SMA SB7000TL-US-22 Inverter - Accessing individual MPPT performance information?

[sensij]

Some violence in the voltage in the morning and evening seems to be a normal side effect of MPPT... there isn't much power there, to the optimization routine has a harder time figure out where to operate. Here is an example from a single panel on my system, showing voltage and power on an otherwise clear day. I have some structural shade in the afternoon, but the morning is about as clear as it can be in a residential area without a perfectly clear view to the horizon. V on 2-15.JPG



The mpp voltage on your SE and SW arrays is probably not as different as you are imagining, and the mismatch won't amount to much in the way of losses... voltage is only weakly affected by the irradiance, and is much more strongly affected by temperature. Current, on the other hand, is much more sensitive to irradiance than to temperature.

Before SMA got into the multiple MPPT game, they aggressively defended the idea of multiple orientations in parallel on a single MPPT (as long as the strings were of equal length). One study showed a mismatch power loss of only 0.25%. In your case, that could easily be wiped out by the difference in efficiency from the MPPT loading.

http://www.smainverted.com/files/201...-TEN122510.pdf

With respect to your daughter's system, there isn't a problem, really. A lightly loaded inverter will have a different efficiency than one more heavily loaded, and from curves I've seen, once you drop below 20% loading the efficiency starts to fall off quickly. In other words, especially in the morning and evening when there isn't much sun, the inverter efficiency is probably a bigger deal than voltage mismatch in the panels. I'm not sure if there is any benefit to jumping the two MPPT's together in her case, but that is an option available in installations that require only a single mppt.

Efficiency test data on the SB6000TL-US-22 is available, and most other inverters you might be interested in can be found here:

http://www.gosolarcalifornia.org/equ...ests/summaries

[DaveDE2]

The SMA data sheet for your inverter indicates each MPPT can be driven up to 18A. So it appears that a large imbalance between the MMPTs is ok. You'd expect that, particularly on systems with arrays pointing in different directions. If it were me, I'd put both southeast strings on the same MPPT.

[sensij]

The SMA data sheet for your inverter indicates each MPPT can be driven up to 18A. So it appears that a large imbalance between the MMPTs is ok. You'd expect that, particularly on systems with arrays pointing in different directions. If it were me, I'd put both southeast strings on the same MPPT.

Imbalance isn't the only factor... by putting both SE strings on the same MPPT, that tracker will see higher peak power than if the SE and SW strings are combined. All else being equal, I think the inverter will be happier with the somewhat lower peak sustained somewhat longer that the current configuration provides.

Ultimately, either configuration is "ok", and the discussion drifts towards what may be more efficient, and what will prolong the inverter's life. Reliable data on both of those are hard to come by, and not super easy to model given the dynamic nature of the system's operation.

[DaveDE2]

My thinking is that the way the system is wired now neither of the two combined strings are operating at their MPP except at around high noon when they are equally illuminated. At mid morning and afternoon there may be a severe mismatch. For example if one string sees 600w/m^2 while the other is seeing 300w/m^2 then the first string will be putting out roughly twice the current as the second whilst both have about the same voltage. This means the MMP (or input impedance of the inverter, R=V/I) for each array will differ by a factor of two. Since the inverter can only operate at one input impedance, it won't be optimal for either array, but something in between. On top of that, since both strings want to put out different voltages, there will be current flow from one string into the other which is wasted power. It's like connecting two batteries of slightly different voltages in parallel. Granted, leaving the configuration as it is now will work one of the MMPTs less hard. Whether that has a significant impact on inverter life is as you say difficult to know. It depends on design specifics of the inverter.

[DaveDE2]

...actually, I take what I said above about current flowing from one string into the other string back. The two strings don't act like batteries, but rather like current sources. No current should flow from one string to the other with different irradiances.

[MinorDeity]

sensij,

From the voltage/power graph you provided it looks like you are able to monitor individual MPPTs too. What software application are you using?

Also, I see you are in San Diego too, just like me. What area are you in? I'm in Clairemont in the Mt. area on Mt. Laudo Dr. You didn't happen to have Rancho Solar do your installation did you? Maybe we can exchange horror stories

[sensij]

The SMA link I provided above was thin on detail, here is a better paper from Fronius:

http://www.fronius.com/cps/rde/xbcr/SID-AF30EAED-B3F2642E/fronius_international/SE_TEA_Efficient_East_West_orientated_PV_systems_with_MPP_tracker_EN_357684_snapshot.pdf

DaveDE2... you've posted your thoughts on this topic in a couple of threads, but seem unable to reconcile your understanding with the data generated by actual studies and experiments. You might want to model numerically the actual irradiance differences and effects on impedance and power output to see if you can explain why that gap exists.

[sensij]

sensij,

From the voltage/power graph you provided it looks like you are able to monitor individual MPPTs too. What software application are you using?

Also, I see you are in San Diego too, just like me. What area are you in? I'm in Clairemont in the Mt. area on Mt. Laudo Dr. You didn't happen to have Rancho Solar do your installation did you? Maybe we can exchange horror stories

I have a SolarEdge system, which offers individual panel monitoring. It reports the input and output of each panel's optimizer, as well as what the inverter is doing. Navigating the SolarEdge portal to get at the useful information takes some work, unfortunately none of the panel stuff is available through their API.

