What is the best data transmission protocol for high noise environment?

[Firejo]

I’d be much more concerned about inducted voltage than noise. I’d be looking for a power line modem that’s designed to deal with the amount of voltages you’re dealing with. Having said that, I’m not sure anybody is making that sort of technology anymore.

 

[strantor]

I am investigating the solutions suggested thus far (thanks for the input).

But if I can't confirm a suitable comms bus, do you think I'd have any trouble with 4-20mA signals through this cable?

All I need is the value of (8) 4-20mA sensors. I wanted to cut a few conductors out of the cable bill by using remote I/O. I've been assuming that I won't have any problems with 4-20mA, but I'll give you all a shot at that theory.

if 4-20mA is all good, then I suppose I could multiplex them to save conductors.

 

[HJTRBO]

Yaskawa do a Matrix drive. That will help with the noise generated from the power conductors

 

[Firejo]

I can't imagine 4-20mA working in that environment.

 

[strantor]

I can't imagine 4-20mA working in that environment.

Why is that? have you ever had 4-20mA not work because of noise? I thought it was immune to induced noise. I don't know about capacitive coupling though.

 

[keithkyll]

Communication should be optically isolated RS422.
At the 4-20 mA end, use a Click PLC and a couple of 4-20 mA modules to process your sensors. Send data to the RS232 port.
Add an Isolated RS232/422/485 converter.
Use another one at the AD Productivity side.
You could also use the repeaters from B&B. Use serial rated cable.
This will allow you to have a one pair (RS485) or two pair (RS422) run that's optically isolated at each end.

 

[jstolaruk]

I have a tool at the end of a 200m cable. Through this cable I am powering a 50kW/3,000V and a 1kW/480V motor, both with VFDs. So there is going to be a lot of noise in that cable. I need to communicate with the tool through the same cable. The comms bus conductors are 2 shielded twisted pair (3 pairs). Is there any industrial comms over copper that have a chance of communicating in the midst of all that EMI? I really don't want to use fiber optics.

A recipe for trouble. Regardless of the length, standard practice has always been to separate comms and low voltage from any type of power, especially VFD runs. I wouldn't be able to get approval from any of my customers.

 

[TomekPLC]

Did you think about wireless?
http://www.bannerengineering.com/en-US/wireless/surecross_web_prodnetwork

 

[BryanG]

This is one of those occasions where everyone will say 'No' and the only real way to know is to try it. There are too many variables for a definitive answer.

Attached shows the Eaton/Moeller smart relay specs for communication, it is based on CAN. It can do up to 1000metres in an industrial environment, so maybe you will be OK in a noisy 200metres. You can slow the comms down to 10K if the noise gives problems. It is cheap enough that you can try it and see what happens. The analog inputs are only 10 bit (0-1023 decimal), but you get 4 of them.

 

[Firejo]

Why is that? have you ever had 4-20mA not work because of noise? I thought it was immune to induced noise. I don't know about capacitive coupling though.

It’s not the noise. The likelihood that the high voltage lines will induce some current onto the communications lines is very high and when working with 4-20mA it doesn’t take much induction to screw up the signal.

 

[GTUnit]

It’s not the noise. The likelihood that the high voltage lines will induce some current onto the communications lines is very high and when working with 4-20mA it doesn’t take much induction to screw up the signal.

This is a very good point assuming the neighboring voltages are high enough.

 

[GTUnit]

Any RS485 based protocol would probably be best for your application.
Differential signals used are highly immune to noise.
http://cds.linear.com/docs/en/product-selector-card/2PB_RS485fd.pdf
Spec says RS485 transceivers can have isolation up to 2500Vrms.
Profibus and Modbus are both based on RS485.
I know a typical Profibus network requires a particular wire with correct impedance.
Modbus is more flexible.

 

[Peter Laing]

Any RS485 based protocol would probably be best for your application.
Differential signals used are highly immune to noise.
http://cds.linear.com/docs/en/product-selector-card/2PB_RS485fd.pdf
Spec says RS485 transceivers can have isolation up to 2500Vrms.
Profibus and Modbus are both based on RS485.
I know a typical Profibus network requires a particular wire with correct impedance.
Modbus is more flexible.

Yes, however, wire impedance (on Modbus or Profibus) is really only important at high baud rates and long lengths. At a relatively low baud rate (19.2k or lower) any shielded/twisted pair cable should be fine.

Here's a nice application note from TI RS485 that nicely covers the technical details:
http://www.ti.com/lit/an/snla049b/snla049b.pdf

I suggest Profibus as your best bet to ensure compliance with RS485 standards, in particular "Termination Methods" (Figure 5 in above link) and "Common Mode Voltage" (Figure 8). Basically, you'll want "Failsafe bias termination" and a Data Ground (C, C') potential connection between devices. This is usually the cable shield, but in a high noise environment I would use a dedicated conductor. Cable shields should be bonded at both ends.

Here is a link to the official Profibus cabling guide. Sift through the BS and check out the important bits - section 2.1 (termination), figure 8 (pin assignment), section 2.6.5 (bonding). http://www.sci-eng.mmu.ac.uk/ascent...deline_Cabling_Assembling_8022_v106_May06.pdf

Good luck!

 

[V0N_hydro]

the comms cables are shielded. there shouldn't be any induced voltage or noise in those pairs?

 

[JesperMP]

I suggest Dupline.
"extremely noise immune", works up to 10 km.