Ground loops

[valdelocc]

Having an issue with a boiler control losing feedback (modbus) from a servo motor and locking out, the problem happens once or twice a week, late at night/weekends when the boiler is lightly loaded. I have swapped the main control and the servo with a known good and the issue persist.
I pretty sure electrical noise is the root cause, they used twisted pair shielded wire between the control and the servos and grounded the shields at the main control panel and to every servo, I'm thinking about removing the shield from the servos grounds and leave it grounded to the main panel only, what do you guys think?

[DaveCR]

Change it; the shield should be grounded at only one point. If the transceivers are three wire capable (they have a common) then upgrade the wire as well.
Other things to consider that generate electrical noise are VFDs, microwaves and UPSs.

[valdelocc]

Change it; the shield should be grounded at only one point. If the transceivers are three wire capable (they have a common) then upgrade the wire as well.
Other things to consider that generate electrical noise are VFDs, microwaves and UPSs.

Thanks! I'm pretty sure the source of the noise is a Honeywell VFD, the nuisance trips happens when the VFD output drops under 28hz, I have bumped the minimum HZ to 30 and it hasn't trip again. I'm not positive at what point the noise gets injected into the control system, there is an isolation transformer stepping the 277V down to 120V to power up the controls, this transformer has a floating neutral and there is 1.7V potential between neutral to ground,(not sure if it matters) but bonding the neutral at the transformer in not an option.

[numbawunfela]

Change it; the shield should be grounded at only one point. If the transceivers are three wire capable (they have a common) then upgrade the wire as well.

I second that. Ground the sheilds in one spot only.

3 wire capable stuff will tke the usual 'shield' trrminal and call it a REF terminal for reference. It is easy for the wire puller to miss if he is not paying attention. Some vendors say put sheild to it anyway (distech, Schneider among others). But be careful with that. An oscilloscope can tell you if noise is your problem quick.

[valdelocc]

Do you think is possible for the noise generated by the VFD to jump from the primary side of the stepdown transformer to the secondary and its being conducted into the control loop that way or is more likely for it to be radiated and injected into the communication wires that way? since my last visit the boiler hasn't gone down again, I have a noise suppressor filter and a bunch of ferrites cores on order.

[BALloyd]

I second that. Ground the sheilds in one spot only.

3 wire capable stuff will tke the usual 'shield' trrminal and call it a REF terminal for reference. It is easy for the wire puller to miss if he is not paying attention. Some vendors say put sheild to it anyway (distech, Schneider among others). But be careful with that. An oscilloscope can tell you if noise is your problem quick.

X3. Usually ground shields at control board end only.

Sent from my SM-G965W using Tapatalk

[numbawunfela]

Do you think is possible for the noise generated by the VFD to jump from the primary side of the stepdown transformer to the secondary and its being conducted into the control loop that way or is more likely for it to be radiated and injected into the communication wires that way?

Noise is introduced from a VFD when the High voltage, high frequency power induces voltage into the low frequency comm trunk. Since what is induced is not from a controller, the controllers can read it as 1s and 0s and wonder 'what is that!?!'. Properly grounded shields drain that noise to keep it from the comm wires. Sometimes you can get the comm wires away from the noise... but not usually when they go into a VFD.

I had a job where the installers had zip tied comm trunk to the power leads to a 208v compressor and mounted controllers inside package units nestled among the 208v in there. All I did was give a foot or more where possible of distance between comm ttunk and anything high voltage and I grounded the shields in 1 place and I had no offline occurances anymore. Only 5 RTUs and only 8 total controllers and the last one couldn't stay online until that.

https://youtu.be/zT_tXupPE4E

I made this for our guys on that site. Bink kind to me HTalkers, it was 102 on the ground, hotter on the roof. I might have gotten my words mixed up at some point. Don't roast me alive if I say condenser fan and not evap fan or something silly.

[kontrolphreak]

Great video, we need more of these "tribal knowledge" type videos of our experiences for junior technicians to gain knowledge.

kontrol out

[pecmsg]

Noise is introduced from a VFD when the High voltage, high frequency power induces voltage into the low frequency comm trunk. Since what is induced is not from a controller, the controllers can read it as 1s and 0s and wonder 'what is that!?!'. Properly grounded shields drain that noise to keep it from the comm wires. Sometimes you can get the comm wires away from the noise... but not usually when they go into a VFD.

I had a job where the installers had zip tied comm trunk to the power leads to a 208v compressor and mounted controllers inside package units nestled among the 208v in there. All I did was give a foot or more where possible of distance between comm ttunk and anything high voltage and I grounded the shields in 1 place and I had no offline occurances anymore. Only 5 RTUs and only 8 total controllers and the last one couldn't stay online until that.

https://youtu.be/zT_tXupPE4E

I made this for our guys on that site. Bink kind to me HTalkers, it was 102 on the ground, hotter on the roof. I might have gotten my words mixed up at some point. Don't roast me alive if I say condenser fan and not evap fan or something silly.

