Optimizers... what are they good for?

[mspohr]

Most of this discussion has been about microinverters (Enphase) rather than optimizers. Microinverters are more expensive.
The last array I installed I used Tigo optimizers. These are panel level and can be used with a wide variety of panel sizes, voltages, capacities. (16-80 volts, 15 amps, 700 W)
These are only $40 each and can be used with any string inverter. You only need to put them on the panels subject to shading. You can leave them off of other panels in the string which aren't shaded.
Nice panel level monitoring along with a useful depiction of "reclaimed" energy (i.e. extra energy available due to the optimizer).
I chose these for this particular array since there is extensive seasonal shading of part of this array.
At times they show up to double the energy due to their ability to compensate for the shading. (Design using unequal string lengths, mixed orientations, or areas of mismatch. Install in shaded areas with a reduced setback ratio. In addition to optimization, the TS4-A-O enables module level monitoring, and rapid shutdown in compliance with NEC 2014, 2017, 2020.)
For this array they were a good investment.

 

[wwu123]

That was probably true ~5 years ago. But today the vast majority of string inverters have 3 or 4 trackers. So you can have 3 strings facing different directions with different shading at different times and your annual production will be about the ~same. I am a huge proponent of each string on its own tracker.

I believe with the new IQ series if you want monitoring you need the combiner box. But it's probably not required for a single string if you don't want monitoring...

Thanks for this video, it's very good. I watched similar videos a few years ago earlier from this thread, and elsewhere, that had similar results, the reason this video is better, is that it shows what is happening with the voltage and current in each of the test scenarios, on both the shaded and unshaded panels. So finally it illustrates what is happening with the bypass diodes (which you mentioned in the very first post starting this thread) on individual shaded panels, and decent modern MPP tracking (which you do mention in follow-up posts), and it is those two things that mitigate shading on a few panels from impacting the entire string.

So even for my shaded 10-year old system, a single string inverter with one MPP tracker would have been fine, as all my panels are on two sub-arrays (roof planes), but they're all in the exact same orientation. So I actually stand corrected, micro-inverters were not the only feasible solution for my shading. Though just like Tesla's low pricing, the non-profit installer was not offering me a choice, they had the lowest price around that happened to include micro's.

The production benefit of optimizer/micro's is not totally zero though - the above video at the end shows despite string inverters mitigating shade fine, there was an overall 10% increase in production in from optimizers. So 20% more cost for 10% more production. Had I had a choice though, sounds like I could have used cheaper string inverter, and with the saving probably put in 20% more panel area for the same price, netting perhaps 10% more production (for the same price) than my actual system with micro's.

Here is where my understanding now lies:
1)partially/fully shaded panels do not really impact unshaded panels in the same string, due to bypass diodes and MPP. This was always my biggest fear, as trees spend 2+ hrs each day rolling across the arrays.
2)optimizer/micro's eke out a bit more yield on a partially shaded panel. Even though this winter time of year, each of my panels spends 2 hrs in full sun and 1.5 hrs in partial, rolling shade, so the partial time is significant, in the grand scheme of things it's small amount of extra production. And in summer when I get 6 hrs of full sun and 1.5 hrs of partial, it's relatively small.
3)optimizers/micro's eke out a bit more yield in the morning/evening, when rolling shade covers say more than half the string - because at a certain point, the remaining string voltage drops below the string inverter's minimum. Like here, at the end of my sun day (currently around 1:30 pm), half of each sub-array is already shaded. But again, this is a small percentage of extra production the micro's would be getting from the 50% unshaded panels, for 45 more minutes, after a string inverter shuts down.

It was all relatively moot for my current system, but 18 months ago when I posted in this thread the first time, I was considering a Powerwall with 4.8kw mandatory solar add-on from Tesla while still before NEM3. At that point, Tesla was no longer using SolarEdge optimizers, so it was just a Tesla string inverter or nothing. And I declined, not because of the Powerwall cost, but because I was concerned about #1 above, that in the my remaining further-shady roof space, that when one or two of the 12 proposed new panels started their daily shading, it was going to wipe out substantially all of entire 4.8kw string - which I now realize is NOT the case.

So totally regret now not going for it, esp in light of PG&E's 20% rate hikes this month and even more expected in March...

 

[nwdiver]

Thanks for this video, it's very good. I watched similar videos a few years ago earlier from this thread, and elsewhere, that had similar results, the reason this video is better, is that it shows what is happening with the voltage and current in each of the test scenarios, on both the shaded and unshaded panels. So finally it illustrates what is happening with the bypass diodes (which you mentioned in the very first post starting this thread) on individual shaded panels, and decent modern MPP tracking (which you do mention in follow-up posts), and it is those two things that mitigate shading on a few panels from impacting the entire string.

