Why does shading over one solar panel reduce the output of the whole string?

[Siwastaja]

I don't think mental modeling / understanding of the panel without bypass diodes does any good, because the diodes will practically be there for a very good reason. Unless you really want in-depth understanding of the semiconductor physics.

This is like "what if your car engine missed one spark plug, how would it perform?" when considering fuel economy of different driving styles, or comparing different cars.

[Berni]

Microinverters per panel are not a very good idea.

The job of the inverter at the end of the chain is a lot more involved than these optimizer modules. The inverter is a isolated DC/AC converter, so it uses much more advanced typologies with enough safety to be directly connected to mains.  It needs to survive the surges it might see and must operate at much higher voltages, at the same time the single panel would provide a fairly low voltage so a lot of steeping up would have to take place on a per panel basis. There is also more smarts in the inverters as they need to monitor and sync to the grid while also reporting status back to a command and control center (aint nobody got time to go check a little LCD on it every day). Since they work at 50/60Hz a significantly higher amount of energy storage needs to happen inside the inverter, tho this is often partially mitigated by using 3 phase mains. However 3 phase is not the solution for microinverters as you need a lot more wiring and complexity to do this. Already running single phase mains to each module would be significantly more complicated than just series connection.

Optimizer modules at the panels minimizes the extra complexity to the minimum required for doing a panel by panel MPPT. They work at low voltages and do not require any isolation, making the switchmode topology inside significantly simpler while also dealing with only DC (So almost no energy storage required). These topologies have fewer parts, are smaller and run at very high conversion efficiency (reduces part cost, material cost and heatsinking requierments) To work they require the minimal of microcontroller brains while also providing insight into individual panel performance as a free bonus. At very low panel output they can easily let all the power pass straight trough by switching the appropriate transistors in the DC/DC converter. This wastes very little power, but across the whole string enough power might be collected for the big boss inverter on the end to harvest some useful power. Often a single optimizer module might handle multiple panels too, so you don't need quite that many of them. These modules also make it practical to have much longer series chains, this allows higher voltages to be used on the system level, this reduces the already lower copper wiring requirement by allowing thinner cables to be used with less loss. Sure this means more voltage to deal with for the inverter on the end, but it is already dealing with 400V AC mains anyway, so 1000V DC is not that far off.

That being said optimizer modules are not default best solution for everyone. They do provide an extra layer of cost and complexity that might not always be justified in the particular solar installation. Going for simplicity the oldschool long chain of panels with a single big boy inverter is the most simple you can do. If you have an array of new identical matched panels that never gets partially shaded, then that simple setup is all you need, you won't see any extra energy production from installing these optimizer modules.

[bdunham7]

* Microinverter is integrated with the panel,
* Output is with weather-proof, safe AC plug anyone can plug in (no electrician needed legally),
* Pre-made wiring harnesses are made in different shapes, so that electrician only does one connection in house fuse box.

These things already exist in practical form, although the second is not likely to be recognized legally in many jurisdictions as they simply won't allow user installs without a permit and inspection.  My microinverter installation doesn't have integrated micros and I'm not sure I agree with that being an unequivocally good thing in the long term unless the 'integrated' module is separately replaceable.  The premade AC wiring harness in various configurations is what I used 10+ years ago, although it is usually a bit more complicated than just connecting it to a breaker.

I realize that costs and regulations are different now and also different in other places, but micros were economically the best plan for me 10 years ago even without considering the safety, detailed data and reliability issues. 

One thing you about your earlier post:  If you have panels that are lightly shaded or just dirtier than others--say 5 of your panels are putting out 200W and 5 of them the full 300--a plain string inverter will have to either draw full current and thus bypass the 5 lower-power panels and produce 1500W or reduce the current and produce 2000W.  And then it has to be smart enough (Global MPPT) to find the correct solution.  Microinverters or optimizers will do per-panel MPPT and produce 2500W.   Whether that matters depends on your installation, but mine is certainly not a 'marginal' installation as I have a nicely pitched roof facing due south in southern California--and I do get some benefit from micros over string, something like 10-15%.

[Siwastaja]

I would assume the MPPT getting stuck at local maximum would be a solved problem long time ago, in popular production string inverters that get installed (e.g., Fronius, Huawei, etc.)

At least my Steca is able to find the global maximum, and it's a decade old product already.

