power and RS-232 signal over distance

[monkeysolder]

Hello again -

You guys were so helpful on the last question so I have another problem that I am curious about.

Short Description:
I am running rs-232 signal over a 50 ft length of cable (28AWG ribbon, shielded). I am also running power (12V, ~1A) and ground, each over 3 of the 28AWG wires. The data and power share ground. Now I know why these are considered bad ideas. So here I am.
I have noticed that there is a +2V difference between the on the ground line and a -0.5V difference on the 12V line. I believe this ground differential is causing the RS-232 to misbehave. This is on a benchtop with cable all bundled up, so I can connect a short ground wire from the power source to the "far" end ground pin and the RS-232 functions perfectly. The 2V differential fluctuates with the current drawn from the power supply, so I believe that it is caused by that.

Does anyone know of anyway the "correct" this problem? Separate data and power grounds? I am going to check for cold solder joints presently.

Thanks for the help.

 

[monkeysolder]

Sorry that should read:
"I have noticed that there is a +2V difference between one end of the cable and the other on the ground line and a -0.5V difference on the 12V line."

The end farther away from the power supply is +2V on the Ground line and -0.5V on the Power line.

 

[itsmoked]

If I where doing what you're doing I would use full duplex RS-485 drivers NOT RS-232.

I would also us multiple grounds in the ribbon cable. 50 foot 28 AWG is a heck of a lot of resistance!

I would stick the comm on one edge of the cable and run the multiple grounds between the comm and anything else.

Also run the lowest possible baud rate to get the job done.

 

[Skogsgurra]

Right, 50 feet is actually not "distance". It is where the specifications end, yes, but if you do as smoked says and use as much area as you can afford (but don't overdo it), you will be fine.

"Cable bundled up" may result in increased inductance in the signal ground - and that may result in sensitivity to transients. If you bundle, avoid coiling up the cable.

Gunnar Englund

http://www.gke.org

 

[monkeysolder]

Alright did some more testing and found a "solution" to the problem. I can hook the ground lines to the shielding (substantial amount of mesh wire and Al foil) inside the PVC jacket on the cable. This appears to get rid of the voltage drop. Also, this will introduce more noise onto the ground path than before, but that is more acceptable than not working so...
Another thing that seems to work is adding large (6800uF) capacitors to the voltage rail on the far end.

I'll probably use both for some peace of mind. Not pretty, but it works.

Good idea, itsmoked. I orginally wanted to convert to ethernet and run CAT-5 and use power over ethernet, but that was vetoed. Doh.

Thanks for the help.

 

[IRstuff]

RS-485 systems are usually half-duplex. RS-422, which has essentially the same specs, is a full-duplex configuration.

In fact, you could potentially get 422/232 converters on both ends and that may get you out of some of the problems you're seeing. The converters usually run anywhere from about $20 to $100, depending on the make:

http://www.dcbnet.com/datasheet/srdcds.html

near the bottom.

TTFN