Reply to Thread

Well in our site builder I couldn't see any modules that said there was a BBMD assigned but I didn't have the tech on site check the modules for a record. This is a possibility.

I would have liked to do a capture but they decided to skip that and went with the sure thing and made the trane system go to port 47908 which solved the problem.

For the integration points between the two systems we took our module and put it on ARCnet to connect to our system, then took the BACnet/IP port on the same network but on 47908.

Turn off BBMD and look for a change. Sounds like a definite broadcast storm to me.

Sent from my DROID BIONIC using Tapatalk 2

I don't see why you couldn't change the port number

"You all agree/disagree using two different BACnet netowrk numbers on the same network is going to cause trouble like this?"

Not sure? Anyone?

Can you move you "BACnet/IP 47808" port to another one? ie:BACnet/IP 47809. I just had to do this on a site becuse they tied to networks/buildings together with over lapping device ids with another controls vendor. Gets use both up and running till the customer decides which one of needs to change device ids and $$$$.

BCU or SC product?

I would run a Wire Shark capture just as control$ suggested. I would be ticked off also if another contractor added equipment to our module "private" network without at least giving us a call. The customer is at fault for that one.

Originally Posted by digo

That would be true on a BACnet/Ethernet network where the IP devices are connected via hubs. On a BACnet/IP network, using switches, no such collisions are going to happen. Exception being for Wi-fi devices on the same wireless network.

I think this topic came up before... here's a break down on the history of collision domains, and how a switch eliminates this:
http://www.scribd.com/doc/17536106/Collision-Domains

Good day Digo,

No, I was not really referring to Hubs, etc but in general if you have one device requesting data from a number of other IP devices and each of these devices happen to be responding at the same time, network performance (to the single device requesting data from multiple sources) will suffer. Indeed, switches have buffers (memory) which can "que" packets, but these have limits depending upon the particular equipment, packet type, Packet sizes, etc... even so packet latencies will be affected. Again, I was making a generalized statement from an out of the box view. Given MaxBurn's recent comments about their network infrastructure (i.e. Gigabit, higher end Cisco gear being used, etc) my comments are no doubt a non issue with problem at hand...

Cheers,

Sam

Also, if you are using different device numbers, I am almost positive, you shouldn't have any collision issues by using the same bacnet object numbers.

Good ole trane.... They like to keep all their information as confidential as possible no matter if your the end user or not.

Bacscope or wireshark the network. I've seen trane put the same device id's on several BCU's on the same network.

This is a gigabit backbone over fiber between about five network switches. No cheap hubs in this setup, I think this is way overblown and using cisco 3xxx series switches if I remember right. We had an independent networking contractor do this aspect, pretty confident the LAN is good here though the customer chose to not go with redundant fiber runs we were initially thinking about.

Originally Posted by s2sam

Remember that Ethernet is a collision based protocol and so adding more and more devices can result in very long packet times .
Sam

That would be true on a BACnet/Ethernet network where the IP devices are connected via hubs. On a BACnet/IP network, using switches, no such collisions are going to happen. Exception being for Wi-fi devices on the same wireless network.

I think this topic came up before... here's a break down on the history of collision domains, and how a switch eliminates this:
http://www.scribd.com/doc/17536106/Collision-Domains

Good day MaxBurn,

How many Trane BACNet/IP devices were added to your Ethernet network? I ask, as looking at it from an Ethernet perspective (ignoring BACNet for now...) the bandwidth available on your Ethernet segment may be too low given you already had 70 units on the one subnet... Remember that Ethernet is a collision based protocol and so adding more and more devices can result in very long packet times (because of the increased number of packet collisions and/or dropped packets) depending upon the bandwidth required by each of the attached devices. By definition if a packet collision occurs the offending device(s) will back off the Network and retry after a random period of time ... which means that your overall response time may be suffering here.

Just a thought from outside the box...

Cheers,

Sam

Trane was likely passed the spreadsheet we were maintaining that documented the IP addresses in use, I can't confirm it. Seems like there was no coordination on the BACnet info side other than one point we are reading on their system, which works.

We do now know that if we unplug their system ours returns to normal. I don't know if ALC modules will give an alarm for duplicate network address like windows will but I haven't seen any duplicate network device alarms.

Comm log for us was filled with "Private Transfer (Devid = 8:560080 Objid = 8:560080) No response in 3 attempts at 2000 millis. --- Adjust <BACnet/IP Connection Comm Timeout (in millis)> tuning parameter to wait longer. ---" with about every one of our device instances.

Also lots like this in the fault log "com.controlj.green.common.CJException: Could not log field trends: SyncRequest failed after waiting 360000 milliseconds. --- Adjust <BACnet/IP Connection Do Sync (in millis)> tuning parameter to wait longer. ---; Could not log trends for trend log:"

I guess a duplicate IP address could account for that, as we would be asking the wrong device.

I wouldnt think the network number should play a roll in it. I would think either you have same device id numbers or you have same ip numbers. Did Trane coordinate what IPs they used or did they just pick something at random?

I would try to get more info on what devices they added and what the settings are.

Trane BCU will generally be device 1 and network 1 is that is what they added.

BACnet basics question

I have a (large) site that has been running for a couple years or so now without any network trouble. It is SiteScan but you can think of it as WEBctrl. It has a little over seventy IP attached devices using the private IP range 192.168.168.x and there is no connection to another network. We are running BACnet/IP 47808 and network number 5600 with device instances in the range of 560xxx.

Just this week we had a trane system add themselves to the network with unknown bacnet information, they didn't coordinate with us.

Now we have dead module alarms, failures to download memory to modules and most importantly to the customer the graphics for the system take a very long time to populate with information. Seems to point to network trouble.

I am thinking that the trane system should have coordinated with us to at least match the BACnet network number with our system because it is on the same network. Also should have reviewd device instances to make sure there weren't any duplicates but I think that is unlikely due to the range we are using.

You all agree/disagree using two different BACnet netowrk numbers on the same network is going to cause trouble like this?