Bacnet Cards Damaged

Above is the picture showing the termination of cable with the card.

Carrier gave you an answer, why are you fighting their conclusion?

He wants to understand the problem in order to fit a soultion before he loses more cards. But the answer might still be the same. Would Carrier specify or make a wiring correction ?

explaining Bacnet and DDC controls would take me weeks, and thats just the programming!

Sounds from the OP as a wiring induction problem. Carrier is pointing to it; how hard could it be lol. In my experience Shield = signal gnd, Drain = frame gnd. OP already stated no grounding in the runs, wonder how this got past installation. 4 volts to frame ground ? could be a lot more. Ungrounded shield / drains always bother me.

Recently, two of the chillers lost their communication with BMS.<<< does sound like it was online to me. no mention of anyone changing wiring

Carrier just concluded about the induction; they did not analyze the root cause. But we will not replace card unless the reason is being found out.

I am not fighting the conclusion made by Carrier. But I should ensure that the root reason for the fault in the system is analyzed and rectified by them prior to replacing the new cards. their conclusion actually help me open the panel and to visualize the wiring and the induced voltage.

Also good info, often overlooked.

All the communication cables from various devices to the BMS are run through dedicated GI Trunking. This trunking is properly bonded at all joints and are finally connected to earth. And the card is along with other module cards in the metallic box originally supplied and installed by Carrier.

For verification, I looped a separate earth wire from last point of Signal Ground terminal and connected with the metallic enclosure of the box in which the last card is installed. Now the voltages across all the wires (A, B & Drain/shield) with respect to the ground is zero.

Then I removed my new grounding to the communication cable. It was reinstated into the previous condition and am waiting the feed back from Carrier prior to replacing the card.

The chiller was installed and commissioned by same Carrier team two years ago. No one did altered any wiring inside; the system was communicating without any issue. And the surprising thing is that, at the end of BMS side, this drain wire is not connected to the terminal of Automation Station (PXUOOU.1 by Siemens) - only the two wires A & B are connected with it. The shield and drain kept tapped.

4 volts what? AC at power line frequency?

Your outermost shield (drain wire) should be grounded to the same reference which your power supply's input power (AC) power is derived from (frame), not a floating ground, which it sounds like your Bacnet power supply uses. Ground the drain at the source end (where the Bacnet's power is provided). Low frequency interference is best handled by grounding at one end only, and managing the length to avoid making an antenna of your wiring.

If you reference your Bacnet ground to frame (unless that violates the warranty or some such) it might solve the issue too, though then you may have too much noise for the system to run at all, eh?

Tech support from Carrier should be able to explain this in the field and demonstrate that their product works as installed.

The only thing intelligent I have to say here is:

And Signal ground is connected to?

and shields are only connected to earth at ONE END. The rule of thumb is the signal source end. ONE END is the necessary condition.

Carrier probably isn't going to be much help.

Ask carrier for an equlivilent circuit of the external connections to the module. Only then can you intelligently solve the problem.

Distance may matter.

You may have a ground loop.

and shields are only connected to earth at ONE END. The rule of thumb is the signal source end. ONE END is the necessary condition.

Very good point! Often forgotten...

Some comments on RS-485 communications in general. These comments are not directed at this particular vendor because for all I know, this vendor's 485 might be the exception to the rule, but I have found that the rule is that industrial serial communications is typically a poor-cousin add-on on most equipment.

Its design is frequently outsourced to those who may or may not understand industrial electrical environments, because, after all, the Modbus works on the test bench, right?

It's not just the service techs who are not communications gurus, it's usually the vendor's engineering staff as well. Their expertise is elsewhere, in whatever the primary equipment is.

Unless the company's core competence is communications, like Prosoft or B&B Electronics, communications is just a tag-along, minor add-on to the real product, whatever that is.

