Line noise crashes Solaredge inverter

[jay2p]

We have a 5 year old Solaredge inverter on our house supporting 38 panels that at peak could maybe deliver 12 kW. One day my currentcost power generated monitor showed 20 W in full sunlight. Obviously something was wrong. I went to the inverter, which is outside the house and it had thrown an x24 or similar error. I called Solaredge CSR (sounded like they are in the Philippines) and they told me that it was due to "Total Harmonic Distortion". I know something of this issue with audio, but I was puzzled and wondered whether some local oscillator in the inverter was going wrong or something else, but did not immediately have an answer. I was shocked when the 5 minute countdown on the inverter timed out, the relays and fan kicked in and the choke (I initially thought it was a transformer) started buzzing like crazy.

It continued to buzz and rattle, for a few seconds and turned off the inverter again, threw the DC injection x24 error and started timing down from 5 minutes again.

After hours on the phone with customer service trying to find out what the problem was (they told me that this was not a customer solved issue, but I would need to bring in my installer) I got them to promise to send me the allowed THD specs. They did not, but my installer was able to connect to a level 2 tech in California who said that the problem was that something was generating noise, That noise either came from a bad connection or something else in the house. I could not believe this, but because the level 2 guy seemed to know what he was talking about, we entered into an email discussion. I challenged him that the inverter must have some kind of phase lock circuitry that locks to the line frequency and a local oscillator that drives the inverter, but he was clear that there was no local oscillator. Previously the offshore support suggested that we reinstall the "digital board" which looks like it has the signal processing for the inverter which I was hopeful for, but the idea was somehow nixed.

In the end we had to (a) Figure out if there was any line noise being generated by a bad pole transformer connection, whether there was a bad tie in to the house or whether there was something in the house that was generating enough noise to cause the problem. Frankly I did not believe that anything in the house could generate enough noise on the line to cause the overexcited humming of the inverter but there was not much left to do except (a) Test the inverter when the house circuits were disconnected but the panel breaker was closed- thus identifying whether the noise was from the house and not the external feed to the house, nor the connections from the panels etc (b) Turn on every device until it crashed the inverter.

The level 2 CSR send an email of an image of one of the 60 Hz cycles tainted with noise on my inverter... So clearly something was up. I did not know they could monitor it remotely.
So we turned off about 40 odd breakers (duhh) in the house and turned them on separately. I have a room with 3 computers one of which is a 27 inch imac from about 2012 and the moment I turned it on, it crashed the inverter. I tried this twice with the same result.
I listened to the device as it turns on - with a number of fans blowing and heard a slight background buzzing noise which was really more of a low level buzzing/sparking noise. Obviously this was the source so I took it offline and all has been well in Denmark as the bard did not say.

My lessons are

(a) Get a good installer who will stand by you. I did and that helped to deal with Solaredge and someone who knew what he was talking about
(b) Yes Virginia, it is possible for a home device to actually create enough electrical noise to kill and inverter
(c) Dont give up and find an authoritative individual you can trust

But honestly, I was really surprised when the 300 W power supply which ran the iMac no problem could generate enough noise to kill an inverter. Of course if you are modulating the entire panel output with noise and injecting it into the grid that's not good. I thought I would share this with folks on this forum as you guys may find yourselves in a similar situation

[peakbagger]

Hard to diagnose power quality problems from a distance. I have been involved on occasion with odd issues that have ultimately been solved with the same tool, something called a power quality analyzer. The problem is they are not cheap and do require set up and to someone who can review the data and make sense of it. When there is an issue, the usual approach is blame the power company. If folks complain enough, they will send a tech out and hook up a power quality analyzer and let it run for few days. Once they have the data, they then can start to chase down the cause, sometimes it is on their distribution system and sometimes its in the home. Sometimes it is in the neighbor's house down the road that you may share a transformer with and sometimes is at a utility substation. Without power quality data in hand, its just a game of hide and seek.

The problem is, who owns the problem and who needs to expend their resources to track if down when they may or may not own it?. Since it was a problem in your home, should you pay Solar Edge or the installer to do the diagnosis now that its fixed?. I seriously doubt the inverter was successfully pumping a distorted waveform into the grid, my guess is the noise the inverter was making was the equivalent of trying to use a feather to move a boulder. If the grid power was dirty and going out of range it would just shut down and then go through repeated cycles of 5 minute countdowns to try to resynch with the grid.

