I^m confused about voltage drop across nodes. In one place, the Specifications says a maximum voltage drop for a network can be 10 volts, but in another it can be interpreted as only allowing 5 volts. Nick Jones^ presentation on Implementing a DeviceNet Network indicates power should not drop more than 5 volts on a network. I have experienced problems with networks that seem to meet the length and current requirements, but he only way to prevent bus-off errors is to add an extra power supply in the system.
Answered by Matt Kuzel, Chairman of the Physical Layer SIG, e-mail: kuzel@voyager.net
A69) The difference of 10 volts refers to the voltage taken between V+ and V-. The reason that it is 10 is because there can be a drop of up to 5 volts on the V- and 5 volts on the V+ wire. These wires will have equal drops since they both carry the same current. The 5 volt maximum difference on V- (or V+) applies to between any two points on the network. Usually the V- voltage is measured with respect to building ground at the two locations and the difference is taken. Care should be taken since the two grounds can vary significantly. It is better to measure between V+ and V- and compare to the 10 volt maximum differential. If you are seeing a problem with bus-off under apparently normal conditions that go away with more power, it may be load transients (like at power-up) that cause current spikes. Devices should not cause current spikes and should not go bus-off when the voltage sags due to such a spike. Devices should not transmit garbage when the voltage sags causing other devices to go bus-off. One of the devices on your network is probably defective.
I read the paper "Building a Functional DeviceNet Network" by Nick Jones where it said that the 24 VDC common should be tied to the shield and connected to a low impedence ground at the power supply. We were looking at a network where the shield and DC common were both connected to the DC common on the power supply and then from there to earth ground. We noticed that between the DC common on the power supply and earth ground there was zero ohms resistance. From the shield to the earth ground, there was 1 M( resistance. The network was in operation at the time these readings were taken. We are wondering if the shield should be connected directly to earth ground and does the DC common on the power supply need to be connected to earth ground also?
Answered by Nick Jones, System Architecture SIG, e-mail: njones@sstech.on.ca
A70) If the shield and DC common are both connected to the power supply DC common and then to earth, they should both read zero ohms to earth. I suspect that the shield is not contiguous (T^s, field-installed connectors and taps are likely suspects for the open circuit) or the shield/ DC Common/earth connection is faulty. The power supply DC output should be completely isolated from earth, and the earth connection must be made externally. The reason is that if you have multiple power supplies you need to be able to control where the earth connection is. Refer to DeviceNet Specifications Version 2.0, section 9-3.9 (page 9-26): "The trunk drain/shield should be attached to the power supply ground or V- with a copper conductor that is either solid, stranded, or braided. Use a 1" copper braid or a #8 AWG wire that is less than 3 meters/10 feet in length. This should then be attached to a good earth or building ground (such as an 8 foot stake driven into the ground, attached to building iron or to the cold water plumbing). If the network is already grounded, do NOT connect the grounding terminal of the tap or ground of the supply to earth. If more than one supply is on the network, then connect the drain wire/shield at ONE supply only, preferably near the physical center of the network." The shield must be grounded or it is just a piece of metal with floating potential and may actually contribute to noise problems rather than help contr