[fourfathom]

I don't think mental modeling / understanding of the panel without bypass diodes does any good, because the diodes will practically be there for a very good reason. Unless you really want in-depth understanding of the semiconductor physics.

This is like "what if your car engine missed one spark plug, how would it perform?" when considering fuel economy of different driving styles, or comparing different cars.

Suit yourself.  Different panels have different bypass diode configurations, and I think it makes a difference.

[Siwastaja]

I don't think mental modeling / understanding of the panel without bypass diodes does any good, because the diodes will practically be there for a very good reason. Unless you really want in-depth understanding of the semiconductor physics.

This is like "what if your car engine missed one spark plug, how would it perform?" when considering fuel economy of different driving styles, or comparing different cars.

Suit yourself.  Different panels have different bypass diode configurations, and I think it makes a difference.

Educate me, is there going to be any other "configuration" than just varying the number of cells per each diode, or the physical shape in which the cells are arranged, which is of course interesting if we have horizontal or vertical shadows.

In any case, if the bypass diodes exist at all in the module (as they practically always do, of course), I can't see how the whole string could be brought to the near zero production of the shaded module.

[fourfathom]

Educate me, is there going to be any other "configuration" than just varying the number of cells per each diode, or the physical shape in which the cells are arranged, which is of course interesting if we have horizontal or vertical shadows.

In any case, if the bypass diodes exist at all in the module (as they practically always do, of course), I can't see how the whole string could be brought to the near zero production of the shaded module.

OK, the 36-cell panels I have used have two bypass diodes (18 cells per section).  I just found a Panasonic 335W panel with 96 cells and four bypass diodes (24 cells per section).  An array of the old 36-cell panels will likely have less power reduction during partial shading than one with the 96-cell panels. 

Of course there are going to be valid reasons why you would prefer one panel design over the other, and for rooftop applications partial shading probably isn't at the top of your list of issues.  But on a sailboat partial shading will be one of your primary concerns.  And yes, I know that sailboats are a microscopically small volume consumer of solar panels, but it matters to me.

[NiHaoMike]

You would need an inductor at each panel if you wanted to MPPT each panel individualy.

What you are trying to regulate is not the voltage at the end of the string. It is the voltage across the actual solar panel. If you PWM the panel onto the series string you get either full current or 0 current trough the panel. This would give you an average of the two extremes in the panels IV characteristic. What you actually want to do is hold the operating point at the maximum power level. So for this you would need an inductor at the panel so that on one cycle the panel is charging the inductor with energy, then once switched over the inductor and panel are sending energy back out to the string.

It can be done with capacitors across the inputs so the output will switch between Vmp and zero/bypass. It would be quite easy to have the frequency high enough that only film capacitors would be needed. In the case of the per cell MPPT, the low voltage would allow the frequency to be increased even more so that the capacitance of the cell itself serves that function. There only needs to be one inductor per string which can be shared with the boost stage that comes into play at low light levels.

I do believe that microinverters will eventually win, but that will be after more rigorous integration, cost reduction, and reliability improvements. What I really mean, inverter becomes part of the panel, so you just basically have a sleek panel with AC output.

Main weakness I found with them is that unlike DC systems that are able to output power as long as there's sunlight, AC systems have the problem of not being able to start up without some other power source providing voltage. I'm sure part of that is the safety aspect of making the output touch safe when not connected, but having all of them agree on the phase without something providing a timing reference probably isn't that easy to solve.

[trobbins]

From a general perspective, I'd suggest we are only really discussing issues in the domestic or small commercial/rural or niche markets.  Large installations either don't have shading designed in to their layout, or use string inverters where only a small % are subject to some form of shading and that has been 'designed in' to their business case.  Even for many domestic installations, a competent PV designer has assessed the site setup as not losing significant annual energy due to shading.  And I'd suggest that many who think they have a shading problem have not adequately assessed the actual measured loss or their particular consequences of any such loss.  Even the OP doesn't have an actual installed system to assess and is only starting on his journey to becoming aware of what is required to be a competent PV installer/designer.

[Siwastaja]

From a general perspective, I'd suggest we are only really discussing issues in the domestic or small commercial/rural or niche markets.  Large installations either don't have shading designed in to their layout, or use string inverters where only a small % are subject to some form of shading and that has been 'designed in' to their business case.  Even for many domestic installations, a competent PV designer has assessed the site setup as not losing significant annual energy due to shading.  And I'd suggest that many who think they have a shading problem have not adequately assessed the actual measured loss or their particular consequences of any such loss.  Even the OP doesn't have an actual installed system to assess and is only starting on his journey to becoming aware of what is required to be a competent PV installer/designer.

