Greetings from cloudy Frederick, MD

[Naptown]

Thanks, Rich. Wouldn't replacing the SB3000US inverters (500Vmax) with the SB3000HFUS (600Vmax) and leaving one of the 40 modules unplugged solve the cold weather voltage problem?

Yes




Yes, I believe option C above is fairly accurate.

There is a PV disconnect box for the utility on the outside of the house below the power meter. Inside, in the basement, the main circuit breaker is indeed 200A, and there is a 50A breaker at the bottom (labeled PV in this photo). Below the main breaker is a PV inverter sub panel with three, 20A breakers.

Then if there is room in that box for another 2 pole breaker it will make a line tap easier. I assume the inverters are outside?
I was looking at aggregating SREC's and what I could find is for systems over 10KW in Maryland they are requiring a revenue grade meter installed which you will have to read once a month and report. The readings from the inverter are not accurate enough apparently. I do not have anything out there right at 10KW DC and have not called the aggregator to check the requirement and the specifics. This is something you would want to do before you make the last payment.

[websterize]

Then if there is room in that box for another 2 pole breaker it will make a line tap easier. I assume the inverters are outside?

Yes, inverters are outside the house: http://flic.kr/p/byyi21

I was looking at aggregating SREC's and what I could find is for systems over 10KW in Maryland they are requiring a revenue grade meter installed which you will have to read once a month and report. The readings from the inverter are not accurate enough apparently. I do not have anything out there right at 10KW DC and have not called the aggregator to check the requirement and the specifics. This is something you would want to do before you make the last payment.

Roger that. We have already paid in full.

[Naptown]

OK from what I see from the photo's a main breaker could be added to the combiner panel. this is a good thing.
So from what I can see to fix your system the following comes to mind
1- replace inverters with the 600V rating inverters
2- disconnect one panel for a total of 39
3- add main breaker to sub panel@ 50 A and do a line tap connection. ( keep in mind this is considered a separate service and grounding requirements will prevail)
4-install a line tap connection to the main panel
5- Maybe add a revenue grade meter between the solar panel and the line taps. (Check with aggregator to confirm)

[websterize]

Rich, Matt Bartlett, who owns AtisSun, just called. They're on the inspection fail and seem committed to making everything right. Appears they'll do whatever needs to be done to make this right. Still awaiting details on the system design from his colleague, Mark.

After reading John Wile's PV Math piece on solar and cold weather, I think I get the Voc issue you've been flagging from the start. (Really do appreciate all your feedback -- this topic is getting a lot of views!). Running the spec sheet data on our 260W x 40 array, the Voc on each 260W panel is 37.9V, with a temperature coefficient of -0.36 Celsius. According to Solar America Board for Codes and Standards, the expected lowest temperature nearest Frederick is -14

[Naptown]

[QUOTE=websterize;44205]Rich, Matt Bartlett, who owns AtisSun, just called. They're on the inspection fail and seem committed to making everything right. Appears they'll do whatever needs to be done to make this right. Still awaiting details on the system design from his colleague, Mark.

After reading John Wile's PV Math piece on solar and cold weather, I think I get the Voc issue you've been flagging from the start. (Really do appreciate all your feedback -- this topic is getting a lot of views!). Running the spec sheet data on our 260W x 40 array, the Voc on each 260W panel is 37.9V, with a temperature coefficient of -0.36 Celsius. According to Solar America Board for Codes and Standards, the expected lowest temperature nearest Frederick is -14

[websterize]

But just to double check what is the Vmp on that module.

Vmmp is 30.9, according to the spec sheet.

The owner of the company that did the electrical work called. The SB3000US (500V) inverters are coming off, and three SB3000HF-US (600V) are on order. Apparently, there was confusion with the similar model number in the Sunny Boy line. (I think the HF inverters are Sunny Beam-ready!)

[Naptown]

Vmmp is 30.9, according to the spec sheet.

The owner of the company that did the electrical work called. The SB3000US (500V) inverters are coming off, and three SB3000HF-US (600V) are on order. Apparently, there was confusion with the similar model number in the Sunny Boy line. (I think the HF inverters are Sunny Beam-ready!)

You still have a problem. Maximum # of those modules on is 13. What happens to module 40. Disconnected?
At 14 on a string the odds of frying that inverter go way up or just pray the temperature stays warm.

They still have not come up with a workable solution that utilizes ALL the modules you purchased and paid for or if they do you run the risk of frying one of the inverters.
This is the easiest/cheapest solution for them as they probably have the strings set up in this manner.
For you who purchased and paid for the system to be installed properly you should insist that it at least be done properly.
I urge you to go to the SMA website. download Sunnydesign and plug in the numbers for yourself.
Atis Sun is not giving you good advice.
Admittedly changing to the Fronius inverter would be more work for them. But this is a workable and safe solution that will give you the full output of your system.
Remember I said earlier that inverters run at their most efficient when operating at near peak input. The Fronius is essentially a three stage inverter. Meaning there are three steps to the size meaning when power is reduced from the array the inverter will go to a lower stage keeping the efficiency at it's peak
In short more harvest

[websterize]

You still have a problem. Maximum # of those modules on is 13. What happens to module 40. Disconnected?

