Data exchange via 10m unshielded multiconductor cable

[Vineyards]

I think I remember your Project. It all depends on the amount of the noise that 10 meter cable will pick up from the environment where it is run. If it is close to the engine, it will probably pick up all the noise from the high voltage Ruhmkorff coil and the cables hooked up to it. This might cut off all data transfer at any baud rate. If that is the problem heavily shielded cables become a necessity. In fact all the cables between the coil and the spark plugs are very heavily shielded but in time, they may degrade and begin to leak.

Other than that here are the considerations for general applications:
1-Every cable has a capacitance per meter value which runs in the picofarads. This can be imagined as having a small value capacitor shorting the signal cable. Up to a certain length it will not do much harm then it begins to limit data transfer rate and gradually makes it impossible to get any meaningful amount of data through. It basically acts like a simple LP filter. As the capacitance goes up permissible signal frequency declines.
2-Twisted pair cables are hooked up to differential inputs with very high common mode input rejection characteristics. Since the two cables have a braided structure they are virtually exposed to the same amount of noise which can easily be filtered by a differential input. For single ended systems, their contribution to noise filtering is only as good as the common mode rejection characteristic of the input op amp which is inadequate in many cases.
3- There are three kinds of galvanic isolation and their main uses are to prevent ground loops and to protect the circuit from transients. These are inductive, capacitive and optical. They might be useful to you as we do have transients in open air/vehicle electronics related projects.

[kreutz]

Optocouplers for noise is BS - optocouplers are used for galvanic isolation, not for noise immunity

Sorry, the above statement is misleading!!

Optocouplers, depending on the way they are used, allow avoiding ground loops and, used on a current loop (4-20mA or 10-50mA) circuit, offer the capability of transmitting data over very long distances (> 1 mile), the galvanic isolation is a plus but not the only advantage.

Anyways, why not to use a wireless link for this application (10 m)?.  Serial over Bluetooth is a good candidate and the transceivers like HC-05 or HC-06 are really cheap. Of course, it will require an additional micro controller (any cheap arduino will work here) and the coil drivers will have to be taken out of the controller and placed near the controlled relays.

Is there any reason why using a grounded flexible metal conduit (or rigid copper tubing) and running the multi-wire cable inside won't work as shielding?

[karri]

Thank you all, I am going to test RS485 via non twisted vires. I tested RS232 before and it worked, but I was not sure, if it is good solution.

[Gallymimu]

Thank you all, I am going to test RS485 via non twisted vires. I tested RS232 before and it worked, but I was not sure, if it is good solution.

RS232 will be "OK" RS485 will be better. 

RS485 can achieve higher data rates and better noise immunity due to being differential (though you lose some of that benefit without twisted pair, it should still help some)

[Resonance]

The problem with noise is that once it's in the signal path, there's not a lot you can do about it upsetting whatever receiver you are using. The previous suggestions will help you alleviate the noise, but unless you manage to eliminate it completely you could be fighting a never ending problem.
It's therefore imperative that you make the communications link recoverable from those errors because once installed in the field you may not realise it, but you could be right on the edge of the system immunity.
So make sure that you have some form of redundancy and error correction in the data thats sent - or as suggested before, use something like CAN which has this built in. I suspect that CAN will be adding too much complexity though.

[vern]

I would go with RS422, no need for optocouplers.
You have already tried RS232 and it worked, RS422 is improving the signal to noise ratio a lot and you are on the secure side.
I would use optocouplers only if you have a large voltage drop on your GND connection, for example if you power the remote unit from your main unit and the current flows over the GND wire.
When picking a transceiver for the RS422 I would choose one with reduced bandwidth like the MAX3070 (3.3V) or the  MAX3080 (5V).