Intermittent BACNet Comm Failure

[I-Con Systems]

We're new to BACNet and are having a problem with intermittent communication failures. We have 2 BACNet networks on this project. Network 1 has several (20+) ABB drives on it and, after we worked out some bugs, have solid communication. Network 2 has a Honeywell BACNet Spyder and 2 York chillers connected to it and we occasionally lose the whole segment.

Details:
JACE: WEBS 6 running AX 3.4.51
Ports: Using an expansion card with 2 485 ports on it
Network 1: Standard 18/3 plenum/shielded cable (before we knew what wire to use), not terminated (I know), shield is only terminated at the JACE, and I believe that the ABB drives bias the network internally. No communication problems.
Network 2: 100ohm 22/3 1 twisted pair/1 reference plenum/shielded cable, not terminated (I'll explain later), shield is only terminated at the JACE, and no network bias.
The JACE is the first device on both networks with a MAC Address of 0.

History:
A few days after bringing Network 2 online the network went down. All devices were powered but were not communicating over the network and showing "down" in the JACE. Power cycling chiller #2, the last node on the network, brought the network back online. A few days later, it happened again and was resolved the same way, by power cycling chiller #2. Originally this network was wired using the same cable used on Network #1. After the second failure the cable was replaced. I tried terminating the network using 100 ohm resistors matching the characteristic impedance of the cable, this would bring the network down imediately so they were removed. The network was fine for a couple of months but went down again yesterday. I will be on site tomorrow and will attempt to bias the network using a method recommended by Honeywell in their BACNet Spyder literature.

Any other suggestions? Any ideas why the termination resistors brought the network down?

Thanks,

Charles

[supertek65]

i can not help but may ask you a question??
what is biasing????
i have heard a lot of guys having this problem and they always say the answer is to add resistors and bias????

[I-Con Systems]

Unlike a LON network where communication is controlled by timing, a BACNet network is controlled by the devices. A device will wait it's turn to transmit over the network. When a device takes over the segment it will "pull up" or "pull down" the network while it has control. When control is released, the network needs to return to a neutral state so the next device knows the network is available. A network that is not biased may remain in this pulled up/down state after control is released halting communication because the controllers think the network is in use and refrain from transmitting.

Some controllers (check with the manufacturer) have built in biasing, others do not. If you have a segment where the controllers do not provide biasing, auxiliary biasing needs to be added. This is usually achieved through a set of resistors.

I've also learned from another board that you cannot terminate a non-biased network. In short, a BACNet network needs to be biased AND terminated.

[carrier_jci_tac_tech]

1. first I would start with address no gap in addressing device. correct 4,5,6,7 not correct 4,8,10,11
2. how long on com bus may need a repeater.
3. check all connection. make sure end of line is installed correctly. at both end of com bus.
this should do it.

[supertek65]

thanks Icon
that was a great answer and even a hammer and pipewrench guy like me gets it
frank

Unlike a LON network where communication is controlled by timing, a BACNet network is controlled by the devices. A device will wait it's turn to transmit over the network. When a device takes over the segment it will "pull up" or "pull down" the network while it has control. When control is released, the network needs to return to a neutral state so the next device knows the network is available. A network that is not biased may remain in this pulled up/down state after control is released halting communication because the controllers think the network is in use and refrain from transmitting.

Some controllers (check with the manufacturer) have built in biasing, others do not. If you have a segment where the controllers do not provide biasing, auxiliary biasing needs to be added. This is usually achieved through a set of resistors.

I've also learned from another board that you cannot terminate a non-biased network. In short, a BACNet network needs to be biased AND terminated.

[bobby m]

http://www.chipkin.com/files/pdf/Bac...echnicians.pdf

Maybe this wil help ya. After reading it i know why some people prefer lon,also look on there website they have some free tools for scanning bacnet networks.There are also more free tools at wireshark.com do a search there on bacnet --cool stuff.