That was in Jersey?

[s2sam]

<snip>
3 wire capable stuff will tke the usual 'shield' trrminal and call it a REF terminal for reference. It is easy for the wire puller to miss if he is not paying attention. Some vendors say put sheild to it anyway (distech, Schneider among others). But be careful with that. An oscilloscope can tell you if noise is your problem quick.

Good day Numbawunfela,

Just be aware that "shield", "Ref", "Reference", etc cannot be used interchangeably and/or are not the same connection on every device. technically "Shield", "Reference", and "Ground" are actually different electrical connections and should be used appropriately. For example:

1. a "Shield" connection could simply be a common connection point with no real electrical connection to a RS485's reference or the controller's ground electrical potential.

2. a "Ref" (RS485 Reference_ connection is common electrical potential as an electrical reference for the controller's RS485's transceiver(s).

3. a "ground" has to be taken in context of the electrical circuitry. In some cases a "ground" could be a chassis ground, or an Earth ground, or simply a reference electrical potential for a particular circuit.

Sadly, the reason for the confusion is that a number of the electrical equipment manufacturers are inconsistent in the terminology... and in a number of cases they have a clear misunderstanding of how RS485 (EIA-485) works and how to design their interface properly.

Cheers,

Sam

[s2sam]

Noise is introduced from a VFD when the High voltage, high frequency power induces voltage into the low frequency comm trunk. Since what is induced is not from a controller, the controllers can read it as 1s and 0s and wonder 'what is that!?!'. Properly grounded shields drain that noise to keep it from the comm wires. Sometimes you can get the comm wires away from the noise... but not usually when they go into a VFD.

I had a job where the installers had zip tied comm trunk to the power leads to a 208v compressor and mounted controllers inside package units nestled among the 208v in there. All I did was give a foot or more where possible of distance between comm ttunk and anything high voltage and I grounded the shields in 1 place and I had no offline occurances anymore. Only 5 RTUs and only 8 total controllers and the last one couldn't stay online until that.

https://youtu.be/zT_tXupPE4E

I made this for our guys on that site. Bink kind to me HTalkers, it was 102 on the ground, hotter on the roof. I might have gotten my words mixed up at some point. Don't roast me alive if I say condenser fan and not evap fan or something silly.

Good day Numbawunfel,

Indeed, VFDs can be nasty beasts with regards to RS485 (Bacnet, JCI N2 , etc) communication and at times shielding may be the only way to obtain reliable communications. However, it is important to understand a few things regarding RS485 and electrical noise so that one can better diagnose and sort out these type of issues.

Firstly, RS485 is a differential signaling technology. What this means is that the digital data is derived from the electrical difference (subtraction) of the voltages as seen on the RS485's "A" and "B" wires (or "N2+" and "N2-" on JCI N2 Buses). Because of this signaling approach any electrical noise on a cable will be seen equally on the "A" and "B" wires and thus cancels out because of the subtraction. Because of this RS485 is actually very noise immune. Issues arise because people use the incorrect cable, do not adhere to proper RS485 wiring practices, or overload the bus (e.g. too many controllers on a bus). This cripples or limits the RS485 transceivers from reducing and/or cancelling any induced electrical noise.

The second issue is incorrectly designed RS485 interfaces. Although only the "A" and "B" wires are needed to extract the digital data, the third wire, the "reference", is actually very important for everything to work correctly. The reason is the "reference" is used as an common electrical reference voltage as seen by the RS485 transceivers. Not having the "reference" could result in excessive voltages seen by the RS485 transceivers resulting in damaged RS485 interfaces and/or misinterpreted digital data.

As for induced electrical noise... As you mentioned in your video, separating the RS485 bus from large voltages is always a good thing. However, keep in mind that the magnitude of the induced electrical noise is inversely proportional to the square of the separation distance. Thus moving the communication wires by a small amount has a big effect because of the distance squared.

As for shields... I avoid using them and only use them when absolutely necessary. If one uses the proper cable, wiring practices, and have proper RS485 interfaces, shields are not needed and at times actually create problems. In the almost 30 years that I have been designing and installing RS485 systems, there has only been a small number of times I had to use a shield... and in this cases I included some additional circuitry to the shield to minimize the shield's unintended effects (noise). Depending upon how the equipment is electrically designed, the shield can act as a long and effective antenna which provides a path for 60 Hz and other higher frequencies to be induced into the equipment and onto the RS485 bus.

Anyway, I hope the above is helpful.

Cheers,

Sam

[MaxBurn]

Some controllers like ALC also play fast and loose with the "reference" wire terminology. On some controllers it's marked shield and others it's marked ground, but it's actually neither, it's a RS485 reference and has no electrical connection to ground. I've seen this on other RS485 equipment occasionally. I think you can generally tell the difference by doing an ohm check to chassi ground, if it's a short it's actually a ground and not a reference terminal, which seems to be somewhat rare. The whole point is you do want to tie your references together but if they are actually real ground connections you don't as mentioned above.