So even for my shaded 10-year old system, a single string inverter with one MPP tracker would have been fine, as all my panels are on two sub-arrays (roof planes), but they're all in the exact same orientation. So I actually stand corrected, micro-inverters were not the only feasible solution for my shading. Though just like Tesla's low pricing, the non-profit installer was not offering me a choice, they had the lowest price around that happened to include micro's.

The production benefit of optimizer/micro's is not totally zero though - the above video at the end shows despite string inverters mitigating shade fine, there was an overall 10% increase in production in from optimizers. So 20% more cost for 10% more production. Had I had a choice though, sounds like I could have used cheaper string inverter, and with the saving probably put in 20% more panel area for the same price, netting perhaps 10% more production (for the same price) than my actual system with micro's.

Here is where my understanding now lies:
1)partially/fully shaded panels do not really impact unshaded panels in the same string, due to bypass diodes and MPP. This was always my biggest fear, as trees spend 2+ hrs each day rolling across the arrays.
2)optimizer/micro's eke out a bit more yield on a partially shaded panel. Even though this winter time of year, each of my panels spends 2 hrs in full sun and 1.5 hrs in partial, rolling shade, so the partial time is significant, in the grand scheme of things it's small amount of extra production. And in summer when I get 6 hrs of full sun and 1.5 hrs of partial, it's relatively small.
3)optimizers/micro's eke out a bit more yield in the morning/evening, when rolling shade covers say more than half the string - because at a certain point, the remaining string voltage drops below the string inverter's minimum. Like here, at the end of my sun day (currently around 1:30 pm), half of each sub-array is already shaded. But again, this is a small percentage of extra production the micro's would be getting from the 50% unshaded panels, for 45 more minutes, after a string inverter shuts down.

View attachment 1007202

It was all relatively moot for my current system, but 18 months ago when I posted in this thread the first time, I was considering a Powerwall with 4.8kw mandatory solar add-on from Tesla while still before NEM3. At that point, Tesla was no longer using SolarEdge optimizers, so it was just a Tesla string inverter or nothing. And I declined, not because of the Powerwall cost, but because I was concerned about #1 above, that in the my remaining further-shady roof space, that when one or two of the 12 proposed new panels started their daily shading, it was going to wipe out substantially all of entire 4.8kw string - which I now realize is NOT the case.

So totally regret now not going for it, esp in light of PG&E's 20% rate hikes this month and even more expected in March...

My old SMA inverter failed at 9 years 6 months. I got super lucky in that it failed with 6 months left on the warranty and even luckier that they were out of refurbished units so they sent me a brand new 3rd generation model with 3 MPPT channels instead of 1. So now each of my 3 strings gets to run independently. Previously the back row would get shaded in the evening which would kill the remaining 11 panels in that string. It's a noticeable difference when you're watching production but in terms of kWh/yr it was only reducing the output by ~200kWh or ~1%.

 

[mspohr]

1)partially/fully shaded panels do not really impact unshaded panels in the same string, due to bypass diodes and MPP. This was always my biggest fear, as trees spend 2+ hrs each day rolling across the arrays.

The problem here is that when a shaded panel is not performing and is bypassed by the diodes, you lose all of the production from that panel.
If you have an optimizer (such as Tigo), it will impedance match to produce power from that module. In addition, if a module is just slightly shaded, it will have a different MPP curve from the unshaded modules and the inverter will not be able to efficiently use the power from that module. The Tigo optimizer reclaims that power.

 

[wwu123]

How much is that really worth? I don't think I have a single customer that routinely checks their systems anymore. It's fun for the first month, then they almost never look again. I actually get texts occasionally from customers for me to check their system because it's been too long since they last logged in and forgot how to.
...

You're basically right, even for a data nerd like myself, panel-level monitoring/logging is not important, though I do examine it occasionally to further my technical understanding. I only stumped up the $400 for Enphase monitoring box (=$0.10/watt, or $25/panel), because I certainly wanted array-level total production monitoring/logging, and that was the only way I could get it, it just happens to also give panel-level monitoring.

