It’s been a couple of years since I first wrote this SolarEdge inverter review. It started out nice. We installed over 3000 SolarEdge optimisers, and I had a tour of the SolarEdge HQ in Israel and had meetings with some of SolarEdge’s founders. But, as time went on, significant problems started appearing so I changed this SolarEdge review to voice my concerns with SolarEdge.
Since then SolarEdge lawyers have sent me threats on two occasions about this blog.
Part one and two are a condensed version of my original review. In part one, I explain the SolarEdge inverter theoretical advantages. In part two, I pull apart SolarEdge’s newest model, the SolarEdge HD Wave.
Part three is where I address the five significant problems I have discovered with SolarEdge: failure rates, monitoring, alerts, underperformance and blocking.
The SolarEdge threats
On Christmas Eve of 2018 at 3:15 pm, I received a letter from SolarEdge lawyers. The letter demanded I remove Part Three of this post. Instead, I amended Part Three to address their concerns, and I added lots of bonus updates. Then on the day before Good Friday 2019, I received further threats from SolarEdge. I’m in the process of updating this blog to removed some, arguably cheap-shots and irrelevant arguments in order to negotiate a truce. I will update here when SolarEdge agrees to withdraw all future threats.
Part One: Why SolarEdge optimisers?
SolarEdge is a huge player in today’s solar inverter market. Compared to their competitors, they are the new kids on the block. They only kicked off in 2006 with the first inverter sales in 2010. Today they are second only to SMA for the number of residential inverters they are moving, and the way I read it, they have been successful are making money hand over fist. So what are they doing right?
To step back a bit, let’s look at one of the key components of any solar inverter: the Maximum Power Point Tracker (or MPPT). Standard string inverters have two MPPT’s. The panels on a house are divided into two groups (or strings) and connected to the two MPPT’s in the inverter. As the sun’s intensity changes throughout the day, the two MPPT’s will continually adjust the voltage and the current from the panels to achieve maximum power.
The potential problem with standard string inverters is that a string of panels works kind of like old school Christmas tree lights: if one panel is affected by shade, then all the panels in the string are affected. The SolarEdge Inverter offers one solution to this problem. It takes the MPPT’s out of the inverter and effectively puts one MPPT (or optimiser) behind every panel. So if one panel is shaded, it does not drag down the performance of the next.
I deliberately didn’t call this section “enhanced safety” or “SafeDC” as SolarEdge brand it, because that infers other solar systems are not safe. If solar systems are installed by electricians who follow today’s standards, your solar is arguably as safe as the rest of the electrical wiring in your home.
Optimizers and microinverters offer a solution that brings another level of safety. SolarEdge’s optimisers require communication with the inverter in order for them to operate. If the SolarEdge inverter is turned off, the optimised panels will produce a safe 1 volt each. This would prevent solar contributing to your house fire, and the firefighters will appreciate that you used a solar solution that enables rapid shutdown.
SolarEdge Inverter Monitoring
Every system we install today has an in-depth level of monitoring. SolarEdge is one option that allows you to monitor every single panel’s production. Seems overkill? Here’s why you may think you want individual panel monitoring.
i) To identify shading for tree growth and soiling,
ii) to identify panel degradation, (this point is dampened in Part Three under “can optimisers underperform”).
iii) to identify bypass diode failure, and,
iv) education. Because there is always more to learn about solar.
In the earlier version of this post, I rabbited on about each of those points. However, while all these features are available on SolarEdge inverter monitoring, I now question their accessibility and their reliability. I discuss that in Part Three. But for now, let’s look at the positive and check out the HD Wave Inverter.
Part Two – The revolutionary new SolarEdge inverter
In the early days, inverters used a heavy copper-wound transformer. Over the last ten years, inverters became “transformerless” which made them heaps lighter and more efficient. All inverter manufacturers today use SMA’s H5 Bridge technology – all except … the new SolarEdge inverter.
SolarEdge claims to have lifted the bar. They have replaced electrolytic capacitors with film capacitors and heavy magnetics with digital processing. There are three claims that SolarEdge makes that I’ll examine. In this updated version of my blog, I have abbreviated my original findings.
i) The HD Wave is more efficient.
In theory, I proved this to be correct in my original testing. But, read on to Part Three. It’s a bit more complicated than I understood in my original tests.
ii)The HD Wave produces less heat and requires less cooling.
Again, this was proven correct by my original testing. However, in my experience, this hasn’t necessarily translated into longevity compared to Fronius.
iii) Film capacitors last longer than electrolytic capacitors.
