Transmitting low bandwith data through long, non-ideal cables

[Renaud]

HI everyone,

I have a small challenge: I have a long multicore cable (assuming shielded) that goes through a building (say 100m).
That mutlicore only carries some 110VDC voltages and low currents (for buttons, switches, indicator lights).

Now I would need to send some data from a sensor through some of the unused spare cores of this cable.
The bandwidth can be very low but I need to reliably get the data without risking introducing big issues in terms of EMI.

Since the cores are not twisted, RS485 may not be suitable so I was wondering if there was any existing transceivers for this type of low bandwidth data transmission across non-ideal cables.
I was thinking about some very low frequency frequency modulation carrier could work but I'm not even sure anything like this is even workable.
The sensor would be queried every few seconds and just needs to return a measurement.

Surely I'm not the first facing this problem, I'd rather not have to pull cables

Thanks for any pointers!

[pqass]

20mA current loop is ideal for sensors over long distances in harsh environments.
Only requires 1 pair and if the sensor is low power, can even power it.

https://www.google.ca/search?q=20ma+current+loop

There is a whole industry around the sensors and controllers.  Although it's primarily an analog system, a low bandwidth digital protocol has been developed on top it; Hart protocol. 

My only direct experience has been with designing a home automation sensor network for myself so I can't direct you to commercial solutions.  I don't know if you are, like myself, designing both endpoints or wishing just to purchase off-the-shelf.

[KrudyZ]

100 m isn't really all that long. RS485 can easily span that even at fairly high data rates.
You can get slew rate controlled drivers if you are concerned about interference.

[mariush]

I imagine ethernet over copper pair (vdsl) converters would work just fine, but they'd be expensive...
example :
https://www.amazon.com/dp/B01BOD8C9W/
https://www.amazon.com/PLANET-VC-231-Ethernet-VDSL2-Converter/dp/B009WWHNWU

100 meters is the maximum length of ethernet ... you have 8 conductors. At 100 mbps or less, only 4 conductors are used, the others could be used for power over ethernet or send power over them (ex just send 48v DC or as much as you're comfortable with through the 2 unused pairs, one pair for +v, one for ground) 
You could have one microcontroller with ethernet get buttons state and set leds on/off and get sensor data and push it over the ethernet cable at 10 mbps or 100 mbps and at the other end your device receives it and does stuff with it.
 

[NiHaoMike]

I imagine ethernet over copper pair (vdsl) converters would work just fine, but they'd be expensive...

Or hack a cheap pair of Homeplug adapters, but that sounds like overkill for this application. I would say go with CAN, it's specifically designed to be resistant to noise.

[Renaud]

20mA current loop is ideal for sensors over long distances in harsh environments.
Only requires 1 pair and if the sensor is low power, can even power it.

That's very interesting and would be a viable alternative to RS485 considering the less than ideal cable I have to use.
Thanks for the suggestion!

I imagine ethernet over copper pair (vdsl) converters would work just fine, but they'd be expensive...

In other types of applications it may be an alternative, but what I have is just straight cores bundled together, no twisting. And I would assume these types of high-speed transmissions would generate a lot of RF that would surely find its way into the other cores.
Good to know this exists though!

100 m isn't really all that long. RS485 can easily span that even at fairly high data rates.
You can get slew rate controlled drivers if you are concerned about interference.

We'll test RS485 but I'm not sure it's going to work well without at least using a twisted pair for noise immunity, that's why I was looking at alternatives that I may not have considered.

I would say go with CAN, it's specifically designed to be resistant to noise.

It is certainly worth trying. Looks like CAN should work on the type of cabling we use. We'll have to check for noise immunity.

Thanks a lot for all the answers guys!

[Medved]

20mA current loop is ideal for sensors over long distances in harsh environments.
Only requires 1 pair and if the sensor is low power, can even power it.

https://www.google.ca/search?q=20ma+current+loop

There is a whole industry around the sensors and controllers.  Although it's primarily an analog system, a low bandwidth digital protocol has been developed on top it; Hart protocol. 

My only direct experience has been with designing a home automation sensor network for myself so I can't direct you to commercial solutions.  I don't know if you are, like myself, designing both endpoints or wishing just to purchase off-the-shelf.

The current loop is quite strong industry standard, so many off-the-shelf component should be available. And contrary to the various digital systems, will have less interfacing compatibility problems...

[MIS42N]

I think as long as you have a differential driver and receiver, and the cables you use are paired, you'll be OK. The twisting of pairs helps prevent high frequency interference but you don't need that. Could try UA9638/9639 driver/receiver and see how they perform. Might need to do some calculation to work out the nominal impedance to get the terminating resistors right. I've used these driving Ethernet cable and powering the driver through unused pairs. Not the harsh environment you are looking at, but IMHO worth a try.

[2N3055]

Yes go with current loop. You can get optically isolated converters that terminate into RS232, so it will be transparent to you, and safe if something wrong happens in a cable.

[tggzzz]

You would be wise to consider failure modes including (but not limited to)...

Any equipment on each end is connected to a different phase.

There is a difference in the ground/earth/line voltages at each end, especially if there is a line fault or lightning strike.

Somebody puts a shovel through the cable, and shorts conductors.

Insulation resistance between conductors when water has crept into the cable.

[2N3055]

You would be wise to consider failure modes including (but not limited to)...

Any equipment on each end is connected to a different phase.

There is a difference in the ground/earth/line voltages at each end, especially if there is a line fault or lightning strike.

Somebody puts a shovel through the cable, and shorts conductors.

Insulation resistance between conductors when water has crept into the cable.

Absolutely good thinking, hence my recommendation to use commercial optically isolated converters....

[kripton2035]

may be you could also get rid of the cable and use a lora network ?