There are several app notes that describe how 2 wire RS-485 requires a 3rd wire as a signal reference, which I hesitate to call 'ground' because all too often, that signal reference 3rd wire in a 2-wire RS-485 network is not brought out to a terminal. So the circuit uses an earth ground as reference. Since RS-485 has very limited common mode ability, it doesn't take much in the way of a ground fault to pump current through the 2 wire comm circuit and poof, the comm circuit is toast. But you mentioned signal ground, so this comm card might be one of the few that has a separate signal ground.

My experience is that one can complain about how the equipment vendor needs to come up with a bullet proof comm network, but it will be in vain. The vendors typically produce a 'working' comm network, in your case, one that worked for two years. And then a fault somewhere blew the cards. It's their fault? That's a hard case to make.

In the competitive world of capital equipment sales, it isn't worth the investment of designing and implementing a bullet proof comm network. When the German government wanted something close to that for its automotive export industry, it paid Siemens to develop Profibus. Big dollar project.

A really robust RS-485 comm port isn't 'sellable'. There's no way to re-coup the investment. I will guarantee that the robustness of the Modbus RS-485 link was NEVER a topic of discussion in the procurement of the equipment.

Some options you might consider:

1) You got two years of use out of the comm cards. Consider replacing comm cards on an occasional basis as the cost of doing business. Does the value from the data or control you achieve from Modbus comm with these units warrant amortizing the cost of replacement comm cards on an occasional basis?

>But we will not replace card unless the reason is being found out.

That statement leads me to believe that there is really little value derived from the communications.

2) Install an RS-485 isolator at each node. They probably cost about $300-$400 each, and optically isolate the comm card from the network. I routinely use at least one on a device whose serial comm port is part of a much more expensive controller card in order to protect the controller or backplane, as much as the comm port.

Again, does the value from the data or control you achieve from Modbus comm with these units warrant using isolators as protection from electrical faults on the network?

Hey, where I worked our computing center was sort of lightning struck. Over the length on the center, the grounds changed say over 400'. and burned out a couple of Ethernet switches that traversed the length. It's now replaced by fiber.

This is one reason why I asked about wire distance and even distance the way a bird flies.

Isolators did a world of good on a X-Ray set that communicated via RS232. It was either replace the $4.00 in driver chips 1488/1489 I think (socketed) when it failed or invest about $100 in an isolator. No problems since. A 100 kV powersupply at 0.1A nearby has potential to inflict some damage. The vendor suggested replacing the chips.

Wire distance is less than 200 mtr out of which 25 mtr is laid at low level on the roof floor of the bldg. Lightning network cables are laid around the periphery of the bldg at high level (elevated at 3 mtr from the horizontal run of the GI Trunk.

Thank you Iris for providing the link about the RS 485. Even though I am not in the field of electronics, especially, in the 01010101 field, it helped me to understand communication path.

Actually, the installation of chillers was done M/s Carrier and the BMS by M/s Siemens. Normally, the operational status for start and stop commands for the chillers are initiated from BMS. Now, as reported by BMS team, since there is no communication between the chiller and the BMS because the Modbus interface withe chiller is lost, it can not be operated in auto mode. At present, the command for start, stop and cycling are provided by manual.

The bacnet card has two communication connectors/ports. one is connected to the its own CCN control section in another PCB and its terminals are marked as CCN Data -, CCN Signal Ground and CCN Data +. The second connector is for RS485 and is marked as RS485-, RS485 Ground and RS485 +. And there is special note also given by Carrier for this terminal in the installation manual that the wiring of this terminal should be as required by the third party's Modbus/BACnet communcation requirement. This communication cable by the third party (by Siemens) is wired using only two cables + and - at their automation station. Now since there is ground terminal found on the RS485 connector side, the guy who terminated the drainwire of the cable to the Ground terminal of the connector. And it is terminated at each connector on every card. As noted by other forum member, the drain wire should be connected to only at one end and should to the direct ground. So the fault might be with this type connection ???

RS-485 is messy because every manufacturer's implementation can be anywhere on the spectrum from excellent to marginal to bad.