Despite a power quality analyzer being an essential tool for tracking things down, it is pretty rare for a small electrical contractor to own one and even rarer if they have someone who knows how to use it. It is usually the domain of engineers and we do not come cheap. When I needed one to support a project, I had to "rent" one from my company and have some other party to bill it to. I would then need to hook it up and configure it, almost always with assistance of an electrician as the analyzers are usually clamped onto the main if there was room, then someone would need to come back in few days, disconnect the device and download the data. Once I had the data in hand I would usually need to spend a couple of hours analyzing the data. If you get out if for less than $3000 you are doing well.

BTW the good old walwart transformers used for power supplies usually were pretty clean but the switching power supplies that have replaced them may be more efficient but can be very dirty and the problem is getting worse as cheap crap from the third world built to no known standard is getting into homes.

[jay2p]

Thanks peakbagger for your reply and for the information about the power quality analyzer are very useful thank you. I agree that when the sh hits the fn that the homeowner is left to figure this out. The inverter company says its not their problem and finding someone at the local utility who would even know where to start would be a challenge IMHO. As I mentioned, our installer got us to resolution because he was able to get 2nd level tech support. Before that dealing with people who really did not know or understand the technical issues had a very poor ROI 2nd level support at Solaredge can connect to the inverter and take recordings of the analog output of the inverter. They sent a waveform back to me which clearly showed distortions that were spikes in the 60 Hz waveform that were about 5-10% of the p-p size of the trace. I would say they were about 100X line frequency. Having the CSR's use terms like Total Harmonic Distortion, made me think that we were getting harmonics from a clipped sine wave instead of noise. They really should have said noise and I would have figured it out sooner.

When you said you spent a couple of hours analyzing the data, what would you actually do. Are you looking at the waveforms cycle by cycle or looking at the spectrurm ?



I wonder what you think about the following

(1) Get a stepdown transformer to about 24 v or less
(2) Build a resistor voltage divider so that it is compatible with a line input
(3) Record the line on an analog to digital converter
(4) Feed the recording to something like Audacity which can both display and analyze the spectral distribution

[bcroe]

If your 2012 imac ia suddenly generating bad line noise, it may
be in the process of failing, needs a new supply. Those switching
supplies can save a lot of power (using many to that end here),
and even adjust to 120 or 240 VAC input. But they also have the
potential of causing line noise. Sounds like it could be seen on a
common O scope, easier with a storage mode. Monitoring it thru
a stepdown trans could filter out the signal you are trying to see.
Typical probes are rated 600V peak.

Some low frequency big iron might be able to keep the imac noise
from getting out on the line. Here have some (heavy) SOLA voltage
regulator units which might do the job, some have a completely
isolated output. good luck, Bruce Roe

[peakbagger]

I am Mechanical Engineer so I look to see if any of the various inputs are out of range and usually look to see if the issues are repeating and also look at the time of day. With industrial systems, modern breaker relays have fairly extensive data recorders and buffers to store the data. If there is grid event a tech can go in and retreive the operating data prior to the event, and during the event and also get first out data.

[jay2p]

bcroe. You are exactly correct. As it happens, this iMac was left at our local dump so I picked it up and found the power supply was dead. $35 or so later and a visit to ebay replaced the supply and all was well for several years. Without wasting a lot of time, the replacement of the existing dead HDD resulted in a thermal sensor which was no longer connected. So I had to install fan control sw because the lack of a sensor was causing the fans to turn on full blast. That sometimes masked the buzz/crackle of the power supply which was now dying. I think it is probably a bad capacitor in the HV stage, but frankly I dont think it is worth reviving this iMac. But that was definitely the source of the problem.

I understand the point you are making about the frequency and the inability to capture - and that makes sense. My main reason was to post this info in case others are seeing similar problems with their inverters and also to see what others had to say, because I was completely taken by surprise when I discovered the source of noise and that it could interfere so much with our inverter.

peakbagger - I wish my breakers had relays on them and I could log the noise. Can they be purchased for a standard 200A panel ? Also it is very clear to me that you are a professional because you appear to have clear methodologies to fix things.


Thank you both for your time. Jay2p

[peakbagger]

I am not a relay or power quality pro. I was an owner engineer for several small combined heat and power plants and prior to that a project engineer on various side project so I had to understand the basis of how plants have to interact with the grid. I have also had to sit through several interconnect procedures to prove to the utility that the protective relaying did what it was supposed to do. These events inevitably happen overnight on a weekend.

I am not familiar with residential grade power quality analyzer but my guess someone with a an oscilloscope would have seen the noise from the computer power supply on the line. SEL will gladly sell you a SEL 751 relay but its kind of useless unless you know someone that can confugure it and access the internal data buffers. SEL-751 Feeder Protection Relay | Schweitzer Engineering Laboratories (selinc.com)