SolarEdge has claimed electrolytics only last an average of 10 years, and by using film capacitors they significantly increase the expected life of the inverter. I consulted leading inverter manufacturers.
An anonymous comment from a well-respected figure was pertinent:
If correctly specified and chosen, and an inverter’s internal heat dissipation properly designed and tested, electrolytic capacitors have no problem to continue to operate for a product design life of 20-25 years. But the same is of course true for ANY component within an inverter.
I contacted SMA, ABB, Sungrow and Fronius. Each had similar statements.
The ugly inbuilt DC Isolator
Design counts. The version of the SolarEdge inverter that was on display at Intersolar in Germany looked small and schmicko. Australia, however, has this stupid regulation about inbuilt DC isolators that made SolarEdge feel obliged to favour function over form.
The isolator works fine and I’m not doubting that it is ‘fit for purpose’. It just feels flimsy. You get the vibe that the engineers snuck it past the design department to get it to market in a hurry.
But good looks is the least of SolarEdges‘ problems
Part Three – The problems with SolarEdge
SolarEdge lawyers suggested some of my headlines such as this are misleading and inaccurate. I don’t agree. Having discussed some of the good aspects of SolarEdge, I now want to discuss some of the not so good aspects. Accordingly, I think the headline is fair and reflects the content of this section. I believe the content is accurate (otherwise I wouldn’t include it) and of significant interest to consumers and installers.
SolarEdge are charging ahead with their ambition to be a one-stop-shop for solar. Their product line up will soon include:
- Home Automation using the wireless ZigBee controllers
- An “Immersion Heater Controller” or a hot water diverter.
- Solar panels with integrated optimisers (made by Jinergy in China).
- A SolarEdge battery
- Electric Vehicle Chargers
- Popcorn makers (just kidding).
I could philosophise about the approach SolarEdge take to developing their business, but I’ll resist and just state some facts as I understand them.
As always, I am happy to correct or update facts if things change or further information is available.
As best as I can ascertain.
- SolarEdge manufactures nothing. They outsource their manufacturing to other companies in China (by Flextronics), Hungary and Romania. They place their own staff in these factories for Quality Control.
- Every other major solar manufacturer has chosen to collaborate under the Sunspec Alliance and enable “plug & play” system inter-operability, promote technology innovation, and accelerate Solar industry growth. SolarEdge has chosen to pull out of the Sunspec Alliance and go it alone. (Their lawyers took exception to this point as misleading. I fact-checked this point with people involved in the SunSpec Alliance and they confirmed my comments are reasonable and true.)
- SolarEdge is at the eve of offering a whole new swag of technically complex products. That is good news as I am always happy to see technological advancements (and, hopefully, improvements). However, in my opinion, after 8 years of sales, Solaredge have not yet proven to the market that their main product has a level of reliability that consumers should be entitled to expect.
It’s this reliability that is my biggest concern, so last month I went to Israel on a “business trip” to investigate.
My tour of SolarEdge Israel
I lived in the north of Israel for several months in 1999. Israel is a fascinatingly complicated country steeped in history, so when SolarEdge invited me, I initially jumped at the opportunity to get back and hang out in Jerusalem.
SolarEdge was well aware of my reservations. At one point I turned down the invitation saying that we had decided not to use SolarEdge optimisers (at least for now) because the issues we had experienced were too significant. SolarEdge convinced me to go, so I went as a sceptic willing and wanting to be convinced. There were two significant issues I wanted to address.
- SolarEdge optimiser failure rate.
- Performance issues due to design restrictions.
SolarEdge optimiser failures
Here are MC Electrical’s updated SolarEdge failures. It has been 4 years since our first install.
- 130 SolarEdge systems installed since Nov 2015
- 138 inverters installed
- 3293 optimisers installed
- 57 approved optimiser warranties
- 1 pending optimiser warranty
- At least 3 “DOA” optimisers (not counted in percentages below)
- 15 approved inverter warranties
- 1 rejected inverter warranty (SE claimed lightning with no evidence).
- 44 jobs we have returned to check and replace. Several multiple times
I’m confident the “1 pending optimiser warranty” are indeed faulty. So percentages work out as a whopping…
- 36% of our SE jobs have had at least one fault related to production.
- 1.7% optimiser failure rate
- 11% inverter failure rate
Is that bad?
Under 1 % is considered an acceptable failure rate for inverters. It could be debated if that is 1% in 5 years or 1 percent for the 12 year warranty period. Our SolarEdge inverter failure rate is eleven times that, and it’s only 4 years since our first SolarEdge install. I know installation companies that claim a far worse SolarEdge inverter failure rates than mine.
