Siemens MBC communication problems

[davidminear]

Hello,

Long time reader - first time poster.

I have a Siemens BACnet MBC running firmware MBCE V3.0 BACnet 4.2g. This MBC is an open processor only, gathering data via FLN1 from a Trane Summit BMS interface. This MBC controls our chillers. The MBC is installed in an enclosure with a second MBC and point termination modules. The MBC communicates with Insight via Ethernet.

I frequently (100+ times per day) lose communications with this MBC. All or most of the points will go "FAILED". The panel can stay FAILED for 30 seconds or 30 minutes depending on it's mood.

I am monitoring the MBC diagnostic points and I get many NO.RESPONSES and my NAK.COUNT climbs continuously.

We have replaced the Siemens Trunk Interface II. We have re-terminated the FLN connection (and made sure only one end of the shield has been landed.) We have verified the FLN is set for 9600 baud. We have also cold-started the processor several times. Siemens techs so far unable to provide much assistance.

Can anybody provide any more ideas of different things we could try?

Thanks,

[davidminear]

Some more info that may be helpful to the experts out there....

I was looking through System Activity Log and noticed that some of my COMM.FAIL alarms come in right after a failed trend collection in that MBC. We had many analog points being trended on COV and there were not enough panel samples available to store the data so they were being dumped. I thought I would delete the COV trends and set them up as interval trends. Deleting the COV trends worked okay but as soon as I tried adding the interval trends the panel coldstarted. I added (4) new trend definitions and the panel coldstarted after each one.

Again, thank you for any assistance you can provide.

[Ex-Man]

I haven't worked on Siemens equipment in 5 years now & I haven't worked on their BACNet equipment but what you describe sounds like the panel is low on RAM. The panel is probably rebooting because it has run out of scratchpad memory (the RAM it uses to do its temporary calculations) due to the trends having already filled the RAM with data. My suggestion is to get out the manual & figure out how much RAM is needed for the points, PPCL, trends, ect. You may have to have the amount of RAM increased to do everything you want to with this controller.

[Knife Switch]

Double checking the trend definitions and making sure that there arent any trying to collect from dead/old points might help? Consider adding SAMPLE(30) or something to the code where it's looking at points that aren't in that particular panel, just to lighten the network traffic. Wireshark can be really helpful in singling out lines of code that are generating alot of unnecessary network traffic

[VAGR]

Dear colleague, it could be a reason of electromagnetic noises in the operating circuit. Try to use shielded cables and also the current filtration devices to make a constant stable current.

[behappy]

I haven't worked on Siemens equipment in 5 years now & I haven't worked on their BACNet equipment but what you describe sounds like the panel is low on RAM. The panel is probably rebooting because it has run out of scratchpad memory (the RAM it uses to do its temporary calculations) due to the trends having already filled the RAM with data. My suggestion is to get out the manual & figure out how much RAM is needed for the points, PPCL, trends, ect. You may have to have the amount of RAM increased to do everything you want to with this controller.

What he said...
Limit your trends and do away with what you really don't need.

T.I.'s are notorious for having loose or slightly corroded terminals, getting week and crashing (possible power supple problems). When we finally went 100 % Ethernet and upgraded memory (and to PXCM's), all these problems went away.

[mrc856]

I haven't worked with your specific driver, but I have worked with the Siemens - Trane integration driver which has the same physical connectivity as what you describe. As 2 of the previous posters have mentioned it's probably memory related.

How many points are you bringing in from the BCU?

Get rid of all trending in the bacnet panel and ppcl if possible.
If needed set up virtual points and write ppcl in the other MBC to look at the bacnet points.
Trend the virtual points on time.

The panel cold-starting after building a single trend is not normal. Although a lot of what you describe sounds memory related, this issue stands out. Have you tried coldstarting the panel (disconnected from Insight) and building one single point and trend? Does the panel cold start?

Might be time to think about upgrading the the modular platform.