Off grid new house build.

The number of kWh available changes by season so in Jan I should get 86% of 66kWh/kW * 25kW = 1419kWh.
With 5 degree tilt your loss will be greater in winter than in summer. You can put your lat/long into PVWatts and play around with different tilt angles. For Halifax and 135 SE / 5 degree tilt and the default 14% system loss it shows me (per kW of panels):


That 40/32/40 in Nov/Dec/Jan is your issue. 50 degree panel tilt gives a bit less annual power but 50% more in December:


You'll need to a better tool to study daily variation. There's a big difference between every December day hitting the 1.6 hour average vs. getting all 49 hours in a couple of 3 day sunny stretches separated by weeks of darkness.
This would be my concern with fixed flat panels (what doggy posted). Also, low slope panels will need to be cleaned. I’d plan for access.

Or a second set of panels oriented for winter max.

On the batteries, it sounds like you have a handle on the maintenance required. For me, it was fine at first. Then as the years went on, less so. I’ve got other things to do and the absolutely zero maintenance of lithium is nice.

This panel is on a 1:12 slope. The water pools at the bottom and collects pollen/dust. Some panels (very few) have a notch molded into the frame to let water out and minimize the issue.

btw-you are way ahead of the game by analyzing this ahead of time. We get a bunch of folks who buy first them come looking for advice.

If that's for winter you could easily compensate that 15% loss with more panels and would probably be cheaper than the racking involved at tilting it up.

No way it is for winter, with 45 degree latitude of Nova Scotia. His panels 5 degrees tilted will be almost edgewise to the sun on winter solstice - 23.5° south of the equator, so 68.5 off normal to the sun. Cos(68.5) = 0.367 as much light. (and then of course attenuated by long path through atmosphere.)

It's an efficiency chart resulting in ratios so it should apply to all seasons.

I don't think so. Cosine of an angle isn't linear.

Is my math OK?

Math maybe.
But 52 in January, 112 in July is likely light intensity, not considering angle. Or power harvesting by panel of a fixed angle oriented for maximum year round.
Your 5 degree tilt is sub-optimum in July (maybe 15% reduction which is OK as you say), but 64% reduction in winter.

The chart Doggy posted is more like what I expect.

Besides the inobtrusive panels on roof (maintaining good appearance of property), can you cover the WSW facing wall with panels?
(Probably not, given the forest I see which likely shades the wall in winter).

There’s a post here recently about a big install of vertical panels (Finland?). May be worth a look given your location.
There’s a post here recently about a big install of vertical panels (Finland?). May be worth a look given your location.

OK, got it. The PVWatt site asks for my tilt angle so it's probably summing it with the sun's angle before calculating COS.

Say I configure the PV to be 100 panels at 250W each for 25kW.

even in the short month of Dec I'll get 669kWh and my genny will need to make up 416kWh.

If I use all of my roof and put up 37.5kW of PV I can get 1003kWh

I think what I need to build now is a spreadsheet with a bell curve ven diagrams where I can put in cost per kW of PV, cost per kWh of battery and cost per kWh of diesel powered genny. Then amortize the whole thing over 15 years and see where the money is best spent.

The engineer I work with mentioned something interesting this morning. Digby NS is notoriously windy, there is a wind farm right across the bay from me. I may be able to generate power in the winter months with a wind turbine.

The results of only one day of audit pointed me directly to where all my power is being used. The business JRElectronics has a shop where we repair industrial circuit boards and the business also has a server rack full of old HP Proliant servers that run the phones, serve the finance software and keep archives, there is also a firewall and routers, switches etc.

The Business is accounting for more than half of the power we use and the server rack is the largest part of that. I need to upgrade my servers to more energy efficient for now and when I retire I won't need them at all anymore.

This means that I can run the house on 25-35kWh per day easily! I don't need 70-80kWh I've just been wasting that much.

I think THIS is using 35kWh per day!

Oh yeah! There's your problem. Back when I was in IT and I ran my own servers in my house, the furnace closet was right next to the server rack. When my furnace died on me I just tore it out and put a huge fan in the floor, cut some holes in the closet wall, installed some vent covers, and let the servers heat the whole house.

I think you'd be surprised just how much power you can save with even semi modern hardware over what's in those racks. I'm guessing old Xeon 2011's or P3's and SCSI drives in those racks? Ouch!
You also will be living on a wood lot. You could go the old fashioned carbon-neutral way, split firewood and let it dry over the summer to heat in the winter. Add ashes to garden compost.

I'm guessing old Xeon 2011's or P3's and SCSI drives in those racks? Ouch!

Xeon? My laptop has dual Xeon.
But I think it is graphics chips which burn the most.
I think THIS is using 35kWh per day!