What about an acceptable optimiser failure rate? An average string inverter system has 1 part: the inverter. A common residential SolarEdge system has over 20 optimisers and an inverter. Logically, the SolarEdge failure rate should be at very least twenty times less, or .05%. That’s being generous. But we’re looking at a 1.7 per cent optimiser failure rate!!
That’s at least 34 times worse than I think it should be.
And keep in mind that replacing an optimiser isn’t as safe and easy as replacing a string inverter. It can be costly to safely get on a roof and remove multiple panels to get to the failed optimiser. While SolarEdge currently has an agreement with installers to pay a small labour component to replace faulty optimisers, their SolarEdge warranty explicitly states otherwise. I’ve asked SolarEdge to change this in their terms and conditions, but they say it’s a standard term. Indeed it is, as my Enphase Warranty document also has this as a “standard term” they are unwilling to change. That doesn’t make it right.
These optimisers are warranted to last 25 years. That’s what they call a “business liability”.
The tour of SolarEdge
During our tour of SolarEdge, we visited their R&D testing labs…..
Before publishing this post in July 2018, I sent it to SolarEdge for feedback. I gave them 4 days to respond. I told them repeatedly that I am open for corrections or omissions and I was willing to delete sensitive information. I also did the same when I reviewed SMA, ABB and Fronius. This time instead of a “thanks for the courtesy, can you change …”, I was threatened with legal action and was given very little feedback. It wasn’t until Christmas Eve 3.15pm that I got a five-page letter from SolarEdge lawyers. I was threatened the second time in the afternoon on the day before Easter holidays 2019.
While I deleted this section about the “tour of SolarEdge” their Christmas Eve legal letter still asserts that the blog still breaks a confidentiality agreement, but they did not state which part they believe is in breach.
I certainly don’t believe I have included anything in this blog which is confidential. If I did, I wouldn’t have included it, to begin with. I take those obligations seriously. If SolarEdge identifies something which they feel was told to me in confidence, I will certainly look at whether it should be removed.
SolarEdge changes their tune
Both leading up to my Israel trip and after it, I had multiple conversations with SolarEdge Australia. Their tune has now changed and they now admit they have high optimiser failure rates across the board. They made two claims that I want to discuss. Edit, their lawyers deny SolarEdge have changed their tune or that they have a high optimiser failure rate.
- SolarEdge’s optimiser problem was mainly “infant mortality”. Optimisers were failing within the first few months of install. Infant mortality usually occurs because of faulty components (like capacitors) and at the manufacturing process level rather than an inherent design fault.
- The “Generation 4” SolarEdge optimiser solution.
SolarEdge had reviewed our optimiser failures and said they were mostly infant mortality issues, which SolarEdge defines as failures within two months. Anecdotally, this didn’t seem right to me, so we checked as much data as we were able to and found that more than half of our faulty optimisers failed after 12 months. Many failed after 20 months. Contrary to what SolarEdge specifically told me, the main problem we have with optimisers is not infant mortality.
The hope of Generation 4?
In an earlier version of this blog, I referred to a “Gen 4” optimiser manufactured in February 2017. I suggested Gen 4 may have solved the reliability issues.
While I have written correspondence with SolarEdge about Generation 4 optimisers, their lawyers tell me that “SolarEdge has never marketed, nor sold, a fourth generation of its product”.
So I’ve removed any further reference to Generation 4 optimisers and until there is a new release or something else changes, any inference to hope for improvement in optimiser reliability in the short term.
Why doesn’t everyone complain about SolarEdge optimiser failures?
SolarEdge is a polarising product. While the majority of solar companies I’ve spoken to complain of numerous optimiser failures; some claim no failures.
But why do some installers claim very few failures while SolarEdge themselves admit there has been an unreasonably high failure rate. Do some installers just have good luck? I have two theories.
1. The average failure time
An interesting although relatively anecdotal observation that an industry friend recently made: the companies that defend SolarEdge reliability have often been installing them for less than two years. I could list numerous companies that have done large numbers of SolarEdge systems over a long period of time. They will no longer install SolarEdge because of the failure rates that some claim is much worse than mine.
The companies that defend SolarEdge reliability have often been installing them for less than 2 years.
2. Temperature-related optimiser failures
After talking with so many installers across Australia, it seems to me there is a strong correlation between the location of the installation and the reported optimiser failure rate. For example, a friend in Darwin has a significantly worse failure rate than mine. Another good friend in Melbourne assures me his optimiser failure rates are fairly reasonable.
