Communication over 5m distance - crosstalk etc

[grevillea]

I am building a project that requires communication of several signals and power over 5m. My current plan is transmit 24V power and ground across 9-way shielded cable (Jaycar WB1578), perhaps with the ground on the shield.

The project involves stepper motor control, and the communication circuit is between a USB controller (SD4DP from stepperboard.com) and logic circuitry that drives LEDs etc and also drives 2M542 stepper motor driver. This is the context for the communication.

The 24V is down-converted to 5V to power the SD4DP (at the SD4DP end) with 1000uF electro cap and 100nF ceramic cap to smooth the power and dampen noise on the line.

The SD4DP accepts input switch signals as CMOS inputs with a 1k pullup resistor. I plan to drive these signals from the other end with 7406TTL inverter buffer open collector. The SD4DP output signals (motor Step and Direction) are driven by ULN8203 darlington array, so they are open collector also; I plan to receive these signals at the other end with a 1k pullup and logic gate input (such as CMOS 74HC logic).

Signal timing is measured in microseconds. A typical Step pulse duration is 10us or more.

I am concerned that there may be problems with cross talk between the parallel wires in the communication cable; this could be crosstalk between the power line and signal line or between one signal line and another. The potential negative impacts of crosstalk include false step/direction signals leading to misstep of the motor.

My questions are:

1. Should I be worried about the potential for crosstalk in this situation?
2. Should I consider filtering the signal at the sender or receiver to suppress any crosstalk?

Thanks
Grevillea

 

[JimB]

My questions are:

1. Should I be worried about the potential for crosstalk in this situation? Yes
2. Should I consider filtering the signal at the sender or receiver to suppress any crosstalk? No

I suggest that you use a cable with twisted pairs such as the Jaycar WB2030 CAT6 cable.
Each of your signal lines should be twisted with a return line (0v, earth, ground, call it what you will).

Using the screen braid as the single return for multiple "fast" signals is potential trouble.

JimB

 

[grevillea]

JimB: Thanks for your suggestion. It leads to further questions.

The input and output signals from the SD4DP use a shared ground. For twisted pair, I believe that I would need a separate ground corresponding to each signal and that the separate grounds can only be connected to the common ground at one end.

I am guessing that some sort of line driver will be required to send signals each with their own ground, or some sort of isolating receiver to receive the signals with their ground isolated. I am using 74 series logic, so can you suggest appropriate driver or receiver?

 

[JimB]

I had initially thought of just using multiple return lines, effectively all in parallel, it is a bit of a bodge but better than your initial proposal.

If you want a good solution, have a look at Maxim MAX485 ICs.
They are RS485 transceivers and will drive into and receive from a twisted pair cable no problem.

JimB

 

[Ian Rogers]

I use a coaxial (satellite type) .. I use a 12 volt supply then transmit uni directional data back and forth on the power (center core),, It works for me up to 30 - 40 metres
I use about 19200 baud.

 

[crutschow]

As JimB mentioned, to minimize noise (both crosstalk and common-mode type) I also suggest using a differential driver and receiver with twisted pair wire such as those used for RS422/485 data transmission. Maxim, for example, sells some in a quad package so you can transmit 4 separate signals with one chip (requiring 4 twisted-pair wires of course).

 

[grevillea]

Thanks for your helpful replies.

After investigating RS485 transmitters and receivers, I agree that they would make a good solution; and it is useful to have learned about them. However, I have decided to try a different approach in this particular project - merge the two boxes into one and eliminate the communication problem entirely. If this works, I will be relying on up to 5m USB connection between the PC and the USB controller, but then there is no need for so much twisted pair communication from USB controller to other parts of the system.

Regards
Grevillea