I'm off of Mt. Acadia, near the Tecolote Canyon golf course... just a couple mi away from you. My installer is actually active on the forum, member ButchDeal.

With your application building abilities, sharing your data on PVOutput.org should be a breeze if you are interested. My system is in my signature, the data is accumulated as part of Team San Diego. We have quite a community of PV owners!

[inetdog]

My thinking is that the way the system is wired now neither of the two combined strings are operating at their MPP except at around high noon when they are equally illuminated. At mid morning and afternoon there may be a severe mismatch. For example if one string sees 600w/m^2 while the other is seeing 300w/m^2 then the first string will be putting out roughly twice the current as the second whilst both have about the same voltage. This means the MMP (or input impedance of the inverter, R=V/I) for each array will differ by a factor of two. Since the inverter can only operate at one input impedance, it won't be optimal for either array, but something in between. On top of that, since both strings want to put out different voltages, there will be current flow from one string into the other which is wasted power. It's like connecting two batteries of slightly different voltages in parallel. Granted, leaving the configuration as it is now will work one of the MMPTs less hard. Whether that has a significant impact on inverter life is as you say difficult to know. It depends on design specifics of the inverter.

Your are looking at it wrong. Let me reiterate what sensij said but in different words.

You can either look at the MPPT as providing the right load resistance to let the panel operate at the MPP or as regulating the panel voltage to Vmp by regulating the current being drawn. Either is valid but one or the other may be more useful in analyzing certain situations.

In the case we are considering, as long as the array temperatures are similar the Vmp for each string will be the same, contradicting the quoted text in bold. By regulating the total current drawn from the two strings in parallel the MPPT algorithm can cause the voltage to go to Vmp for both strings. They will just deliver different amounts of current depending on their irradiation.
If we look at the variable load resistance, you can figure out what the ideal load resistance, R1, is for string one and the ideal load resistance, R2, for string two.
All that the MPPT algorithm then needs to do is to figure out that for the two strings in parallel the correct load resistance is the parallel combination of R1 and R2. It is not necessary to have the inverter input impedance match either string directly, just the parallel combination.
Of the two analyses, I find the one that says the MPPT algorithm needs to seek the same voltage for both strings to be easier to follow.

If the two strings had different lengths or some of the panels of one string were shaded, then the two Vmp values would not be equal and running them in parallel into the same MPPT input would not work as well.

But even with a difference in the string voltages the effect on the overall efficiency may not be as great as you might expect. Because of the slope of the total power versus voltage (or current) curve of a silicon panel is by definition flat at the MPP, small variations from the MPP will not have a proportional effect on the output power.
The accepted rule of thumb is that if the two string voltages are matched within 5% and the strings are then put in parallel the loss of power will be less than 1%. Maybe only .5%?
Similarly if two panels have the same Imp within 5% they can be put in series without losing more than 1% of the output power.

[DaveDE2]

Your are looking at it wrong. Let me reiterate what sensij said but in different words.

You can either look at the MPPT as providing the right load resistance to let the panel operate at the MPP or as regulating the panel voltage to Vmp by regulating the current being drawn. Either is valid but one or the other may be more useful in analyzing certain situations.

In the case we are considering, as long as the array temperatures are similar the Vmp for each string will be the same, contradicting the quoted text in bold. By regulating the total current drawn from the two strings in parallel the MPPT algorithm can cause the voltage to go to Vmp for both strings. They will just deliver different amounts of current depending on their irradiation.
If we look at the variable load resistance, you can figure out what the ideal load resistance, R1, is for string one and the ideal load resistance, R2, for string two.
All that the MPPT algorithm then needs to do is to figure out that for the two strings in parallel the correct load resistance is the parallel combination of R1 and R2. It is not necessary to have the inverter input impedance match either string directly, just the parallel combination.
Of the two analyses, I find the one that says the MPPT algorithm needs to seek the same voltage for both strings to be easier to follow.

If the two strings had different lengths or some of the panels of one string were shaded, then the two Vmp values would not be equal and running them in parallel into the same MPPT input would not work as well.

But even with a difference in the string voltages the effect on the overall efficiency may not be as great as you might expect. Because of the slope of the total power versus voltage (or current) curve of a silicon panel is by definition flat at the MPP, small variations from the MPP will not have a proportional effect on the output power.
The accepted rule of thumb is that if the two string voltages are matched within 5% and the strings are then put in parallel the loss of power will be less than 1%. Maybe only .5%?
Similarly if two panels have the same Imp within 5% they can be put in series without losing more than 1% of the output power.