Of course I know nowadays that wrapping a set of $10 CT's around the 30A wires to my solar breaker would give me that raw data too, but there were no off-the-shelf solutions 10 years ago, or even really today, to integrate that with the cloud. I'd have to get the local CT data, then integrate with local server for Internet connection, and finally find some compatible upload solution to pvoutput.org or similar. So probably $200 in cobbled-together parts for a DIY solution, even today. Array-level production monitoring certainly easier with a string inverter that's cloud-connected...

 

[nwdiver]

The problem here is that when a shaded panel is not performing and is bypassed by the diodes, you lose all of the production from that panel.
If you have an optimizer (such as Tigo), it will impedance match to produce power from that module. In addition, if a module is just slightly shaded, it will have a different MPP curve from the unshaded modules and the inverter will not be able to efficiently use the power from that module. The Tigo optimizer reclaims that power.

Depends on the shade. If only one or two cell groups are in shade the results are the same with a string, micro or optimizers. It's only full or partial shade across all 3 cell groups that there's really a difference and then you're only capturing maybe ~20w.

 

[wwu123]

The problem here is that when a shaded panel is not performing and is bypassed by the diodes, you lose all of the production from that panel.
If you have an optimizer (such as Tigo), it will impedance match to produce power from that module. In addition, if a module is just slightly shaded, it will have a different MPP curve from the unshaded modules and the inverter will not be able to efficiently use the power from that module. The Tigo optimizer reclaims that power.

Well I think that's where the 10% extra production from the end of the recent Youtube video nwdiver linked comes from, is basically what that might amount to. 10% over the course of the day is not small, it's meaningful. But financially, it's possible I could have put up 20% more panel area with a cheaper string inverter for the same price, and make up that 10% with the 20% more panel area, and then some. Maybe.

I think it depends on the nature of shading issues. If someone has a chimney or powerline sitting across a panel for half a day, the optimizer/micro can make up well more that 10% more production. With my rolling shade, I'm partially shaded for only 1.5 hrs per panel where it helps, but it's only recovering maybe 50% production for 1.5 hrs. I may be fully shaded for 4 hrs where it doesn't help any.

 

[mspohr]

Depends on the shade. If only one or two cell groups are in shade the results are the same with a string, micro or optimizers. It's only full or partial shade across all 3 cell groups that there's really a difference and then you're only capturing maybe ~20w.

Actually, it can be more than that:
The following example shows the IV curves of 2 identical PV modules; one exposed to high irradiance (A) and
the other exposed to low irradiance (B).

Figure 1: Example of 2 identical PV modules exposed to different irradiance levels: module A exposed to high irradiance, and module B
exposed to low irradiance
Since the same current should apply to both modules, the Maximum Power Point Tracker (MPPT) at the
inverter or charge controller can select between module A’s power point (high irradiance, around 8.7A),
module B’s working point (low irradiance, around 5.2A), or any point in-between.
At 8.7A, the strong module, A, produces its maximum power, however the weaker module, B, is completely
bypassed (no point on its IV curve that matches 8.7A). At 5.2A the weaker module, B, is at peak power, however
the stronger module, A, can only produce ~185W. With module-level optimization by Tigo, each module (A or
B) can produce its maximum available energy, independently of the other modules in the string. The result in
this example can be quantified for all 3 options by summing the available power from both modules:
With a string current of 8.7A: 320W + 0W = 320W (strong module’s output + weaker modules’ output)
With a string current of 5.2A: 185W + 190W = 375W
With a string optimized by Tigo: 320W + 190W = 510W - conversion efficiency (~0.5%-1.5%) = 505W

 

[nwdiver]

Actually, it can be more than that:
The following example shows the IV curves of 2 identical PV modules; one exposed to high irradiance (A) and
the other exposed to low irradiance (B).
View attachment 1007212
Figure 1: Example of 2 identical PV modules exposed to different irradiance levels: module A exposed to high irradiance, and module B
exposed to low irradiance
Since the same current should apply to both modules, the Maximum Power Point Tracker (MPPT) at the
inverter or charge controller can select between module A’s power point (high irradiance, around 8.7A),
module B’s working point (low irradiance, around 5.2A), or any point in-between.
At 8.7A, the strong module, A, produces its maximum power, however the weaker module, B, is completely
bypassed (no point on its IV curve that matches 8.7A). At 5.2A the weaker module, B, is at peak power, however
the stronger module, A, can only produce ~185W. With module-level optimization by Tigo, each module (A or
B) can produce its maximum available energy, independently of the other modules in the string. The result in
this example can be quantified for all 3 options by summing the available power from both modules:
With a string current of 8.7A: 320W + 0W = 320W (strong module’s output + weaker modules’ output)
With a string current of 5.2A: 185W + 190W = 375W
With a string optimized by Tigo: 320W + 190W = 510W - conversion efficiency (~0.5%-1.5%) = 505W

In real world conditions if 1 or 2 of 3 cell groups are in hard shade those diodes are tripped and you lose the cell groups regardless of optimization. Only in full shade are you going to see a difference and full shade is generally closer to a ~95% reduction in irradiance levels. Optimizers can't recover energy from cell groups that are in shade. Only if current thru the entire panel is the ~same. Only if ALL cell groups are in shade.