Yeah, that'll do it... You can likely replace most of the functionality with a single dedicated server in a data center and upgrade that PBX to a small box.

Xeon? My laptop has dual Xeon.
But I think it is graphics chips which burn the most.

These servers likely don't even have a GPU. The Xeons (if that is indeed what they are running) on a server are an order of magnitude more power hungry than a laptop version - especially the older versions. Those SCSI drives, yeah, those are power hungry as well.
I wouldn’t enjoy a genny. Esp if there was a solar option.

In small wind, there is only sadness.
There will definitely be a wood stove in the living room to heat the common area. I've probably got a season or three just from downed trees that I'll clean up the first years. After that I'll have to decide if I want to drop my own trees or buy firewood pre-cut.

That doesn't affect electricity needs or generation but the more wood I use to heat the house the less propane I'll burn.

I'll start on the rack next month, I think I can get SSD drives for the Gen5 GL380 on the bottom and put everything on it. (PBX, Apache, Finance, Storage Archive, Desktop) I'll also swap out the top one for a PFSense firewall box that only draws about 25W.
Hi All

I've been off for the past couple of weeks doing an audit of my power requirements and trying to reduce my consumption by changing old servers for new. I'm now convinced I can survive on 35kWh per day.

My next step was to use the numbers from the PVWatts site to create a spreadsheet and try to calculate how much power I can generate and capture with a solar array and then add how much it will cost to generate shortfalls with a diesel genny in the darker months.

It's curious that on a 20 year lifespan (I'm 59) the most efficient system is NOT to completely cover the roof with panels. I get the lowest price per kWh at about 100 panels. This scenario has me running the diesel in Nov, Dec and Jan but not needing it the rest of the year. I can add more panels to get rid of the need for the genny but the price per kWh actually goes up because the extra panels are only required three months a year. and won't pay for themselves in 60months (20yrs*3months/yr). If I reduce the number of panels and run the genny more often the price goes up. The sweet spot seems to be 100.

Here are my assumptions - please LMK if I've estimated anything incorrectly.

Cost of initial system:

Solar Power Generation from 100 panels:
Diesel (@75 cents per kWh running cost)

100 panels x 250W = 25,000W (STC)
December 22kWh/day, a bit under 1 effective hour. Maybe accurate. That's where generator comes in.
$0.75/kWh for diesel, maybe accurate. I've heard $1/kWh total.

$20,000 for 25,000W pf PV is $0.80/W
You can probably do better here. 250W panels especially are older. Spending $0.80/W should get new premium panels.
You can get good older ones for $0.35/W or less.
Threads from a couple years ago had people paying similar price for brand new from China, not sure the prices today.

Used panels can be a bargain. Get top quality names, less likely to degrade. Closer to 200W/m^2 vs. 130W/m^2 will save on area, labor, hardware costs.

FLA batteries and ICE generator may be a poor match, if generator has to keep operating for 2 hours absorption. Better generator for bulk, PV for absorption middle of the day. The dollars you show is OK, but assumes optimal output is accepted by batteries.

Don't know what size battery, but one guy said he buys forklift battery direct from China for half the cost here.
I think FLA cycled 15% most nights can give 20 year life.
LiFePO4 server rack battery is getting quite attractive. Current Connected now offers 5x SOK server rack battery system for $8000, which can deliver 25kW and charge at 12kW. Lithium can charge at high rate from generator and doesn't need long slow absorption (but may need similar for balancing.)

I don't think you've quoted consumption by month. Servers won't care the season, not sure how much of total consumption is them. Can you throttle performance and use less power in winter?

I don't know which inverter, I would use Sunny Island especially for lead-acid. While my experience is short, some people say SI is very good at stretching life of FLA.

$1000 in charge controllers is low for 25kW.
Midnight Classic around $800 for 5kW
Victron 200/600 - is that $2000 for 10kW?

I use Sunny Boy GT PV AC coupled to SI. You might pick up used or new old stock for $0.10/W, $2500 for 25kW. I like PV strings of multiple orientations, could peak around 15kW for less GT PV or SCC capacity needed.

With system costing $3100/year and diesel $420/year, and batteries + PV being 60+% of total, seems like less PV and less battery but operating generator more could reduce cost of ownership.
That said, I favor over-paneling to make up for days with less sun. Cheapest PV panels (old, deteriorated, fewer W per unit area) can be had for $0.12/W, maybe less. If you can mount and wire cheaply, saves on fuel and saves on battery.
Those are pretty expensive 250W panels. Can you use bigger panels, like 320W or more or so? Cost per Watt tends to be lower with bigger panels. Still, $200 (CAD I assume?) for a 250W panel is quite high. Over here I pay the equivalent of 265 CAD for a 410W panel.