The Arrhenius equation is a rule of thumb that asserts every 10°C increase in temperature reduces component life by half. While I hope it isn’t the case, if the Arrhenius equation can be applied to SolarEdge components, then my friend in Melbourne may be subject to the “Pantene Promise” :
“It won’t happen overnight but it will happen”.
3. Restricted SolarEdge inverter monitoring
My second theory relates to monitoring. The SolarEdge inverter monitoring platform has three levels of accessibility. While the full view gives individual panel monitoring and alerts, the most basic view does not allow the customer to view “alerts” or “layout”. Most companies I talk to only give the customer the most basic access when they set up the SolarEdge inverter monitoring portal. This is obviously unfortunate as panel level monitoring is half the reason to buy a SolarEdge inverter.
The Devil’s advocate
But let’s play devils advocate and suggest that as long as the installation company closely monitors all of their SolarEdge systems, a restricted customer view would work, assuming:
- The installation company remain in business for the next 25 years.
- They give a rats about their past customers – for the next 25 years
- Busyness does not get in the way of them making your failed optimiser high on their priority list – for the next 25 years.
- SolarEdge remains on their product line up and they regularly log into their monitoring platform – for the next 25 years.
- The alerts actually work for the next 25 years.
Alerts don’t always work
Let’s assume an exceptional company like MC Electrical installs your SolarEdge inverter, and we adhere to all of the above points for the next 25 years. I call it my “Platinum-Gold-Ironclad 25-year Exceptional Customer Service Guarantee”.
All we need to do is regularly check the alerts of the 2000+ system we would have installed, and repair the failures that are likely to happen.
The problem is the alerts don’t necessarily work.
Over half of our failures have been picked up because our customer looked at the individual panel monitoring, or we have happened to stumble across it. Clearly, not all customers bother to check their monitoring so we regularly check all 130 systems manually.
Can optimisers underperform?
A few months ago, a diligent customer called us insisting that some of his Sunpower panels were underperforming by around 5 per cent. Given the system was installed in the shade, it seemed like an unlikely story, but we took the time to look into it. The short of a very long story is that the panels were not underperforming nor was it a shading issue. We spent countless hours identifying that 3 SolarEdge optimisers were underperforming. SolarEdge told us they could see these optimisers were running hot. They eventually replaced all three optimisers, (albeit one by one).
We have identified several other jobs that appear to have underperforming optimisers. Some of these are pending optimiser warranties listed above.
Now think of all the companies that report a low optimiser failure rate, but don’t give their customers access to panel level monitoring. Think of all the underperforming optimisers that are almost impossible to identify. Do we really know the full extent of the problem?
SolarEdge tells me they are working on updating the alerts in their portal. If this is applied retrospectively, will we see a massive glut of optimiser warranty claims? SolarEdge lawyers did not comment on this section.
Production Loss and “Blocking”
Designing a SolarEdge system on a residential home is far more complicated than keeping the stringing rules and using the SolarEdge design tool. Consider this install we did earlier this year. This installation is in full sun, in two strings of 10 panels. However, four of the northern panels are producing significantly less than the other ten. What’s going on?
Optimisers work on a string level by increasing or decreasing the voltage on individual panels so that the current in each panel in that string is the same. To complicate this even more, the SolarEdge inverter needs optimiser voltages to add up to ≈380 volts. A P500 optimiser boosts to a maximum of 60 volts. In this example, the voltage and current figures of the red string look like this:
- The four northern panels boost the voltage to their maximum of 60 volts and 2.2 amps (≅132 watts).
- The six western panels buck the voltage to about 24 volts and 2.2 amps (≅52 Watts).
- The voltage of all the panels add up to ≅380 volts and 2.2amps.
However, if the four northern panels were not “blocked” by the 60 volts limit, they could have produced the same power as the other group of 10 panels. We worked out the production due to blocking to between $6 and $9 per quarter.
SolarEdge lawyers pointed out that this situation is rare. I looked at the “charts” and the “optimiser voltage” on the SolarEdge monitoring platform for many of our more complicated roof layouts. It takes some time. You need to select all the optimisers and zoom into day view. Winter often has the worst results.
I found many of our complicated layouts had some degree of blocking. Adjacent is one of the worst examples. (The panel stringing we used is recommended by the SolarEdge site designer.)
