Optimizing optimizers, Tigo vs. SolarEdge August 1, 2013Posted by Maury Markowitz in solar.
Tags: micro-inverters, optimizers, solar power
I’ve touched on inverter concepts in the past, but I’ve only talked about “optimizers” in passing. I see lots of questions about these on the ‘net, so I think it’s time for a little intro to the two market leaders in this space. Those happen to be Tigo, the incumbent here in North America, and SolarEdge, a powerhouse in Europe. They both take different approaches to the task, each with its own ups and downs, and I’ll try to cover these in this article.
The basics of panels
Solar panels consist of a bunch of individual solar cells wired in series. Each of those cells behaves like two different electrical devices at the same time, a battery-like device that’s putting out DC power at 0.5 volts, and a very small of resistor that’s sucking some of that power back up. Each of these processes, production and resistance, is dependant on the amount of light.
In bright sunlight the battery is putting out much more than the resistor is sucking, so cells produce net output power. But in shading, or at night, the resistor side of things overwhelms the production – not entirely surprising.
When you wire things in series, they “add up” the voltages and resistances. So when you wire all of these cells together each of those cell’s 0.5 V gets added together. Modern panels normally have 60 cells, so they operate around 30V (actually closer to 35V) in normal conditions. But the resistances also add up too.
When you build a solar power system you normally have multiple panels wired together. The plugs on the panels are designed to be connected in series, so each time you add a panel the voltage goes up – 35V with a single panel, 70V with two, and so on. The cables and connectors are normally limited to either 600V or 1000V depending on where you are, so it’s very common to see strings of 12 to 20 panels. When you wire like this, it’s like you have a single bigger panel, with 720 cells instead of 60.
So here’s the problem… imagine you have a string of panels and the one on the left gets shaded by a tree. So that panel is putting out less power, but it’s also got increased resistance. And since every electron from every cell has to go through every panel on it’s way out to the world, that single panel is sucking up power from the entire array.
This is bad.
One solution would be to wire the panels in parallel, like modern Christmas tree lights, the ones that don’t go out when one bulb fails. In this case, if a panel was shaded, or broke, it wouldn’t have any effect on the string as a whole – the power from the other panels would “go around” that panel, not through it.
But this runs smack into a very practical problem. When you wire in series the voltage goes up and the amps stay the same, but when you wire in parallel the volts stay the same and the amps go up. Any wire is limited by the amperage, or current, it can carry. The 14 gauge wires in your home are limited to about 20 amps, so we put 15 amp breakers on them to be safe. A typical panel might put out 6 to 8 amps, so with that wire you could only put two panels together. If you want to collect the power from a string of panels in parallel, you need honking big cables. And copper is expensive these days.
So they don’t do that. We wire in series and live with the downsides.
The basics of inverters
Every “grid interactive” PV system looks the same when viewed from far enough – there’s panels, an inverter, and the grid. When I visualize these things I always imagine systems with panels on the left and the grid on the right, and the inverters sitting between them in the middle. Most of the heavy lifting is in the inverter. Lots of magic has to happen in the middle.
Basically, every inverter system has three parts. On one side (the left) are the panels that are connected to the inverter. In order to get the most power out of them, the inverter has to use a system called “Maximum Power Point Tracking”, or MPPT. After that is a DC-to-DC converter that takes the output from the panels and turns it into a fixed output voltage, and finally there’s the DC-to-AC converter, the only part rightfully called an inverter, that takes the fixed DC output and turns it into AC power for the grid.
There’s no reason that all three parts have to be in a single box, although most systems do just that. Actually, there is a very strong argument for breaking them up.
Breaking up is easy to do
Recall that the output from a panel changes with the light. That’s the whole idea of the MPPT, to pull out the maximum amount of power for any given conditions. And since the lighting changes over time, we keep changing the MPPT settings as fast as we can, ideally hundreds of times a second.
But those conditions don’t just change over time, they also change from panel to panel. Imagine that one panel in a string is shaded, so its resistance is higher and its voltage is lower than the rest of the panels. The inverter at the end of the string sees the whole string as a single big panel, so it adjusts it’s MPPT to suit the system as a whole. But what that really means is that all of the panels are no longer on their MPPT point… the shaded panel is being asked to operate “too high” and the rest are being asked to operate “too low”.
So ideally, we’d want to put an MPPT on every panel – or every cell if that was possible.
This is the whole idea of the micro-inverter. By putting individual inverters on individual panels, you isolate each panel and get individual MPPT. And since you’re boosting the voltage from 30 to 240V, the wire losses aren’t such a problem, so it’s OK to use parallel wiring. Now if one panel fails, no big deal.
But think about it for a second… if the part we want on the panel is the MPPT, why put the entire inverter on the panel? Why not just the MPPT itself?
And that is the optimizer concept.
Tigo Energy of California is one of the big players in the optimizer market. They make a very small box that clamps to the frame of the panel and optimizes that panel. Actually their latest versions have two of these optimizers in a single box, which reduces the relative cost of the box – for a $95 item that’s a serious consideration. Using Tigo’s couldn’t be easier, you simply connect the panels to the Tigos, and then wire the panels together like any other array. The output from the strings is fed into any normal string inverter.
The Tigo uses a very clever concept known as “impedance matching“, which helps improve the collection of power from the array as a whole. This may sound counter-intuitive, but I’ll try my best here…
Imagine a string of panels with one shaded one in the middle. That panel not only produces less power, but also slows down the flow of current from all of the panels. So now the inverter at the end sees a certain voltage and current, and adjusts the MPPT trying to get those two values to their maximum. The problem is that with that extra resistance in the circuit, the point it will select will be wrong for every panel – too “high” for the shaded panel, and too “low” for the unshaded ones.