It's a good idea to simulate with a tool such as PVGIS (https://re.jrc.ec.europa.eu/pvg_tools/en/tools.html), using a horizon file.

Partial shading might look nasty to the eye, but it could still be just a few % in annual production.

If you have a decently sized string inverter where minimum voltage of MPPT does not become a problem, and the MPPT algorithm handles global maximum - both assumptions are true with modern string inverters - even if you shade half of your panels, you won't lose that much of production. For example, you get 5*300W + 5*0W instead of 5*300 + 5*50W. And if the panels are shaded for more than an hour or two per day, it makes absolutely no financial sense to install those panels at all.

Spot shading within one module is then dependent on the module, and microinverter or optimizer cannot make it magically better.

I have an issue where 6 out of my 10 panels are shaded around 7am-8.30.am by a tree, coupled with misdimensioning the inverter, one designed for normally 16 panels. Now this ratio 4/16 is just too small so minimum voltage is not satisfied and MPPT cannot track below 200V, so production is very minimal. Still not a big deal in annual production. But this has started to bug me lately, because this happens to be the most expensive hour in the Nordic spot market.

[Alti]

Hi all,

I've been reading up about solar, and one thing confuses me (well lots of things do, but you have to start somewhere!)

From what I have read, if you have a string of identical panels in series and one of the panels becomes shaded, the power output of the whole string drops to match the shaded panel.

Obviously this is confusing, because it is a total and utter made up lie. It does not work that way. For example, I have a string of 10 panels where I get 1 to 2 completely shaded panels in late afternoon. No such effect.(..)

I believe this is the consequence of misinterpretation/extrapolation of marketing claims of PV market players that advertise optimizers and microinverters.

What they claim is that in a traditional string:
 A: shading one panel to 0% results in Pstring = 100% (implicit)
 B: shading one panel to 50% results in Pstring = 50%
and now OP et al. extrapolated A and B, adding another claim to that:
 C: shading one panel to 100% results in Pstring = 0%

Since this is an advertisement, IMHO it is to be reasonably expected that Joe Average without technical background understands B is true only in very specific narrow case selected for the benefit of batterizer advertizer.

http://www.gsmlimited.com/wp-content/uploads/2013/09/Solar-Edge.jpg

[Berni]

It can be done with capacitors across the inputs so the output will switch between Vmp and zero/bypass. It would be quite easy to have the frequency high enough that only film capacitors would be needed. In the case of the per cell MPPT, the low voltage would allow the frequency to be increased even more so that the capacitance of the cell itself serves that function. There only needs to be one inductor per string which can be shared with the boost stage that comes into play at low light levels.

Ah yes a capacitor like this could work.

Tho in practice this means that you are putting sharp high frequency square waves on the series string. In a typical installation this string has quite a large loop area as it snakes around panel to panel with the return then taking the shortest route (rather than retrace the string back along the same path). This means that we might not even need an inductor on the end since the string itself is likely going to have a self inductance that comes close to the miliHenry range. However making this pass EMI regulations is another thing. Just laying out a small low power DC/DC converter in a wrong enough way can make you fail EMI compliance tests pretty badly. So sensing square waves in such a long string might make it transmit more RF power than a WiFi router.

[Berni]

I believe this is the consequence of misinterpretation/extrapolation of marketing claims of PV market players that advertise optimizers and microinverters.

What they claim is that in a traditional string:
 A: shading one panel to 0% results in Pstring = 100% (implicit)
 B: shading one panel to 50% results in Pstring = 50%
and now OP et al. extrapolated A and B, adding another claim to that:
 C: shading one panel to 100% results in Pstring = 0%

Since this is an advertisement, IMHO it is to be reasonably expected that Joe Average without technical background understands B is true only in very specific narrow case selected for the benefit of batterizer advertizer.

http://www.gsmlimited.com/wp-content/uploads/2013/09/Solar-Edge.jpg

Yeah this picture is completely wrong.

What happens instead is that when a panel gets slightly shaded the other panels keep pushing the same current trough it. As a result the slightly shaded panel that might be producing 50% full power starts sliding down its IV curve rapidly, since it can't maintain a voltage at this high of a current. As a result the array pulls the panels voltage to 0 and then down into negative where the pass diodes catch it. As a result the panel that would be otherwise producing 50% of power is actually burning a few % of power into heat on the pass diode. So not only is it not producing any power, it is burning a bit of the power that other panels produced.