It gets worse
SolarEdge optimisers is a product we were recommending for shaded situations. However, if we have a string with 6 or fewer panels in the sun, the system will suffer voltage blocking at best, or it will just shut down at worst. This is even when the electrician installs to SolarEdge guidelines.
Unless a shaded SolarEdge optimiser system is very thoughtfully designed by someone experienced in the intricacies of SolarEdge optimisers, it is highly likely that performance will be restricted due to voltage blocking.
(SolarEdges lawyers commented on this section stating that I “wrongly imply that a string of 6 or less [sic] panels is permissible in accordance with SolarEdge’s installation guidelines”. To be clear, I was not saying that. I bolded “in the sun” for clarification.)
Solution 1 – Long strings
There is a simple way to solve this. In the example above, if all 20 panels were in one string, then the average panel would run at 19 volts. Lowering the average voltage of an optimiser would give it more room to boost and less chance to clip. But lowering the voltage, in turn, increases the current, and we could end up with a problem by limiting the current at the optimiser’s maximum current of 15 amps. Panel currents are fast approaching 10 amps. I’ll leave the solar installers to chew over that little mathematical nugget.
However, assuming current limiting will not occur with the design, long strings are a great idea. That’s assuming you don’t care about your 25-year SolarEdge optimiser warranty. But if you want your warranty intact, SolarEdge inverter design guidelines allow a maximum of 5.7kW of panels per string. If we exceed 5.7kW, we may put undue stress on the optimiser, and your warranty may be void.
The loophole – if SolarEdge trusts you.
Fortunately, SolarEdge trusts MC Electrical’s ability to design correctly because they gave us a waiver to install up to 25 panels in a string.
Now that’s a strange suggestion from SolarEdge because if we connected just 22 x 330W panels (keeping it to 6.6kW) in one string, our average voltage would be: 380volts/22panels =17.3volts.
If a 300W panel was running at 17.3 volts, it would need to run at: 300w/17.3v=17.3 amps.
But the maximum an optimizer can run at is 15amps. So we would just be replacing the voltage blocking problem to a current blocking problem.
Incidentally, the below SolarEdge Waiver debunks any suggestion that our high optimiser failure rates are because we don’t know how to design SolarEdge systems properly.
Solution 2 – Larger SolarEdge optimisers
A second option is to install an optimiser with a higher voltage. The P404 has a maximum voltage of 85 volts. This means it needs fewer panels operating in full sun to reach the 380 volts required by the inverter. Still, if you have shading or split orientations, you will need much longer strings than the minimum.
Either way, a combination of longer strings and larger optimisers and less shade is an ideal way of reducing optimiser blocking.
Solution 3- Smaller panels
This isn’t really a solution. It’s a problem I dressed up as a solution. The problem is that panels are getting bigger and bigger. The standard panels we install in 2019 are 330W, 350W and 360W panels.
Let’s say you choose a 350w panel and you want to install 6.48kW of panels. You can’t do a long string of 18 panels (unless SolarEdge gives you a waiver) so you do 2 strings of 9 and 10 panels. If the 9 panels are on different orientations or have significant shade issues, you may be prone to blocking. If you get some clearance stock 275W panels for a bargain, you could have 2 strings of 12. Problem solved!
A word from SolarEdge
Finally, SolarEdge insists that there is no flaw in the way they have designed their system to operate. They say that, to ensure optimum performance, any solar panel system must be thoughtfully designed and installed in accordance with the applicable manufacturer’s instructions and that SolarEdge is no different.
Additionally, they say I have relied on some specific and rare situations and that there are no inherent design defects in its general application. (I’d suggest a good example of a “general application” may be one orientation per string with minimal shade.)
As indicated previously, I have no issue with SolarEdge. If my experience demonstrated to me that their products were performing without problems arising, I would say so. If things improve, I will report that. I will continue to engage with SolarEdge on these issues in the hope of a better outcome for everyone.
The SolarEdge optimiser solution has, in theory, some technical advantages: optimisation in the shade, rapid shutdown, panel level monitoring.
The new HD Wave is a redesigned and in theory more efficient. SolarEdge believes they have solved a significant problem by replacing electrolytic with film capacitors. However, the industry does not entirely agree this problem needed to be solved. SolarEdge does not admit the optimiser failures is bad. But the failure rate I have seen is worse than bad. I also suspect the full extent of failures has not been reported as a result of restricted access to monitoring and an alert system that is not fully working.
SolarEdge has an inherent design problem in many applications because of a fixed string voltage and low SolarEdge optimiser voltage. We can sometimes solve this with longer strings, more expensive optimisers, and an experienced SolarEdge designer.