Cause of RFI during chiller compressor operation 1

[PQ101]

As soon as the newly installed chiller starts-up the reception on AM radio signal disppears for appx. 5-6 blocks. The 200Ton chiller is equipped with harmonic filter. Equipment manufac. has replaced all circuit boards, the SCR panel, installed isolation transformer, chokes around the cables, installed PVC on incoming power conduit, etc. but made no change.

You can basically hear compressor motor whining-distortion on AM radio anywhere from AM500 thru AM1500.

To start with grounding from 480V Main panel to isolation transformer to chiller starter enclosure to compressor motor was checked for proper installation as per NEC and verified OK. Plan was to set power analyzer to check for waveform distortion including the equipment ground.

As soon as chiller is shutoff the AM radio signal comes back to normal. Appreciate any help to narrow down the cause.

 

[PQ101]

Thanks waross.

Yes I did. Dranetz 658 power analyzer is curretly connected on 480V main to collect data. Data that you see on Chiller disconnect was captured on Dranetz Powervisa. Plan is to visit the facility after Thanksgiving when York folks return. Plan is to connect monitoring equipment at several locations to identify the cause. I can pick that up quickly. I'm hoping that this leads to cause of RFI but difficult to say right now.

No radio---When chiller starts-up basically there is so much RFI noise in the area (based on monitoring display on Radar Engineers Model 240 RFI locator) that entire parking lot area is buzzing so bad that you cant hear anything (AM reception is balanketed out for about 5-6 city blocks). No noise when the unit is shutoff-AM station nice, crisp-clear.

Plan is to check wiring connections starting from 480V Main-everything is suspect right now. Attached is the picture when I discovered someone had disconnected & taped the two ground conductors at main swgr neutral-ground bus. Ground wiring was re-connected back & tightened.

With respect to sag, I further reviewed the collected data, and there is indication that voltage drops to 250-300volts (from nominal 480volts) at no load. Pointing to the fact that something is wrong with supply voltage.



Providing power quality solutions that empowers you!!!

 

[itsmoked]

A large drop could certainly be a loose connection. A loose connection could easily be a cause of EMI. This would fit with; 'the last 100 didn't have this problem!'.

When using a radio to hunt EMI you need to attenuated its ability to receive the signal. Wrap it in a Faraday cage or use one that's sheilded. You may need to even shield the antenna some. You want it compromised so badly that it can be used as a strength meter in your strong noise environment.

If you can get its sensitivity down far enough you should be able to work your way to the source.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Marke]

Hello PQ101

I think that this is not a VFD, rather it is a soft starter.

If it is in fact a soft starter (uses 6 SCRs) in series with the supply and no DC Bus capacitors, then it would normally make a burst of noise at start up and then go pretty quiet as the SCRs should be turned hard ON.
If it is a soft starter with an energy optimising option, then the voltage is reduced when the motor is under light load.
The energy optimizing option is not worth the trouble and is best disabled, that will reduce the SCR switching noise during run.
Your description sounds like conducted noise from SCRs that are being held in partial conduction rather than full conduction.
If it is a VFD, then the speed will change as the load is changed. If it is a soft starter, it will run at constant speed.

Best regards,
Mark.

Mark Empson
L M Photonics Ltd

 

[PQ101]

Thanks.

The emitted noise is so strong that no matter which direction you point the RFI locator (or the portable radio) you will not be able to determine the source. But I understand your point.

Providing power quality solutions that empowers you!!!

 

[Skogsgurra]

Smoked, the interference is a constant whining. You do not get that from loose connections - they produce 'static'.

Mark, I have yet to see a York compressor with a soft starter. Could be, but I doubt it.

For OP: Is it a soft starter or is it a VFD?

I think that the 100 with no problem is more anectdotal than factual. Most VFDs produce EMI. But some areas are listening to AM radio and some aren't.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[PQ101]

Thanks Skogsgurra.

Chiller is equipped water cooled SCRs. It is also equipped with harmonic filter package added (inductor coil & capacitor) . I will call to verify 100%.

Providing power quality solutions that empowers you!!!

 

[Skogsgurra]

I think it would help to have the exact York compressor type.

You now, again, say that you have SCRs. Inbetween you were certain it is a VFD. That is something that we need to be certain about if we shall progress at all.

I have been looking at your voltage recordings again, events #58 - 130 11/19/2009 15:17:12 - 15:19:11, and I am quite positive that this is not an actual problem with the grid or bad connection between grid and compressor. It cannot be anything but a loose connection in the measurement set up.

The reason I say so is that you can see very fast voltage edges and slow release when falling back to reduced voltage. This is what you get when you have a loose contact or bad cable between measurement point and measurement cable. What you see is voltage discharging (typical capacitance discharge curve) and then, when contact is established again, a very fast jump up to nominal sine value.