Yes, confirmed today: design is 13 modules per string, for a total of 39. The 40th panel was left unplugged deliberately. I guess I'd feel better about this if they had disclosed that from the start, but they did not. Granted, the array would look odd with three rows of 10 and one row of 9, but these panels aren't cheap. The fit and finish of the project turned out exactly as promised, but as you correctly pointed out, technically we paid for a 10.4 kW system, not a 10.14 kW system.

I urge you to go to the SMA website. download Sunnydesign and plug in the numbers for yourself.

I tried, weeks ago, but SMA marginalizes users of Mac OS X. I don't own a Windoze PC.

[QUOTE=Naptown;44223]Atis Sun is not giving you good advice. Admittedly changing to the Fronius inverter would be more work for them. But this is a workable and safe solution that will give you the full output of your system. Remember I said earlier that inverters run at their most efficient when operating at near peak input.

[Naptown]

Then it sounds like they are doing the right thing.
Consider the unplugged module as a spare for just in case.
40 x 255 = 10,200W
39 x 260 = 10,140W so with the increase in module wattage you in reality only lost 60W

Yes they are a major competitor in my market. That doesn't mean I am going to make up fairy tales and trash them for no reason. The 40 modules on 3 inverters is what caught my attention. The rest came from a quick look at the SMA site and doing the calculations.
As long as you are OK with the final product and satisfied then that is what matters.
As far as being skeptical that is a good thing it makes you dig into what a salesman tells you, although before you sign on the dotted line is a better time to do the research.
At this point the only things I could fault them for would be not telling you up front that there would be 1 unplugged module and the installed wattage would be less than promised. I understand that people make mistakes, it's how they correct them that matters.
The thing that got my goat was a statement you made in an earlier post to the effect of they scrutinize the engineering very carefully and the manufacturer also checks the integrity of the design. This obviously had not happened or was done incorrectly. Too many mis steps in the process for that to happen.
Are they still exceeding you expectations?

[websterize]

Having a commissioned Webster Solar Billboard by April 30 would exceed my expectations.

[Naptown]

Having a commissioned Webster Solar Billboard by April 30 would exceed my expectations.

don't count on it
If they had the inverters installed and inspected tomorrow Allegheny will take 3 weeks to install the net meter. If I had to venture a guess I would say late May to early June. (They have no control over the utility Co.)

[websterize]

We're rethinking the inverter swap -- replace mistakenly installed 3000US (500V) inverters with 3000HF-US (600V) model -- for two reasons. The dimensions of the HFs are different enough that affixing them in the same spot might not work. More importantly, we'd have to leave the 40th panel unplugged (3 x 13 panels).

Instead, we're considering two inverters: a 7000-US with 26 panels and a 3800-US with 14. That's one less inverter to replace, and 40 hot panels, for a 10.4 kW system. This seems to make the most sense.

Also confirmed with Solsystems, a local SREC aggregator, that to qualify to produce SRECs in Maryland, our >10 kW array will need a revenue-grade meter. Right again, Rich.

[Naptown]

We're rethinking the inverter swap -- replace mistakenly installed 3000US (500V) inverters with 3000HF-US (600V) model -- for two reasons. The dimensions of the HFs are different enough that affixing them in the same spot might not work. More importantly, we'd have to leave the 40th panel unplugged (3 x 13 panels).

Instead, we're considering two inverters: a 7000-US with 26 panels and a 3800-US with 14. That's one less inverter to replace, and 40 hot panels, for a 10.4 kW system. This seems to make the most sense.

Also confirmed with Solsystems, a local SREC aggregator, that to qualify to produce SRECs in Maryland, our >10 kW array will need a revenue-grade meter. Right again, Rich.

All inverters in the US are limited to input voltage of 600V
So the 3800 will handle the extra wattage of the 14th panel. Problem is the same as the 3000HFUS with 14 of those modules on a string the voltage is over 600V at your lowest record temperature.
The only viable option I found that will work is the Fronius IG10+
Because of the lower start voltage and stepped power levels you could string 10 modules on a string, use all 40 modules and down the road you will have only one inverter to replace. And if you act fairly quickly that can be extended to 20 years for about half the cost it would be normally.

Here is the crux of the problem
Atis Sun is a Schuco dealer. Schuco provides "packages" to the dealers. These are pre engineered package units designed for a dealer with no knowledge of how to engineer solar to enter into the business. Generally in pre determined configurations, with inverters chosen for the application. Not much thought on the dealers part required. Your's has gone outside of the pallet so to speak.
I do not know if the Fronius is available to them through Schuco I believe Schuco is strictly SMA.
The problem is that SMA does not make a product I could find that works for your application. The problems come in when determining voltages for the most part. 10 on a string is not enough voltage and 14 on one is too much.
The problem did not start with the unexpected change from a 255W panel to the 260's you got. The voltage characteristics of both modules are identical.
As far as the meter goes we are a platinum dealer for Sol Systems we know the requirements although they are the same for everyone regardless of aggregator.

[websterize]

Another scenario: stringing 13 panels to the 3800US and a 13 and 14 to the 7000US. Then we're within voltage spec on the 3800, and we can use all 40 panels. Now this seems to make the most sense.

[Naptown]

Bill
let me think about that one
I still think there would be voltage problem with the 14 modules on one string.
how it would react with the other string in parallel I will have to think about more.
With unequal strings there are going to be losses as they are not matched.