The reason shading within a panel is not as bad is because there are also some parallel connected cells in there, so the cells can help each other out a bit. Also shadows tend to be pretty fuzzy from a distance. Certainly does affect them but MPPT can still milk some usable power out of them.

Would have expected better from a big solar electronics manufacturer like SolarEdge, but then again the actual technical details of why partial shading is bad is not as simple of a topic even for tech minded people on this forum (Hence we have this thread) but now imagine trying to explain this to the typical end costumer that works as an accountant at the local bank and calls an electrician to wire up a wall outlet. There goal is to sell you there fancy solar tech after all.

[tautech]

Dave did a recent video on this:

[Siwastaja]

(picture)

Yes, that attachment is an outright lie. It is not some kind of exaggeration or "optimal" case, it's total bogus and would belong to the "dodgy technology" section. There is no way I could buy any product from this kind of company. In fact, this is a scam that should be handled by the authorities.

No wonder this misconception lives so strong.

[Alti]

(picture)

Yes, that attachment is an outright lie. It is not some kind of exaggeration or "optimal" case, it's total bogus and would belong to the "dodgy technology" section.

I tend to disagree, you misinterpret the advertising claim (B), expecting this property to hold in all but exceptional circumstances. That is not a lie but the way advertising works. A shift of a wonky MPPT due to partial shading results in output power drop whether you like it or not. Yes, that requires high noon, no clouds, primitive and greedy MPPT, string must include only three panels, shading must cover 50% of cells gradually, etc, etc.. But it is a fact, narrow case, not a lie.

I attach some SolarEdge claims.
These vary from 50% shading / 50% Pstring to 10% shading / 90% Pstring but the overall concept is identical. I suppose percentage varies country by country.

https://www.solaredge.com/solutions/residential/system-benefits


https://swb-herten.de/solaredge/




And a pdf in Polish.

 solaredge-najlepsze-rozwiazaniePV.pdf (2504.14 kB - downloaded 33 times.)

No wonder this misconception lives so strong.

Extrapolation.

[Siwastaja]

Well, they use words "typical" or "traditional" to describe practically nonexistent systems. Competitors on the market do not perform as described. This is the part where the scam is.

This is not OK, like it or not. In marketing, descriptions of "typical" competitor systems must be realistic. You don't need to do careful analysis to find average or median system, but you can't take a malfunctioning system, or something that has not been sold in years. You also can't cherry-pick some special edge case without mentioning it.

FUD is one of those marketing tricks that really piss me off. If you don't have a viable business case on your own merits, you should be improving your product or doing something else, instead of lying about how your competitors perform. And this is the problem with microinverters and power optimizers: they usually offer very little production gains. Thus, manufacturers have no other fair option but to accept they are filling in a niche. I know it is tempting to try to sell these to everyone, but that easily escalates into iffy marketing or outright lying.

Power optimizers are in specially tight spot, because at least with microinverters, you have the obvious use case in very small systems, or systems that are very distributed by nature: two panels here, three panels there, one panel over there. Power optimizers still require the string inverter, and it needs to be a specialized one - only on PowerPoint made by an engineer it is a "simpler one".

[Alti]

Well, they use words "typical" or "traditional" to describe practically nonexistent systems. Competitors on the market do not perform as described. This is the part where the scam is.

This is not OK, like it or not. In marketing, descriptions of "typical" competitor systems must be realistic.

So in your opinion 50% shading/50% Pstring claim is exaggerated and unrealistic?

SolarEdge 60% shading/40% Pstring. 

http://www.elektromuench-ulmen.de/html/solaredge.html


OT(or maybe not): Pictures of claims are excerpts from marketing materials from SolarEdge, it is quite tricky to reach the original pdf's from their servers, these are just pictures with no text to search for. If anyone knows how to find original documents - please attach pdfs.

[Siwastaja]

It's not an opinion, it's a simple fact. Anyone with a PV system can observe this, me included. I just observed it a few hours ago as my output was around 80% of normal with two panels out of ten shaded pretty severely. I will observe it again tomorrow, and every day, as the chimney shades first one, then one more panel in the afternoon. (Knowing what exactly is "normal" is of course challenging, as the illumination angle gets worse to the non-shaded panels at the same time, too. But you can obviously see there is no any kind of dramatic reduction.)