It's not like panels in the same array are going to be exposed to different irradiance levels without shade unless they have different orientations and that's what multiple MPPT channels should be used for.

 

[mspohr]

In real world conditions if 1 or 2 of 3 cell groups are in hard shade those diodes are tripped and you lose the cell groups regardless of optimization. Only in full shade are you going to see a difference and full shade is generally closer to a ~95% reduction in irradiance levels. Optimizers can't recover energy from cell groups that are in shade. Only if current thru the entire panel is the ~same. Only if ALL cell groups are in shade.

It's not like panels in the same array are going to be exposed to different irradiance levels without shade unless they have different orientations and that's what multiple MPPT channels should be used for.

This is panel level shading.

 

[sunwarriors]

What would I have gained spending >$6k more for micros?

Finishing a 12kW project this weekend. Micros would have added >$4k to that project. What would ~$4k have bought me?

Another 15 years of warranty coverage? Is that worth $4k?

Slightly increased redundancy to maybe avoid losing ~a month of production? Spend ~$4k to maybe save ~$300?

You speak as someone who is an installer (and is an installer). Most people aren't you with ~a month of lost production or less. I see 3 months, 6 months regularly in threads all over prior. Why I didn't go with strings initially. It's sorta like saying why pay or have anyone install your solar at all? Just DIY and save $15k! It's easy. Why stop at strings right?

Why have anyone fix your car, just DIY and save mega $$. Strings and micros aren't the same tech. They have different costs/benefits and warranties. You think micros are never worth it, I disagree and it's as simple as that. We're never going to convince each other so we can just stop and stay on our respective sides.

You are pushing a product that is new from a new company. It has s 10 year warranty. What if they all failed in 11 years and the company is bankrupt and gone? See how a regular consumer (me) sees your constant bias on saving $$ for lower warranty and "possibly" more problems later? It's not just about $$ too really, but that's almost all you/string people keep spouting. It's about hassles/time/etc. Like I've never said there was a ROI on batteries, but having power when it's out is pretty priceless and can be life impacting as well. Same with strings, if it goes out, now I have no solar at all and I have to simply deal with hassles that I would prefer less of.

Is having all panel monitoring worth it for folks? It's definitely worth it for me and again, I did not see the large $6k difference in my own research/quotes. I'll honestly keep an eye for any new ones I find, but I just don't see the MASSIVE $$ you save from my own research or quotes I see posted, but I could be very wrong. Of course, if you are doing a commercial install with hundreds or thousands of panels, then yeah, strings will save a lot. For someone with 20 or 30 panels or typical small home installs in a small bay area home, I doubt the $ is that big.

 

[sunwarriors]

How much is that really worth? I don't think I have a single customer that routinely checks their systems anymore. It's fun for the first month, then they almost never look again. I actually get texts occasionally from customers for me to check their system because it's been too long since they last logged in and forgot how to.

Like I posted earlier. You don't need panel level monitors to spot, diagnose or find an issue. $300 IR camera or $4k for panel level monitoring....

I'm a few years in my install and still check my stuff like daily...

I will add I've called Enphase support directly and we debugged/fixed my issue with the 1st line tech support who answered the phone. (Zigbee issue).

Hoymiles is pushed by a youtuber I watch, but it looks like they are almost trying too hard, but they are cheaper with their multi-inverters to save on cost if someone wanted to go that route.

 

[nwdiver]

You speak as someone who is an installer (and is an installer). Most people aren't you with ~a month of lost production or less. I see 3 months, 6 months regularly in threads all over prior.

It's not like micro-inverters are a panacea for that. One of the systems I repaired was micro-inverters. It was down for over a year. That's why it's so important to use a local installer. The hail damaged system we repaired had two non-functioning M215s. Since it wasn't our system it wasn't worth the cost to the customer to pay us to chase down more M215s so those two panels will likely remain non-functioning for the remaining life of the system and this is a ground mount. It would be interesting to know how many micros go out and it's just not worth the hassle of climbing over a roof to replace a micro-inverter.