In many cases you can improve the total collection by turning that panel off entirely. Now the inverter’s MPPT sees only the good panels, and tunes them properly. But what happens if you have two shaded panels, or four? At some point you can’t just turn them off any more, the losses will be too great.
So what Tigos do is add more resistance to the circuit… that’s right, more. But they add this in parallel to the individual panels. When you do this, you end up with two paths for the power to take, one through the panel and one around it. Now you might think that you just want to add zero resistance in the around path, but that’s not right – at a minimum you want to add the resistance that the panel would have when it’s operating perfectly, small but not zero. But you can actually tune this even further, if you carefully control the resistance at every panel you can maximize the amount of power being taken from the array as a whole.
The amazing thing about all of this is that the Tigo does all of this with nothing more than some clever programmable inductors. There’s very little circuity in the panel-side boxes.
Since the system has to work at the string level, someone has to be in overall control. Tigo has another box, the MMU, which can see all of the optimizers and then send them instructions on what to do. They do this using a wireless connection, through a box known as the Gateway. Is this starting to sound a little more complex now? Another issue is that the wireless connection isn’t exactly very high performance, so the Tigo’s are only sent updates on what they should be doing every so often. This makes the system unable to deal with individual clouds or similar temporary effects.
Nor does Tigo help with the Christmas Light Effect – if one of the panels goes down the Tigo can cut it out of the string, but if one of the Tigos goes down… the string is down.
SolarEdge is out of Israel, and while they’re not as well known as Tigo here in North America, they’ve been very successful in Europe.
Like the Tigo, the heart of the SolarEdge system is a small box that goes with the panel, but their box is slightly larger and heavier than the Tigo, so it clips to the mounting rails rather than the panel frame. Mechanically, that’s about the only difference.
Electrically, they couldn’t be more different.
The idea behind SolarEdge is to split the traditional inverter into two parts. One is the DC-to-AC stage which they put in a box that goes on your wall. The other two stages, the MPPT and DC-to-DC, they put on the roof. So it’s basically 2/3rds of a micro-inverter.
Now how does this improve things over a normal micro-inverter? I touched on this in a previous article – basically the hard part of inversion is the last stage, DC-to-AC, because it needs all sorts of energy storage in the form of capacitors. So if you leave that out of the rooftop portion of the system, you’re left with the MPPT and DC-to-DC, which is a lot smaller and more reliable. Meanwhile the “bad” part, the DC-to-AC with all its capacitors, goes on the wall, where it’s easy to service. This also provides a convenient place for all the associated bits and pieces, like monitoring and communications.
SolarEdge boxes are true MPPTs, so each panel is individually tuned to it’s MPPT point, all the time. The system then controls the output so that the string as a whole is always outputting 350Vdc. This simplifies the inverter at the end of the string, because it’s always doing 350Vdc to 240Vac (or 220/230 in Europe, of course).
So which approach is better? In pure theoretical terms, the SolarEdge is going to get more power out of the system, it’s running MPPT on every panel rather than the array as a whole. But in practical terms, they get this by adding more circuitry on the roof, so the cost should be higher for the system as whole. They can offset this slightly because they don’t have a full string inverter at the end, their inverter is greatly simplified because it doesn’t have the MPPT and DC-to-DC stages. Tigo uses a normal, complete, inverter.
Tigo actually dismisses the SolarEdge approach in their white paper. If you look on page 5, the second section is saying that if you put a DC-to-DC on the roof, like SolarEdge, you’ll lose about 2 to 3% of the power. But read it carefully… they’re talking about an additional stage. SolarEdge removes the DC-to-DC in the inverter so there’s no “additional”. From the numbers I’ve seen, SolarEdge trashes Tigo-based systems in overall production and efficiency terms.
And here’s the thing… in every scenario I’ve run, the SolarEdge system is also cheaper. Here, try going to this site and using the search bar a bit. Let’s say you’re building a 10 kW system with 40 panels and two 5 kW inverters.
Ok, two Power-One 5k’s are 2x$2222.33 = $4,444.66 = 44 cents a watt. Now add 20 Tigo dualies, 20x$92.40 = $1,848 and a Gateway/MMU kit at $335.89 = $336 + $1,848 + $4,445 = $ 6,819, or 68 cents a watt.
Ok, now compare that to SolarEdge, where I need two 5k inverters at 2×1,608.41 = $3216.82 and the optimizers at 40x$71.63= $2,865.20. $2,865 + $3,216 = $6,081, or 61 cents a watt.
I’ve tried all sorts of scenarios large and small, and SolarEdge always seems to come out on top, either because of the free monitoring, or the low cost of their simplified inverters. And I should point out, those are retail prices.
So it seems that Tigo remains an excellent solution where the inverter already exists, like adding on to an existing array or when you’re using central inverters. But for all of the small and medium sized systems that are going up fresh, if module-level optimization is a requirement, SolarEdge seems like the way to go.
Update: I feel I should also point out that in spite of all the advantages of optimizers versus microinverters, micros still have one enormous advantage. In any optimizer system the output eventually runs into an inverter, and that inverter needs to be sized to the array as a whole. That means you need to have all sorts of different inverters to hit different system sizes. In comparison, a single micro can be used on any system, from a single panel into the megawatts. This design-time simplification should not be underestimated. That said, the PV market is ridiculously price sensitive (to its detriment, IMHO), and it seems the lower costs of the optimizer approach will win many designs in spite of any theoretical disadvantage.