Microinverter update May 1, 2013Posted by Maury Markowitz in solar.
Tags: micro-inverters, solar power
Much to my surprise, my previous article on microinverters turns out to be the most popular one on this site. Now more that a year old, it continues to generate a stream of readers each and every day. I guess that means this is a topic of some interest out there on the interwebs.
Much has happened since I wrote that article, and much hasn’t. So here a little update for everyone on what’s been going on in the last year in the microinverter space.
When I wrote the last article on micros I didn’t know a whole lot about power conversion. Well I still don’t, but I’ve learned enough to solve a few mysteries. So let’s start with a little more tech.
The purpose of an inverter is to take the DC power from the panels (or other sources, like batteries) and turn that into standard AC power, like what’s in your wall socket. If you look at the graph on the right, we want to take the red line and make it look like the green one.
Actually what comes out of the panels is more like the grey line, because the output varies with the sunlight falling on the panels. But inverters don’t really like variable voltage, it murders their efficiency.
So modern inverters generally have two “stages”. The first applies the MPPT algorithm and tries to pull out the maximum power possible from the panels, and then converts that to some other fixed DC voltage. The second stage starts with that fixed DC voltage and then converts it to AC. Or, following the picture, we convert from the grey to the red, and then from the red to the green.
Now here’s the issue. Follow that green line through the graph as it moves up, peaks, and then comes back down to zero. At that instant in time, at zero, there is zero power coming out of the AC side of the system. But look at the red line at that point, it’s definitely not zero. So there’s power on one side of the system and not on the other.
Now there’s two ways we could deal with this. One would be to ignore it, and let that extra power dissipate into the circuit. This would work, but it puts enormous stress on the rest of the system, and it also means you’re simply throwing power in the garbage.
So what we really do is store the power, briefly, when the voltage is low. When it picks back up again on the other side of the cycle, we dump it back out. Now we’re not losing any power, and we’re also making life easier on the rest of the components.
And what do we use to repeatedly store power for brief periods of time? Capacitors. And what sort of capacitor can store a couple hundred watts or so? Electrolytical capacitors, or ecaps. And this is why ecaps are so commonly brought up when talking about microinverters.
The other takeaway point here is that we can see why optimizers are so much smaller than inverters. Inverters consist of two stages, and need capacitors to buffer power between them. Optimizers are only the DC-to-DC side (grey to red), theres no point in the process where they have zero output that they need to buffer. What little capacitance they have is just to filter out little bumps. They don’t need nearly as much energy storage, a fraction.
Check out the picture on the right (sorry about the quality, it’s the only one I could find). That’s an early model Enphase micro, probably a M210. Notice those brown cylinders on the left? Those are the ecaps.
It you take them out, you can see how much smaller the case would be. In fact, one of those two grey gizmos in the center would go too. Rearrange the rest and you can see that it would be much smaller.
It would look something like this, the inside of a Tigo. It’s not obvious, but the scale changed – do you see that brown spring-looking bit at the top? That’s the same size as one of the grey gizmos in the Enphase above. The board is about 1/3rd the size. That black area at the bottom of the picture is just wiring. It’s also much lighter; the Enphase is filled with a thermal epoxy to “pot” the components and take the heat out to the metal box. The Tigo doesn’t need this, it’s in a plastic case.
So are ecaps really an issue?
I’ve been reading a bit about whether or not ecaps really are a problem. The string inverter companies claim it is, because they say their own ecaps will only last 10 to 15 years, so how can the micros claim their’s will last longer?
So, in theory, how long will these things last? There’s a couple of different answers to this one, depending on who you ask. But generally there’s a couple of things that stress them out. One is the frequency of the power; more frequent cycles are generally easier on them. Temperature is also an issue, especially how much they heat up when they’re charged and discharged. And finally, like batteries, they don’t like to be fully charged and then fully drained repeatedly, which in capacitor terms means they don’t want to operate near their maximum design voltage.
And that last one is the bone of contention. Typical capacitors are rated around 600 to 650V. In a normal string inverter, the input is, say, 500V, which is the output of a whole string of inverters. That means the capacitors are working at 80% of their design limit, all the time. But in a micro inverter you only have one panel’s worth of power, working at perhaps 40V. So in this case the caps are running at maybe 7 or 8% of their design limit.
Depending on who’s math you use, that can completely overwhelm any other factor. It’s entirely possible that micros will far, far outlast a string inverter for this reason alone. Now a comeback here is that micros are up on the roof where it’s really hot, so if temperature really is important, that might overwhelm the positives.
It’s so hard to call it one way or the other the only solution is to test it. Which is why this paper discusses all the reasons for doing a big test and finding out once and for all.
So, onto the updates
Time for the real story. We’ll start with the obvious, canonical, example…
But in that year panels have trended upwards in power considerably. 235 and 240W panels were common when it was released, but now I’m seeing all the A-grade monos in the 265 to 270W range, and polys are anywhere from 240 to 255W. Since the M215 peaks out at 225W AC (or a little under), the M215 is simply too small for today’s panels.
And I still don’t like the Engage cabling system. The connectors are frigging huge, and scarily expensive, and come all in a string. So if you need, say, a string of eight, you need to take the big run of cable and cut it. That leaves an open end that you have to cap. In comparison to their M190’s daisy-chain system, this is a major step backwards.
But other than that, they just keep going. I have M190s on my roof, and they’ve been flawless. In fact, one of them noted low output on one of my panels that turned out to be a blown diode. If I was using a string inverter I’d have never noticed this, and lost 1/12th of my power, forever.
