured a device with the right com port and...
Replies
7
Views
188
Does anyone know of an AOI using the user ASCII protocol select on the L6x controllers that will talk Modbus RTU using RS-485? Thanks, Trevor...
Replies
1
Views
129
Hello, I need to access the bits from the fire system control panel through RS-485. So, I used Read Var block to read from the panel...
Replies
0
Views
181
Back
Top Bottom