That 40% 40% 40% 40% 40% 40% is not going to happen in actual reality, it makes absolutely no sense, PV systems do not work that way. Totally made up.

Really it is 0% 100% 100% 100% 100% 100%. Or to be exact, a tad below 0% because of the Vf of the bypass diodes, but this is not much. Can also be over 0% if shadowing is very light, in which case all the others might be a tad below 100% (like 98%).

The best thing optimizers or microinverters could offer for me would be the per-module monitoring which could be very interesting.

[fourfathom]

That 40% 40% 40% 40% 40% 40% is not going to happen in actual reality, it makes absolutely no sense, PV systems do not work that way. Totally made up.

I know, I'm going on and on about sailboat installations, but similar shading across multiple panels really does happen, and this is one reason why we consider shading sources when we look at panel array orientation.  I've got three panels where the shadow of the boom would fall on all three if I had the panels arrayed fore and aft.  Instead, they are arrayed port to starboard so the shadow only hits one panel, or at most two partially.  The same issue arises with the mast and to a lesser extent the shrouds and stays.  A flagpole might create a similar shadow on a terrestrial installation, but the shadow would be much less severe -- not a 40% situation.

[Siwastaja]

Yes, panel orientation matters if you have straight line shadows or something like that.

It's worth noting though this is orthogonal to the string vs. micro discussion. If you have such shadow shapes, it is worthwhile to build the right configuration, micros or string.

[bdunham7]

Yes, that attachment is an outright lie. It is not some kind of exaggeration or "optimal" case, it's total bogus and would belong to the "dodgy technology" section. There is no way I could buy any product from this kind of company. In fact, this is a scam that should be handled by the authorities.

It's not that bad.  The case they show is a simplistic version of what happens may happen if the string inverter doesn't have global MPPT, the global MPPT is not enabled or as in your example, the loss of one or more panels results in the string voltage going below the MPPT threshold voltage.  But in their simple example, even if the global MPPT was available and working, the output would be 0 + 100 +100 or 2/3 full capacity, whereas the per-panel optimizers would make that 50 + 100 + 100 or 5/6 full capacity. 

[Siwastaja]

Very strong "may happen". And this isn't relevant because global MPPT algorithm is needed in any case. All modern string inverters have it, but those optimizers need it too because they are also dealing with series strings like it or not (within a module) which is still susceptible to non-monotonous power curve.

I don't understand why people try to somehow justify the wrong claims by trying to come up with some rare edge case in which their claims are true.

This is like claiming opamps are bullshit and you need our new xyzqwerty-amp because typical opamps suffer from phase reversal every day in every circuit, while in reality phase reversal only happens in certain cases, and only with some ancient opamps easily avoidable.

[bdunham7]

I don't understand why people try to somehow justify the wrong claims by trying to come up with some rare edge case in which their claims are true.

But you've provided a specific example of this very thing happening to you!

I have an issue where 6 out of my 10 panels are shaded around 7am-8.30.am by a tree, coupled with misdimensioning the inverter, one designed for normally 16 panels. Now this ratio 4/16 is just too small so minimum voltage is not satisfied and MPPT cannot track below 200V, so production is very minimal. Still not a big deal in annual production. But this has started to bug me lately, because this happens to be the most expensive hour in the Nordic spot market.

I can't speak for Solar Edge, but a microinverter system like mine would not have that issue at any power level.  And global MPPT or not, the micro based system generally outperforms the string in shading situations--just not as much as the simplistic example that offends you.  It has to choose between two generally non-optimum solutions and the non-optimumness of those solutions is not rare, it is normal.  Now whether that bit of additional power is worth any additional cost, that is another issue--and a more important one, IMO.

[Alti]

A marketing masterpiece. Or at least a clever one, an example of resources well invested IMHO. Either investing into improving a product, or hiring consultancy agency, marketing specialists and analysts for shading claims.. I'd say that this direction might have been based on the fact 98% of the decision makers won't/don't care how PV works, these SolarEdge presentations are clear, simple and convincing and target Joe Average.

What puzzles me is: why have they gone only to 60% shading claims and stopped there? I think SolarEdge won't ever go as far as: X% shading/(100-X)% Pstring, for any X, as in OP but why 60%? Is this 60% some "decency level" that should not be crossed?