I found a document on the Enecsys design and now I know how it works. Basically they boost the voltage in their first stage right up to 600V. Since power is volts times amps, and amps is what you’re storing in the cap, by increasing the voltage you lower the amps and can get away with smaller caps. To improve on this, they beef up the DC-to-AC side of the system so it can handle a little more unfiltered surging from the DC side.
Put together, these changes allow them to dramatically lower the amount of capacitance they need. So they get rid of the ecaps and use the smaller, but extremely long lived film capacitors in their place. These should last for decades.
Now the downside to this approach is that it meant some power was being thrown out. When they first introduced their micros they were around 94% efficient, which was enough to compete with the Enphase M190, but certainly not the 96% efficient M215.
Well Enecsys has just started posting information on their new “second generation” product line. To start with, it runs at 96.5% efficiency, even better than Enphase. On top of that, it can handle panels all the way up to 315W, which gives them lots of room for future growth in panel ratings.
And that higher efficiency and lack of ecaps makes for a cooler running inverter. The new ones come in a plastic case. That might not seem like anything important, but this generally means you don’t have to ground them. Depending on the installation, this can save you time and money.
And if that’s not enough, they’re field-upgradable to supply any grid standard. In theory this means they can make a single inverter for anywhere in the world, and then just patch it in software to work wherever it ends up being used. This could have significant effects on the economics of making these, so I’m curious to see what happens to the price.
One thing that’s curious is that they appear to be releasing two models . One is a 240W nominal (for panels up to 275W) and the other is 300W nominal (315W tops). This seems to fly in the face of good practice, so I’m going to try to find out why they’ve done this. I suspect it’s simply component costs.
But generally I’m pretty impressed; if they can get this to market I’ll definitely be wanting to try it out!
Self-launching any product is difficult, and when that’s going up against an incumbent like Enphase it’s all the harder. So I’m happy to report that Sparq has indeed launched, and is being sold into the Canadian market. Heliene, a panel manufacturer here in Ontario, is selling them pre-installed on their panels to make an AC panel. Apparently these are selling OK out west.
Beyond that, though, I have little to report. The model being sold is the 215W output, which can draw only 230W from the panels. Basically the same issue here as the Enphase, and it’s not clear if they have any plans to address this.
But with that exception it’s all good. Their cabling system is daisy-chained like the original M190’s, a major advantage over Enphase and, to a lesser extent, Enecsys. The monitoring can work offline or on, so if you’re installing in a remote location without ‘net access, no problem. This is a major sore point for the other systems.
Ahhh, SMA, the company that loves to hate its own products. What’s that SMA? You still haven’t released your micro-inverter? You mean you’ve set four different release dates in the last three years and failed to actually ship on all four? Wow.
The cabling system is better than anyone’s (except maybe Enecsys’ new one, haven’t seen it yet), it’s ecap free, and handles 240W. That’s all good. What’s better is that you can mix and match their micros with their strings, which is a super-big advantage where you have lots of panels in one place and a few in another you need to wire in. And their monitoring works across strings and micros, which is definitely cool.
But come on, is there any reason to believe they’re ever going to actually release these things? Well, maybe one reason…
Power-One has been eating SMA’s lunch just about everywhere. Their three-phase string inverters ripped a whole in the market, especially up here in Canada where they were the first to support our oddball 600V 3-phase standard. For some reason most of the other companies still don’t have a competitive 600V model, including SMA.
They had been talking up their Aurora Micro for some time, but it seemed they too were slipping into the indefinite future, SMA-style. But then they actually shipped. And it’s a pretty great product. It too comes in two versions, a 250W and a 300W. They peak out at a very respectable 96% CEC, 96.5% max, which puts them in the ballpark of their competition.
One interesting feature is a very high operating voltage on the panel side, up to 65V. This makes it compatible with both the common 60-cell panels, as well as the larger 72-cell models, and even some of the new 96-cell ones. This is something that I’d like to see the other company’s address, but for now it looks like Power-One is the only game in town for the larger, higher voltage panels.
It also has a lower operating voltage, down to 12V, but it’s not clear to me there’s any advantage here. My panel that blew a diode has a lower voltage, around 20V, so you might thing that this model would keep it running. But the problem is that it needs 25V to start up, so no help there, that’s the same as everyone else.
The only real downside is the cabling, which is a “trunk cable” system like Enphase’s Engage, which is just bad all around. A more minor issue is the wireless monitoring, which I’ve always been uneasy with. And it comes with a surprisingly short 10 year warranty, compared to 20 for Enecsys and 25 for Enphase and Sparq.
There’s no real news to report from Tigo. They switched their production from their original model to one with two optimizers in a single case, but that’s all they seemed to change. And that’s it!
SolarEdge is the hybrid system that combines smart optimizers with a dumb inverter, isolating the circuits so that only what’s needed appears in either place. The result is a system that has almost all of the advantages of the micros, with all of the cost performance of a string inverter system.
The only real downside to the SolarEdge approach is that they have fixed-size inverters. The nice thing about micros is that a single model can be used with any system from one to a million panels. A single SKU and some cables and you’re ready to install anything. With SolarEdge (and Tigo or any other optimizer) you need to have the right-sized inverter for the system. And that means you have to stock a number of different models, which drives costs into product all along the line.
That’s in theory. In practice, they’re less expensive than either micros or anyone else’s string plus optimizers, which seems like a little bit of magic. Built in monitoring without an external box? Does wireless monitoring just by plugging in a little card? Uses standard MC4 connector based cables? It’s all good.
And that’s why they’ve been tearing a hole through the local Ontario market. As much as I like micros, I think this will be my next install.
Well that’s it for this update. Hopefully I’ll have more to report on the new Enecsys kit in the near future, and who knows, maybe SMA will actually release something!
In the meantime, I’m going to write a little on the whole AC panel concept, which I think is dumb, dumb, dumb.