DeviceNetCable

DeviceNet is using CAN, that means there are no big differences if you want to apply the following to another CAN network like CANopen.

See also CanCabling and CanCablingChecklist? ( ProblemSolvingQuestions) .

DeviceNet uses a standard 24 VDC power supply. More than one power supply can be used to meet big loads. However you need to make sure that the common voltage is less than 5 VDC between any two points on the network. Also make sure that the voltage drop in the cable does not exceed 5 VDC between a power supply and each station.

Regarding the cable you must consider that DeviceNet uses a trunk-drop topology. The trunk cable is a thick cable and needs to be terminated at each end. Using a thin cable is not recommended in the trunk section. For a small segment, a thin cable might work but I wouldn't use it for 50 meters. In addition DeviceNet use 4 wires plus shield. The best thing to do is use cable than meets the DeviceNet specification. For more details you can go to http://www.odva.org

Sergio Diaz; Momentum Technical Consulting; sergio.diaz@momentum-inc.com

A good overview about RS485 cabling is given at http://www.maxim-ic.com/appnotes.cfm/appnote_number/763. Its comparable with a CAN cabling. The first CAN networks available were build using rs485 transceivers.

oe@port.de

http://www.can-cia.org/downloads/?39 gives an overview about used CAN connectors.

This schematic was created by Steve Corrigan s-corrigan1@ti.com

Comments given by John Dammeyer johnd@autoartisans.com : You can see in Steve's drawing that the ground is common between all nodes. Under ideal conditions CAN will function with isolated grounds but, if for some reason you have a common mode voltage difference between node 1 and 2 that exceeds the device specifications, the current will find it's way through the virtual ground loop. the result will be heat. CAN is 3 wire: CAN_H?, CAN_L? and GROUND. That's unless you isolate with transformers or optical. Ignore the ground or let it find its own way at your peril.

"Corrigan, Steve" <s-corrigan1@ti.com>: Without boring you to death with the math, the max stub length = tr/(10*5ns/m) is a good rule of thumb.

"Corrigan, Steve" <s-corrigan1@ti.com>:

A very common problem in any application is that the local ground is NOT zero volts. In buildings, noise from sources such as switching power supplies and/or florescent lighting load cause third harmonic currents to flow in the ground plane. The resistance in this path cause the ground voltage to rise, sometimes significantly. There are very many computer latchups that old Bill is blamed for that are a symptom of this condition.

The ground voltage between two nodes in a CAN application often differ by several volts. This is why the transceivers are designed to operate with this difference between the ground voltages. When one is signaling, its CANH may be 3.5V, but its ground voltage may be 10V higher than the other nodes. So if you put a meter on the bus at this node and run a ground wire outside to a ground rod stuck solidly into the ground, you would measure 13.5V on CANH, but if you measure between CANH and the node's ground plane, you would only see 3.5V

The other transceivers on the bus would latch up if they were not designed to operate over a wide range of voltage. They simply "float" on the rising ground voltage and operate as if nothing is wrong.