My priority is $/kWh. Reliability factors into that. It's not like we sell these systems and walk-away. It's 'ours' for 10 years. Anything happens and we fix it. If that GroWatt inverter fails it's up to us to resolve it.

No doubt there are people that like to follow their system performance closely. But they're the 1% if not less. Definitely none of my customers... I was afraid one of my 'old fashioned' clients was going to have a nervous break down when I was walking her thru the app....

 

[cali8484]

How much is that really worth? I don't think I have a single customer that routinely checks their systems anymore. It's fun for the first month, then they almost never look again. I actually get texts occasionally from customers for me to check their system because it's been too long since they last logged in and forgot how to.

Like I posted earlier. You don't need panel level monitors to spot, diagnose or find an issue. $300 IR camera or $4k for panel level monitoring....

I look at it regularly but certainly as often as I used to. It's very useful to uncover an expected seasonal shading from chimney early on (and move the panel) and periodically determine when trimming of certain trees is needed. I doubt IR camera would help much with that.

 

[wwu123]

It's not like micro-inverters are a panacea for that. One of the systems I repaired was micro-inverters. It was down for over a year. That's why it's so important to use a local installer. The hail damaged system we repaired had two non-functioning M215s. Since it wasn't our system it wasn't worth the cost to the customer to pay us to chase down more M215s so those two panels will likely remain non-functioning for the remaining life of the system and this is a ground mount. It would be interesting to know how many micros go out and it's just not worth the hassle of climbing over a roof to replace a micro-inverter.

I know this was not installed by you to bother with, but as a M215 system owner, I've been worried what happens when one fails under the 25-year warranty (which I think 100% of them are still within 25 years), given they were discontinued years ago and Enphase has no remaining inventory. I thought the reassuring thing I discovered within the past year, is that Enphase now manufactures a modified IQ7 replacement for warranties, that are compatible with the the older 4-wire systems, so basically swap-compatible as well as compatible with the PLC comms on the older Envoy monitoring systems? So that owner should have been able to get two of these modified IQ7's from Enphase, right? ( Again, not something you were obligated to bother with, since you weren't the installer, but the owner could have?)

I've read these modified IQ7's are so backward compatible, Enphase will not sell them to expand an older system, only as a like-for-like warranty replacement. So the only way to expand an existing M215 system would be to find used M215's on the secondhand market.

 

[zƬesla]

This is the first GroWatt we've installed since SMA and Fronius have fallen behind. Really impressed so far.

Can you elaborate on SMA? We have SunnyBoys 7.0 & 7.7 installed in April 2020. Is the issue one of reliability or new features not being added in the newest models?

 

[skepticcyclist]

I know this was not installed by you to bother with, but as a M215 system owner, I've been worried what happens when one fails under the 25-year warranty (which I think 100% of them are still within 25 years), given they were discontinued years ago and Enphase has no remaining inventory. I thought the reassuring thing I discovered within the past year, is that Enphase now manufactures a modified IQ7 replacement for warranties, that are compatible with the the older 4-wire systems, so basically swap-compatible as well as compatible with the PLC comms on the older Envoy monitoring systems? So that owner should have been able to get two of these modified IQ7's from Enphase, right? ( Again, not something you were obligated to bother with, since you weren't the installer, but the owner could have?)

I've read these modified IQ7's are so backward compatible, Enphase will not sell them to expand an older system, only as a like-for-like warranty replacement. So the only way to expand an existing M215 system would be to find used M215's on the secondhand market.

Why can't you replace an IQ7 with an IQ8? The plugs and wiring are exactly the same in my IQ7 system as they would be in an IQ8 system. There are only 2 wires, not 4. Solar panel plugs directly into the micro and their is a seperate plug to parallel multiple micros together

 

[nwdiver]

Can you elaborate on SMA? We have SunnyBoys 7.0 & 7.7 installed in April 2020. Is the issue one of reliability or new features not being added in the newest models?

Mostly new features. Most newer panels output a bit over 10A and SMA inverters clip at 10A. Growatt can take ~13A per channel and it's a hybrid. They also don't have anything larger than 7.6 in their residential line while GroWatt has a 11.4kW. SMA has a new generation nearly identical to GroWatt but it's not available in the US yet.

 

[cali8484]

So the only way to expand an existing M215 system would be to find used M215's on the secondhand market.

Other M series models (e.g. M250) should also work.

 

[mspohr]

I have 16 M250 on 285W panels that have been in service for 9 years. No problems.