[numbawunfela]

That was in Jersey?

No sir. I leave the location in my profile as NJ to keep the man off my tail..... ;-)

[numbawunfela]

Good day Numbawunfel,

As for shields... I avoid using them and only use them when absolutely necessary. If one uses the proper cable, wiring practices, and have proper RS485 interfaces, shields are not needed and at times actually create problems. In the almost 30 years that I have been designing and installing RS485 systems, there has only been a small number of times I had to use a shield...

Anyway, I hope the above is helpful.

Cheers,

Sam

Interesting, I never heard that before....

I can tell you that aside from moving the controllers all I did was land the shields (which were unlanded anywhere, but all tied together) and the site has been happy since.

Out of curiosity, anybody else avoid using shileds?

[numbawunfela]

Just be aware that "shield", "Ref", "Reference", etc cannot be used interchangeably and/or are not the same connection on every device.

Super reminder... did I get them mixed up in my video? I tend to be careful about those things....

I think MaxBurn is on track. Smart guys know what these things are, sometimes in spite of the Manufacturer. Some say ground the Reference, but in reality when you do it can bite you hard. It works great until it doesn't and then you are screwed. Your post outlined the difference well.

[s2sam]

Interesting, I never heard that before....

I can tell you that aside from moving the controllers all I did was land the shields (which were unlanded anywhere, but all tied together) and the site has been happy since.

Out of curiosity, anybody else avoid using shileds?

Good day Numbawunfela,

Interesting, I never heard that before....

If you have ever used Ethernet (which I am sure you have), you have used cable without a shield and it works really well. To be fair Ethernet twisted pair signaling technology is not the same as RS-485... but you get the idea. Also, Ethernet is significantly faster and yet it works fine even in really noisy environments.

My intent was to remind those who use shields to understand why they are used, as opposed to simply use them "because that is how it was always done".

As for your project... one cannot argue with success! As I mentioned, sometimes a shield is needed, but it is important to understand why, what it is trying to achieve, where the shield is connected, and what affects the shield could impart on the communication bus. Depending upon where the shield is terminated and how this connection relates to the controller's and line voltage electrical connectivity a number of things could result:

a. can add additional parasitic capacitance to communication wires resulting is inadvertent filtering of the signals... which can limit data rates and/or signal integrity

b. can create a line voltage ground loop which could violate electrical codes (thus be unsafe)

c. create an antenna and thus provide a pathway to introduce electrical noise into the equipment. If the shield is connected to a reference which is connected to the RS-485 transceiver, then electrical noise will create a localized ground bounce in the equipment causing problems with the equipment (offlines, hangs, inaccurate analog values, etc).

Cheers,

Sam

[bolt950rider]

So where does this reference terminal come up with its reference? Like do all controllers that have it on a trunk share the same, does the router/super provide it or are they individual per controller?
Is it like a common to the + and - terminals?
Is there down sides to having some controllers with a reference and some not on the same trunk?(as long as all references are wire nutted)
Ive been getting more into JCI stuff and it’s a little frustrating even for their own equipment I hear to use it on the PCGs but not VFDs, and not TECs. I’m more used to just a +,- and shield.



Sent from my iPhone using Tapatalk

[numbawunfela]

My intent was to remind those who use shields to understand why they are used, as opposed to simply use them "because that is how it was always done".

a. can add additional parasitic capacitance to communication wires resulting is inadvertent filtering of the signals... which can limit data rates and/or signal integrity
Sam

Very interesting! I also hate the 'we always do it that way' mentality.

[numbawunfela]

So where does this reference terminal come up with its reference?

Reference is a ground reference for use by the communications only. All controllers need this, but 2 wire bacnet uses what the controller is tied to as a reference. Hopefully that is a grounded secondary on a 24v transformer. Otherwise the controller tries to generate 3v (give or take) and send bacnet packets buf it may be that one controllers 3v is different than another controller's 3v, because they do not share a common ground reference. This is especially hard when mounting controllers in a bunch of residential heat pumps - separate, possibly ungrounded units.

So a reference takes comms away from all that by creating a 0 reference against which that 3v is measured. The transceiver generates it internally. All of the transcievers have a ground on the integrated circuit, but 2 wire manufacturers route it to a ground terminal for the controller, 3 wire manufacturers route it to the reference terminal on the comm bus. Then when tied together, all the references match each other, and therefore everyone's 3v match. Now your comms dont care if the unit or transformer is ungrounded. The HVAC hacker from 10 years ago now plays no role in whether my comms work today. This is why I personally feel 3wire bacnet is WAY better than 2, but only Siemens and Johnson use it (to my knowledge).

I hope that helps!

[numbawunfela]

Great video, we need more of these "tribal knowledge" type videos of our experiences for junior technicians to gain knowledge.

kontrol out

Thanks KP! Glad it did not suck..