I think you shall review the 658 connections and also try to forget that you may have a problem with the grid or connection between grid and compressor. No compressor would continue operating when subject to disturbances like these. Most (all, I would say) drives would output alarms and stop operation on the first disturbance. That is obviously not happening and that is also an indication that you have to look for other explanations to your EMI problems than bad grid or grid connections.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[waross]

Thanks for that explanation Gunnar. A bad meter connection makes sense. I missed the tid-bit about whining. That information makes my contributions much less valuable.
I apologize if I have caused any confusion.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[REDDOG]

The conduits entering the transformers load side compartment appear to be steel.
Is the required bonding from these conduits to ground missing?

 

[PQ101]

Point well taken. But I assure you with good degree of certainty that all voltage probes were tightly grabbing the lugs on all 3-phase (straight delta connection). Furthermore, the first 15-20 minutes of recording the monitor was collecting data with no voltage disturbance. As soon as load was applied (compressor disconnect switch closed and controller start button pressed ON) the voltage sagged. Prior to that voltage values maintanained well within 495volts nominal. What you see is a fraction of data becuase there are so many events but I attached only a few. All 3phase are affcted though.

Plan is to review collected data on 480V main, and compare with data collected on isolation transformer, and data collected on 480V chiller disconnect and check for any problems. Once I verify voltage stability on chiller disconnect, I can closeout the suspect on incoming service and focus on chiller end.

Providing power quality solutions that empowers you!!!

 

[PQ101]

Pictures may not be that great but someone has installed 6" inch piece of PVCs on (2) incoming conduit piping from 480V mains (right before isolation xfmr against the wall), on (2) departing conduits to the chiller and right before the flex conduit (top entry to the chiller enclosure). Plan was to bond across by placing jumper on all 4 locations and bring back to original configuration, and go from there.

Providing power quality solutions that empowers you!!!

 

[rasevskii]

For PQ101:

What with bits of PVC inserted into conduit runs, no bonding of the conduits to the padmount (as reddog says) it seems that the installation needs a complete inspection.

You say a big voltage drop on startup-does that mean the lights in the school building actually flicker? It seems that the school is fed via the second dry-type transformer at 120/208V...readings taken with the Dranetz or other digital meters can be far off due harmonics. It may be old-fashioned but try an old moving-iron voltmeter instead.

Meters such as a Simpson or AVO usually are rectifier type and will show errors on heavily distorted waveforms. By several percent.

Try direct connecting the chiller via temporary cables laid on the ground directly to the 480V board, as a test.

I still feel that there is some HF coupling between the overhead conduits and the roof or steel trusswork in the building. You may have a giant antenna operating at a low
level of HF voltage but of immense area. Maybe you can get a patent on this...

Get the power company to inspect the HV side of the padmount-are the cable shields grounded and where..

If the bldg has not a metal roof, it can be the wiring on the 120/208V side is radiating. Did they use metal conduit or non-metallic?

just some more thoughts..

regards, rasevskii

 

[PQ101]

Thanks rasevskii.

Keep in mind that equipment manufacturere had implemented 10 different line item in order to suppress the RFI, and none worked. So, I have to bring back to code as required.

The roof is sheetmetal roof. With respect to utility padmount, primary concentrics are bonded to ground rod at 500kVA service transformer, system neutral goes from the padmount to 480V swgr bus ground connection, bonding jumpers installed between swichgear GRND and incoming neutral. In previous life I was a utility power line investigator doing all sort of power investigation/RFI/TVIs, can tell you service side has a question with respect to loading but I need little more field data to prove the facts. Typically the goal is to clean things first, let the dust settle and then see what else is left to go. Appreciate the comment.

Providing power quality solutions that empowers you!!!

 

[PQ101]

Appreciate all the comments.

I just learned a few minutes ago that when a portable generator was brought in to power the 480V chiller disconnect (from 3-phase delta generator) the RFI was completly eliminated-gone(with 480V service & equipment ground disconnected from 480V main panel). Generator was then removed, isolation transformer was installed and noise was back in. It appears RFI could be related to 480V panel.


Providing power quality solutions that empowers you!!!

 

[rasevskii]

For PQ101:

A tough nut to crack, this. Looks like a lot of galvanic paths may exist between the infamous conduits and the roof.
Maybe a conn. at floor level would help...Conduits entering the top are hard to seal against water, anyway.

Perplexing...

rasevskii

 

[PQ101]

Found the bonding jumper loose on 480V Mains, connecting system neutral to ground bus, placed jumper across all PVCs and RFI is gone.

Providing power quality solutions that empowers you!!!

 

[Skogsgurra]

Does that explain your Dranetz reading?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

Thanks for the fun ride PQ101. Nice response support.

Glad you got to the bottom of it.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[PQ101]

I wanted to appreciate everyone for the comments.

First inspection on bonding jumper way in the back of the 480Vswitchgear, with safety gloves pulled on system neutral and all connections looked tight at first glance. Checking the voltage increase to 529volts and decreasing to 250-350volts I figured system unbalance is due to bad neutral-ground bonding connection (which I always get that on power monitor), so I went back to double check this time with a lug driver. The bonding jumper was less than adequately tight- about 3/4 of a turn. For some reason this year I had 5 chiller cases, but this one was a odd ball. Thanks.

Providing power quality solutions that empowers you!!!

 

[ozmosis]

thanks for coming back with the result. It's always good to get a result and know about it. We all learn in this way.