What was the reason again to use FLAs instead of LiFePO4?
Thanks for the tip, I'll have to look around, it sounds like I can do much better on the panel costs. I've got lots of room on the roof and I'm not set on any particular size or wattage panel so I can do some bargain hunting. If I reduce my spend on panels to $0.25 per watt and raise the number to 150 to cover my entire roof I can get the price per kWh down to $0.21, and almost completely eliminate the genny use.


Monthly consumption would be 35kWh*30=1050kWh.

I was thinking of a pair of the Schneider Connext SW4024. I like the idea of a pair so that I'm not dark if one fails. I've been working on Xantrex/Schneider inverters for years so I may be able to repair them if they fail and I can afford them at $3kCdn each. My power audit says that my peak monthly 15 minute draw is 8kW and an average around 4kW. I think having a pair of 4kW inverters would keep them cool but with enough power for everything required.

I think I want to DC couple the system, the panels can charge the batteries directly with MPTTs and the inverter chargers can interface between the batteries and the fusebox with the genny fed into the inverters as a backup for both extra power and for battery charging when necessary. I'll have to do some reading to understand why you suggest that one inverter is better with FLA than another.

The reason I want FLA is that I don't trust Li yet. I've been repairing inverters for RV's and industry applications for 20 years and I've seen first hand that Li batteries don't always live up to their hype. I've had to deal with many unhappy customers who take it out on me when the Li battery they bought 5 years ago has failed. They were told it would last 10-20 years and it didn't. Too many times I've heard someone say "I know the batteries are good, they're only XX years old" but they aren't and the company that they bought it from is gone in the wind. Now that's not to say that all Li batteries are like that, I certainly think that there are some good quality Li batteries but I can't afford Battle Born and I don't want to buy a fly by night brand on Amazon. I'll buy Li when I can afford to buy from a company has a factory in North America with hundreds of employees, I want the executives to have some skin in the game and shareholders to report to. I won't buy from a company that has an office in North America and a supply chain in China because it's just too easy for them to disappear. Rather than take a chance on cheap Li I'll stick with FLA. Another factor is familiarity, I'm used to FLA, I know how to clean them up and change the acid, I'm familiar with equalizing charges and I'll be retired and have time to maintain the water levels.

Lots to research but I should get back to work now and make some $$ so I can pay for all this...

Thanks for all your help!
Schneider should be good.
Victron too, though not UL listed.
SMA is good. Usually expensive, but still some liquidation deals thanks to DC Solar
Also Midnight.

"open box", actually says pull-out. So might lack battery temperature sensor, Ethernet patch cord, plugs for sensor and relay connections.

Sunny Island is particularly robust, good with lead-acid battery, generator, AC or DC coupling.
Using a battery shunt it can be informed of any DC coupled charging or loads.
If it has older firmware, should also work with Midnight Classic SCC and an adapter board to receive charge current data and control charge voltage (but not current). It can then adjust charge current from generator or AC coupled GT PV inverters to hit desired total charge current.

With AC coupling, SI only supplies starting surge and does charging during the day, or for loads greater than PV. At night it carries the loads. You can easily have 11.5kW of battery inverter and 24kW of GT PV inverter, supplies A/C and other loads while the sun shines. (Some model Schneider and others also now support AC coupling at least to some extent.)

I understand FLA is best maintained with a target charge current, so smart control of charging is best.
Schneider may do that with communication to SCC. Victron does that with Cerbo. Midnight does it with Wiz Bang Jr., SMA of course does that (unless DC coupled PV exceeds desired charge power.)

Also take a look at Midnight Rosie. That is from the guys who designed the orignal Trace, Outback, and others. It is high frequency, with 60 second surge rating. They also have newer SCC I think in the 10kW range.
Last edited:
Making those FLAs last 20 years is key to your economics. Also, solar companies usually budget inverter replacement after 10 years. But your experience with both and repair abilities give you the edge.

Thinking outside the box a bit, you might consider co-generation aka Combined Heat and Power (CHP). Honda developed a residential generator that also provides hot water. They sell NG and LPG versions in Japan, not sure about elsewhere. Others like Yanmar also have micro-CHP systems. Or you could rig something up with any water-cooled generator. Co-gen can cut your effective fuel cost 65-75%. That might tilt the economics and let you cut back to 15-20 kW of panels.

Low generator efficiency during FLA absorption sounds like a potential gotcha. Co-gen might also help with that, as low electrical efficiency could just mean more waste heat going into the water. Or not, I don't know enough about those units. Running at very low load is also bad for the generator. Using solar for absorption sounds great, but your winter solar may be too unreliable day-to-day for that. I don't know, maybe shut the main generator down and have a small unit that handles absorption? Sounds pretty awful, but would also give you a backup for times the main generator is down.