RS485 for 30m cable

[A2]

I am using a board with wired communication on RS485. I am using SN65HVD72, it specifies cable length to be 2000m.

The board on other side which is in field, i.e 30m away has 4 wire connection, i.e Vcc, Gnd, A, B. (where A, B are RS485 signals).


1. I am planning to use four core cable.
2. My field board has 80mA current consumption. Vcc is +5V. This +5V goes to 3.3V regulator which is used for board finally.


Question:
1. Will +5V be able to show at filed board which is 30m away from centeral board. Also keeping in mind 80mA cuurent consumption.

2. Will RS485 will be done fine?

3. What type of four core cable I should select

[Artraze]

1) You won't get 5V at the board because the cable will have some resistance and drop some voltage.  So if you put 5V in, you might get 4.9V out, but that depends on the cable.  However, if all you care about is getting >3.6V to drive an LDO, that's no problem:  Your cable just needs to be less than (5V - 3.6V) / 80mA / 30m = 0.58?/m.  Just about anything meets that, including cheap fine gauge ethernet wire.  Do make sure you have a fair amount of capacitance on the field board, however, as that long wire isn't going to be able to respond very quickly to changes in loading.

2) It will depend on you signaling rate, of course, but supposing you are under ~100kbps you won't likely have any issues.  TI has a design guide, but if it's just point-to-point like that RS485 is pretty simple.

3) Unshielded twisted pair is the standard choice here, with one pair for power and one for data.  If it was me for a hobby project I'd probably use Cat 5 ethernet wire as it's cheap and would work fine.  If you're in an industrial environment, you'd probably want an overall shield to help control noise.  Size isn't likely to be an issue, but I'd recommend 24AWG / 0.2mm^2, give or take a little.

[rs20]

1) You won't get 5V at the board because the cable will have some resistance and drop some voltage.  So if you put 5V in, you might get 4.9V out, but that depends on the cable.  However, if all you care about is getting >3.6V to drive an LDO, that's no problem:  Your cable just needs to be less than (5V - 3.6V) / 80mA / 30m = 0.58?/m.  Just about anything meets that, including cheap fine gauge ethernet wire.  Do make sure you have a fair amount of capacitance on the field board, however, as that long wire isn't going to be able to respond very quickly to changes in loading.

2) It will depend on you signaling rate, of course, but supposing you are under ~100kbps you won't likely have any issues.  TI has a design guide, but if it's just point-to-point like that RS485 is pretty simple.

3) Unshielded twisted pair is the standard choice here, with one pair for power and one for data.  If it was me for a hobby project I'd probably use Cat 5 ethernet wire as it's cheap and would work fine.  If you're in an industrial environment, you'd probably want an overall shield to help control noise.  Size isn't likely to be an issue, but I'd recommend 24AWG / 0.2mm^2, give or take a little.

+1 to all this. I did something very similar a while ago, RS485 over a (cheap 2-pair, not normal 4-pair) CAT5 cable that would have been around 30m. I only ran it at 4800 baud or something ridiculously conservative like that, just because that's all I needed (and then some). Worked without any problems, and has been working for a couple of years now even in the wind, rain and sun (I got black CAT5 cable because maybe the black pigment plastic is better at handling UV).

[bobcat]

Don't forget to use the same value termination resistors as your cable impedance! Lots of signal quality  problems can occur if you don't.

[SeanB]

If using Cat5 cable you can use the 2 unused pairs for power, and get a lower voltage drop along it. If you need a little more power at the far end use 12V and use a small buck converter to drop the voltage down and up the current. As above use a lot of decoupling on the input, and a large transient suppressor as well along with a reverse polarity protection diode is also good to have.

[SL4P]

If using Cat5 cable you can use the 2 unused pairs for power, and get a lower voltage drop along it. ...

... and clearly label the cable as NON-STANDARD wiring.
Your next IP phone, switch or modem wil thank you for it.

[rs20]

Don't forget to use the same value termination resistors as your cable impedance! Lots of signal quality  problems can occur if you don't.

At 100kbaud, this isn't an issue until the cable is significant fraction of a kilometer long. 2.4kbaud; 40 kilometers.

[A2]

I have selected this 4 core cable 0.5mm^2.
http://edgecdn.lappgroup.com/fileadmin/documents/technische_doku/datenblaetter/unitronic/DB0034302EN.pdf

1. Since loop resistance is 38.9ohm/Km. & my requirement is 50m or 100m in loop so total resistance = 3.89ohm, so drop at 80mA is .32V. Will work fine according to me?

2. Other is I didn't find characteristic
impedance in datsheet for termination in RS485 connections.?

I have only master & slave device (2 devices on Rs485)
& at 9600bps. Do I need termination resistance or I can ignore

[David_AVD]

I would be using 12V or 24V and a small switching regulator at each sensor point.  The Murata OKI-78SR are a economical solution that's good up to about 30V input.

Then you can use (stranded) CAT5E for power and data.  I do this all the time with an RJ45 pinout that minimises risk if connected to phone or Ethernet.

[GeorgeOfTheJungle]

I have selected this 4 core cable 0.5mm^2.
http://edgecdn.lappgroup.com/fileadmin/documents/technische_doku/datenblaetter/unitronic/DB0034302EN.pdf

1. Since loop resistance is 38.9ohm/Km. & my requirement is 50m or 100m in loop so total resistance = 3.89ohm, so drop at 80mA is .32V. Will work fine according to me?

Yes, at these speeds almost any cable would do really, but a CAT3 or a CAT5 cable would do as well and they're pretty cheap and common.

2. Other is I didn't find characteristic
impedance in datsheet for termination in RS485 connections.?

I have only master & slave device (2 devices on Rs485)
& at 9600bps. Do I need termination resistance or I can ignore

Plug the scope probe onto the bus line and do it empirically using a couple of trim pots as terminators. Feed a signal in a loop and adjust until you see a nice clean wave then use a fixed resistor of the closest standard value you can find.

I did something like that a long time ago in a bus a bit longer than that with a signal @230.4 Kbit/s (UART) and all went perfectly well.

To make the bus I used a long flat cable in a single piece with IDC DB-9 connectors crimped on to simplify the construction as there was a device every 1-2 meters or so.