If there only 2 terminals, then the signal ground can only assumed to be the case connection, an earth ground. This presents the most real-world problems because RS-485 is used in multidrop situations where the local grounds between two different devices can be significantly different resulting in a ground loop. The design works great on the test bench, but can be problematic in real electrical environments.

Even when there are 3 terminal connections for 2 wire 485, sometimes the 3rd connection is not a separate 485 signal ground, it's case/earth ground. Not a good design, but I've seen a number of them. Same issues as above, just a 3rd terminal.

I suspect from Carrier's terminology that their 3rd terminal is signal ground. But where does it connect on the Siemens side? You can disconnect the shield on one card, connect an ohmmeter and see whether the 3rd terminal has a direct connection to the earth ground/case connection. If it does, then the 485 is using the case as its signal reference. Even if the 3rd terminal is not case ground, but a signal ground, what do you do then? There's no signal ground connection on the other end! Are you going re-engineer the automation station?

Could connecting the shield a signal ground cause a problem? Yes. If the Carrier 3rd terminal is a signal ground, you can disconnect the shield and move the shield to a ground connection.

The reality is that the 485 network works under normal conditions. But then there was a fault somewhere which took out the comm cards with their long cabling that acts like an inductive antenna or ground loop path. Even if you discover, somehow, the root cause, what can you do about it? What if it was a close lightning strike? Disconnecting a shield from the signal ground is a start. But that doesn't guarantee a trouble-free future.

FYI, Siemens, who engineered the very robust RS-485 network for Profibus DP connects shields at both ends of its 485 cable drops. Here's a screen shot from the Profibus DP installation guide (V1, 1998, the current version posted on the US Profibus site as of today).

Of course, Siemens' expectation is that the entire site is "equipotential". Uh-huh.

Your control scheme uses Modbus. You really need to consider replacing the comm cards and adding RS-485 repeater/isolators.

Isolator/repeaters all isolate input to output. The more expensive ones have power supply isolation. Some have surge protection on the driver lines, but even so, they're relatively inexpensive and even if they're damaged by faults, they're typically a DIN rail mounted unit that can quickly be swapped out which leaves the main comm port intact and undamaged. I buy the ones with removable connectors so that the wiring doesn't have to be disconnected with if the isolator needs to be replaced.

Thank you Bud;The third terminal on the card did not show any continuity with the ground/case when I checked with the ohmmeter. When I grounded the shield dir ectly to the case, the induced voltage is zero. I removed all the terminals from start to end; also removed the new ground connection, then also voltage exists. This increased my curiosity, and I continued my attempt. I found another fact also. The power supply to the chilled water valve actuator, which is 24V ac is run along with the communication cable in the same trunking. Not one, there are such six nos of power supply which is laid within the same compartment. May it may be a reason also. Since the actuators are on duty, I did not switch off the power supply to them to verify the induction.

Another thing which I forgot to inform you all is that, both the communication cards were not damaged simultaneously. The first one lost its communication four month back, when the Carrier guy came to service the chillers. He normally switch off the chiller, service it and put it back to the system. We informed them about the issue and simply told that it is BMS problem. BMS guy told it is their problem.The second one lost the communication on the same day they had done another periodic service last month. Carrier replied the same sentence that it is BMS issue. They admitted the fault only when we asked them to show the integrity of the card interface with another monitor.

I have to second Bud's recommendation. But then again we know each other from previous lives and have some war stories in common.

What a surprise, dude. Long time, no see. I thought you'd sworn off serial comm after the XXX episode. Check your private messages.

Bud has the right idea,

The cable along the rooftop might be part of the problem. You do have to identify signal grounds and Earth grounds in the SYSTEM.

They have to be connected at ONLY ONE POINT! otherwise you may need some isolation somewhere.

And as a reminder, shields only connect at ONE end.

The further systems are apart, the more likely they potentials are unequal. Surge protection might help.

If you don't have to return the boards and if the RS485 transceivers can be identified, it's possible that that is the only problem.

You can always afford the time to do it over, but never can afford the time to do it right.

You might have to resort to fiber and/or ethernet as well.