Although I believe we are all in violent general agreement about all of this, I respectfully disagree with your assertion that "as long as the array temperatures are similar the Vmp for each string will be the same". I'll say it again, different irradiances cause different cell currents and voltages and different MPPs. Look at any solar panel data sheet and the family of I-V vs irradiance curves and MPP plots. The MPP voltages vs irradiance are not the same value.They're almost the same, but not exactly. Some vary by several volts from 20-80% irradiance. I think therin lies our argument. Yes, I agree that if 1%ish loss in efficiency numbers are good enough for someone, then go ahead and combine strings with different irradiances. On the other hand, if you're trying to eek out every last 0.5%-1% like so many of the people here are, then I submit that combining strings that are facing the same directions and putting them on the same MPPT makes sense.

[inetdog]

...
....They're almost the same, but not exactly. Some vary by several volts from 20-80% irradiance. I think therin lies our argument. Yes, I agree that if 1%ish loss in efficiency numbers are good enough for someone, then go ahead and combine strings with different irradiances. On the other hand, if you're trying to eek out every last 0.5%-1% like so many of the people here are, then I submit that combining strings that are facing the same directions and putting them on the same MPPT makes sense.

Let's take a closer look at your argument on that:

Assume two strings, one producing 1kW at peak sun and facing West and the other producing 1kW at peak sun and facing east.
Look first at the time when the sun is at SW in the sky. The two strings will be producing equal power (less than 2kW, but equal contributions from each string.)
No loss of efficiency at all in paralleling them since they have the same irradiance and presumably same temperature.
Now lets move to the time when the sun is due S of the arrays.
One array is producing 1kW, more or less while the other is at at most 40% irradiance and is producing roughly 400W. The Vmp of the second string may be 3% lower (1 volt out of 30).
If the two strings had equal power output the overall power loss could be as much as .3% as a result. But since one string is producing 1kW and the other is producing only 400W, The loss in total power is more like .15%. and for only a low production point in the solar day. Look at the time when the irradiance is only 20%. The voltage difference is twice as large but the power is half as much. Result is still less than .15% overall. Compare that to the constant loss of 2 or 3% or more through the whole day from mixing a 7 panel and an 8 panel string.
I agree that there is a difference, I just do not think that it is enough of a difference to override other concerns or to pay extra for an additional MPPT input in most if not all cases. Remember that any static solution has to be considered averaged over the whole production day.

[DaveDE2]

Really? Are you sure you aren't confusing temperature effects with irradiance effects, at least for the common panels being installed today? Does 4:1 accurately reflect the ratio of irradiance for a SE and SW array for a meaningful amount of time?

I'm getting this from some plots I've seen on the internet. One such is attached here. In the case of this plot, the power loss from half irradiance to full would be about 1-2% if Vmp were fixed. Having said that, I have not analyzed MPP plots for common panels being installed today and maybe I should and my mind would change if I saw that MPP differences caused insignificant loss at fixed V. According to you and inetdog sounds like this is the case for most panels. I read the SMA article you referenced. 0.25% loss for their case is indeed low and impressive and quite surprising to me. Btw, I also get what you're saying about load balancing possibly being an advantage, not just about possible longevity increase which I'm not sold on yet but more about the possibility of inverter efficiency increase with increased power and *possibly* nullifying any advantage of rearranging the strings in the OP's case.

[DaveDE2]

Let's take a closer look at your argument on that:

Assume two strings, one producing 1kW at peak sun and facing West and the other producing 1kW at peak sun and facing east.
Look first at the time when the sun is at SW in the sky. The two strings will be producing equal power (less than 2kW, but equal contributions from each string.)
No loss of efficiency at all in paralleling them since they have the same irradiance and presumably same temperature.
Now lets move to the time when the sun is due S of the arrays.
One array is producing 1kW, more or less while the other is at at most 40% irradiance and is producing roughly 400W. The Vmp of the second string may be 3% lower (1 volt out of 30).
If the two strings had equal power output the overall power loss could be as much as .3% as a result. But since one string is producing 1kW and the other is producing only 400W, The loss in total power is more like .15%. and for only a low production point in the solar day. Look at the time when the irradiance is only 20%. The voltage difference is twice as large but the power is half as much. Result is still less than .15% overall. Compare that to the constant loss of 2 or 3% or more through the whole day from mixing a 7 panel and an 8 panel string.
I agree that there is a difference, I just do not think that it is enough of a difference to override other concerns or to pay extra for an additional MPPT input in most if not all cases. Remember that any static solution has to be considered averaged over the whole production day.

Thanks for the example, not sure if I follow your numbers here but I believe you're accounting for power output loss numbers by only considering changes in voltage but not resulting changes in current when the voltage is not at the MPP. Given that, I believe your calculations for loss percentages are optomistically low, but I see where you're going with the argument.

By the way, the OP would not have to pay for another MPPT, just move a few wires in the dc disconnect. Cheers.

[inetdog]

I believe you're accounting for power output loss numbers by only considering changes in voltage but not resulting changes in current when the voltage is not at the MPP..

No. When the voltage is above MPP the current is below MPP and vice versa. Exactly at MPP the slope of the power curve is zero.

[sensij]

Let's reintroduce temperature into the discussion. In the isothermal case, the orientation with less irradiance will have a slightly lower Vmp. However, the orientation with less irradiance will also have a lower cell temperature, which should *increase* Vmp for the lower irradiance array.