Harmonic Distortion and Noise

Status
Not open for further replies.
P

PWDickerson

Senior Member
Location
Clinton, WA
Occupation
Solar Contractor
This is a weird one!

The installation:

Residential installation on a barn, 320A service, 120/240 split phase. Grid-tied PV, no batteries.
25 kva pad-mount utility xformer 110' from service equipment
Interconnect is line-side connection between meter and 200 amp main breaker in one of two main panels.
Conductors between interconnection point, fused disconnect, and production meter are #2 CU, about 10' in length total.
Conductors between AC combiner panel and fused disconnect are #1/0 AL, about 120' in length
We have two SMA SB7700US-TL-22 inverters adjacent to the AC combiner panel, combined AC max output is 64 amps, 15.36 kW
33 solar modules to inverter #1, 32 to inverter #2, 280-watt 60 cell modules, 5 strings of 11 and 1 string of 10

The system has been installed for about 10 weeks. Under high production conditions, the system generates a high pitched noise that is really obnoxious. This is not fan noise, it is an electronic noise generated at the interior of the inverters. When the inverters are making the noise, the same noise can also be heard at the service equipment and the utility transformer, though not as loud as at the inverters. The noise does not happen when the system is generating less than about 11-12 kW. When the noise happens, it comes on quickly and stays on. If either of the two inverters is turned off, the noise instantly goes away.

It sounded like high frequency harmonic distortion to me, so I borrowed an o-scope from a friend to check out the AC wave form. It is a 2-channel scope, so I was able to monitor both legs of AC power. I put the test leads at the bus in the AC combiner. The first photo is of the AC output waveform with only one inverter operating (no noise). Pics 2 and 3 are with the noise happening with both inverters running. The two photos are at 2 different time scales. It shows a wicked 5 khz harmonic on the primary 60 hz waveform. I would not have thought something like this was even possible. The peak-to-peak voltage of the harmonic is about 175 volts! Any ideas what might be causing this.

SMA tech support is scratching their heads. They initially thought might be an undersized utility transformer, but 25 kva should be adequate. They have sent us some data logging equipment so they can have something to analyze, but we have not received/installed it yet.
 

Attachments

  • IMG_7293.jpg
    IMG_7293.jpg
    139.1 KB · Views: 1
  • IMG_7295.jpg
    IMG_7295.jpg
    140.6 KB · Views: 1
  • IMG_7297.jpg
    IMG_7297.jpg
    139.3 KB · Views: 1
1. Possibly the switching frequency of one of the two inverters is out of spec and the 5kHz is a beat frequency between the switching frequency of the two inverters?

2. The high level of the current switching transients on the AC line might be causing a voltage transient too which is somehow causing an oscillation of the control circuitry of the two inverters as they try to compensate for each other's presence?
 
Would need to examine the internal schematic of the inverters to be certain, but appears that at high load the EMI/RFI filters resonate with a harmonic, causing amplification of the harmonics.

Possibly the inductance of the EMI filter(s) drop slightly at high load bringing the L&C into a resonant condition or frequency content of the PWM changes (some PWM systems are designed as such) . AM unfamiliar with your particular model inverters.
 
It's a two channel 'scope. The two live conductors of presumably a 120-0-120 arrangement are monitored. Where was the common connected?
 
Has anyone else run into a situation where two grid-tied inverters interfere with each other and cause this kind of problem? We have installed hundreds of systems with multiple inverters and have never seen something like this happen.
 
Have you used an IR camera to look for high resistance connections? That might be worth a try given that the problem could be a function of power level.
 
It sounds like feedback loop noise between the inverters. Since these inverters are not isolated you can get weird interactions between them if current loops develop. I would check the following:

  1. Make sure all the ground connections are good. If the ground on one inverter is good and on the other not so good then you can get feedback through ground loops.
  2. Make sure the neutral and line connections are good for the same reason as the ground connections. One bad connection can cause a loop.
  3. Double check that all the software settings on the two inverters are the same. Every once in a while something can get tweaked and you only find out in multiple inverter setups.
  4. While it's not supposed to matter, making sure both inverter L1 connections go to the same phase might make a difference.
  5. If nothing else works putting an isolation transformer on one of the inverters will most likely fix the problem since that will break any loops.

Let us know how you resolve this problem. It sounds like it will be instructive. Good luck.
 
We went back and checked every single wire termination to make sure it was tight. We also observed all of the system components, wire terminations, and wiring with a thermal imager during peak output. No hot spots. I suppose we could try swapping L1 and L2 in one of the inverters to see if that makes a difference.
 
PWDickerson said:
We went back and checked every single wire termination to make sure it was tight. We also observed all of the system components, wire terminations, and wiring with a thermal imager during peak output. No hot spots. I suppose we could try swapping L1 and L2 in one of the inverters to see if that makes a difference.
Click to expand...
Did this work?

Sent from my SAMSUNG-SM-G900A using Tapatalk
 
No, it didn't work. We swapped L1 and L2 on one of the inverters and no change.

We did some other troubleshooting that yielded some rather interesting results though. We were on site when the inverters were generating about 85% of their rated capacity, and they were not creating the interference. There were a few lights on and a couple small fans, but no real loads in the building. We tried turning off the loads and opening the main breaker. When we did that, the interference and noise immediately start up. The same thing happened when we shut down a sub-panel at the site. Keep in mind that this is a line-side connected system, so shutting down either the main or the sub panel will not shut down the PV system.

Could it be that the the building wire has a capacitive property that prevents the two inverters from misbehaving?

We have been recording data for a few days and will be sending it in later this week to SMA to evaluate. I will post more as we learn more.
 
PWDickerson said:
No, it didn't work. We swapped L1 and L2 on one of the inverters and no change.

We did some other troubleshooting that yielded some rather interesting results though. We were on site when the inverters were generating about 85% of their rated capacity, and they were not creating the interference. There were a few lights on and a couple small fans, but no real loads in the building. We tried turning off the loads and opening the main breaker. When we did that, the interference and noise immediately start up. The same thing happened when we shut down a sub-panel at the site. Keep in mind that this is a line-side connected system, so shutting down either the main or the sub panel will not shut down the PV system.

Could it be that the the building wire has a capacitive property that prevents the two inverters from misbehaving?

We have been recording data for a few days and will be sending it in later this week to SMA to evaluate. I will post more as we learn more.
Click to expand...
How about some unwanted reactance (like a PFC capacitor bank that really should not be there) on the POCO side?
That problematic reactance could cause a resonant condition at the inverter output which is damped by the resistive characteristics of the building loads when they are present.

If you had mentioned this in the original post out speculations might have gone in an entirely different direction.
 
Here is a white paper describing your issue

What is the power factor with and without your residential load? I would ditch the scope for a power quality analyzer.

Your inverter choppers run at...? I'm guessing about 5 kHz, which would mean that without a load they free run. Does your arrangement utilize internal or external freewheeling diodes? If so, have they been tested? These are part of some designs to prevent...well...freewheeling. A small current limiting reactor arrangement might also keep the whine down.

EDIT: And another good article
 
Last edited:
Status
Not open for further replies.
Top