Bacnet MSTP shield grounding?

[fxb80]

What is the real world experience with shield grounding on an RS485 network? Do you know of networks that had problems caused by the shield being grounded to earth (equipment ground) at each node that were fixed by grounding the shield at just one end?

Why I ask: We just installed our first Bacnet MSTP network. It has 8 Honeywell Smart VFDs and a JACE. I did a fair amount of reading and searching threads to find out what good and normal practice is for the cable shield. It seems there are two schools of thought. One, ground the shield at one end. Two, ground the shield at each node.

I chose to ground the shield at each node. The Honeywell instructions said to connect the shield to the VFD cabinet, equipment ground. There is no REF connection on the low voltage board. It seems to make sense that grounding each node would eliminate node-to-earth potential differences between locations. I do understand that connecting to equipment ground provides a path for objectionable flow through the shield, but it seems that would mostly be the case if there is a problem with the equipment grounding system that is part of the power supply wiring.

The network was analyzed on startup and it has a good signal with very little noise. However, all possible noise makers are not yet active or will be active intermittently. Our sales rep was surprised that the shield was grounded at each node, said that is not standard practice (in apparent contradiction of the installation instructions) and that a first thing to try if problems do develop is to ground the shield at just one end.

It seems to me that the preferred way is to connect shield to REF at each node. But what is the best thing to do, based on actual experience, not just theory, when there is no REF connection?

[trouble time]

My experience with it is limited to the early days of VVT systems. The reasoning behind the grounding only at one end was to prevent it from becoming an FM antenna and picking up stray signals. ......

[doubleduece]

Ditto on the Earth ground only once. Some manufatures require the ground landed on each device some dont, bottom line dont make a antenna and olnly earth ground once.

[fxb80]

Ditto on the Earth ground only once. Some manufatures require the ground landed on each device some dont, bottom line dont make a antenna and olnly earth ground once.

So, if the manufacturer's instructions call for landing the shield to ground at the device and there are 100 devices, follow the instructions for one device and deviate from the instructions on the other 99 devices?

[MatrixTransform]

there are parts of the industry that say that both MSTP and LON dont require shielded cable.
This is good in theory and usually works just fine...it is mostly idiot proof ....But not always.
Be strategic about it ... make ENGINEERING choices.
running UTP in and out of VSDs or across light fittings is simply asking for trouble
stick strictly to distance recommendations.
select the cables carefully! Use a specified signal cable ...The most expensive cable will probably be the right one to use. It will have a low capacitance, 120ohm impedance, a foil shield AND possibly a braid as well.

Shielded cable is designed to defend against electromagnetic interference ... and buildings are inherently noisy environments.
the shield goes to GROUND. Typically only at one location. Maybe best at the place where the cable originates.
the REF connection is NOT a shield connection... if there is a SHLD point at each device it will probably not be connected to earth.
if there is a REF(or COM,or 3rd wire) connection at each device this is probably the RS485 reference. It pay or may not be connected to earth internally via pull-apart resistors. This 3rd wire connection is NOT where the shield wire goes.

when you do use shielded cable to protect from noise, you introduce other problems and sources of noise.(and the cable can act like an antenna and pick up all sorts of crap and couple it into your comms)

If there is noise and it is getting onto the cable then you need get heavy ...

I suggest taking it to ground through an RC network at multiple points.
We use 470kΩ, 1/4W, ±5% resistor in parallel witha 0.1 µF, 10%, Metalized Polyester, ≥ 100V

How the parallel RC actually acts is complex (im not even gonna try to know) :
1 - prevents static build up in the cable
2 - provides a path to ground for high freq AC. (such as 5/10/20kHz noise from VFDs)

Our policy is to always use shielded cable
Always use proper EOL terminators and shield terminators.
avoid star topologies like the plague.

[fxb80]

Matrix Transform, thanks for taking the time to respond. Would you clarify a few points?

Since we are a Honeywell dealer, we used the cable specified by them, Belden 9841. We actually used the SmartWire equivalent, 042002-S. This is 120 ohm cable so we used 120 ohm termination resistors. Do you know if this is a best cable to use?

What do you do when a device has an REF connection, do you use 3 conductor twisted cable with shield?

If I understand correctly, you add the RC only when noise is an issue? Or should the RC be added at the single point ground termination as standard practice, and then added at multiple points when noise is an issue?

Obviously, we are still very much in a learning phase. But, we always use daisy chain topology and maintain at least one foot separation from lighting and other sources of interference when possible. Most of our work lately has been industrial. In these places I chose to run the comm cables in a dedicated EMT conduit and still maintain separation from noise makers. Do you think a dedicated conduit is overkill? Is it OK to run comm cable with low voltage sensor and power cable over long distances?

Thanks again,

CW

[kontrolphreak]

One of the major mistakes installers make is running low voltage conductors in parallel to hi volt, this should be avoided as well. All crossovers should be at 90 degrees / perpendicular with as much separation as possible.

kontrol out

[PM_Dude]

I worked for JCL for about 14 years and also had my own controls company that was an authorized dealer for Automated Logic & Reliable Controls.

I can tell you that the majority of projects of any complexity we executed the biggest challenges were those that had troubles with communication.

We standardized on low capacitance network communication cable (< 25 picofarads/ft) and with installation procedure improvements aside, this cable resulted in significantly reduced downtime troubleshooting network issues.

ALC & JC used true RS-485 3-wire connections, but Reliable used only 2 terminals. (The REF terminal is actually the RS-485 driver ground which can be significantly different potential than the ground or shield connection we anchor at one end.)
Reference this article...
http://www.chipkin.com/articles/rs485-cables-why-you-need-3-wires-for-2-two-wire-rs485

Hope that helps.