diy solar

diy solar

Adding storage to my Enphase system

Since I know my solar is very unlikely to give me a negative bill, I switched back to monthly billing. After just 4 months, I knew the yearly bill was not a good idea as my "True Up" was going to be over $500 last year, and I would much rather pay $50 a month, then get slammed with that at year end. But when I changed it, they said it could not change until the first year completed. My first year just ended in August. My first single month bill, just happened to be the highest consumption since the solar went in. On the yearly true up, they would bill the "non bypassable charges" each month, and it would also show the running total of the full year adding up. They would not accept any payment towards that though.
 
Always nice to know the power utility is willing to be accommodating to the customers they serve.
 
If I do add some DC solar later, maybe it would have been a better deal gettign one of the al in one solutions, but I doubt I will go that route. If the grid here does end up getting far more unstable with wild fires, earthquakes, etc., I may add 4 panels, and a cheap charge controller to just make sure the batteries always have some charge, even if the inverter has to shut down. But with a stable grid, the Enphase micros are more efficient powering my home with the battery just time shifting some power and waiting for a power failure. The greater round trip loss on energy going through the battery is a small portion of my energy as I am typically using 2/3 of my production as it is generated.

(Very long thread by now. I don't see a way to search just within a thread to see if this was discussed.)

How does your system behave during grid failure if batteries become discharged?

The original (DC coupled) off-grid PV systems of course just charge batteries from PV, and if voltage isn't below low voltage cut-off, inverter can make AC.

For my Sunny Island, I have the recommended "load shed" relay, which disconnects house from system at 70% DoD (for AGM) so enough battery remains to provide AC for Sunny Boy PV inverters to work. In that case it waits for charge to reach 50% before reconnecting loads. DC coupling is another way, but without load shed, SoC would reach the lower level before shutdown.
I plan to look into "battery preservation mode". Sunny Island consumes 25W (each) operating, 4W standby. With batteries drained too low, it would shut off, only powering up once every two hours during daylight to see if AC coupled PV provides power for charging.

When grid returns, does your XW recharge batteries from grid (or generator, if you have one) even if battery voltage got so low it shut off? Unless you're sure your loads will never leave battery drained in the morning, I'd think you'd want to put in that DC coupled PV so it works reliably during blackouts. Schneider (Trace/Xantrex) started out with DC coupling. Instead of a cheap charge controller, you could use one that talks to the XW, maybe get additional PV feeding the grid as well. You've got roof space, and panels are so much cheaper than whatever surprise the utility gives you each month or each year.
 
So far I have only done one short off grid test, and I had an issue where most of my Enphase micros went offline during the transition. 5 stayed up and charge current went to about 600 watts. I only had about 200 watts of load on the output at the time, the 5 working inverter were making about 800 watts. I did not leave it off grid long enough to run the batteries low. I am waiting for word back from Enphase on why the inverters locked out. The 5 that stayed working did not even glitch, they just kept producing with no log entry. The 11 that quit all said it was a grid frequency error. They should have gone back on after 5 minutes, I gave them close to 20, and still no good. Even after going back on grid they did not recover. And once on grid I cut the solar disconnect and turned it back on, and the same 5 plus one more came up, the other 10 didn't. I did get 3 more to come up by covering the panel to cut the DC going in, and then they came up. The other 7 dead ones were on my upper roof. They did come up the next morning after they went dark by nature.

So this was a bad test, Enphase is searching the logs, and they did admit seeing this happen before, it does seem to be a problem with the Rule 21 grid code.

I do not have much load moved into my backup panel yet for various reasons. Mainly, I am having trouble tracing the circuits, and I will have to cut house power, which means shutting down a bunch of delicate gear. I had not realized that the original solar installer moved all of my breakers, and didn't bother transferring the labels. So they are all marked wrong, and using an electronic tracer has failed. Some device in my home is putting out a noise that evidently looks like the tracer signal, causing it to keep reporting multiple breakers, even without the tracker source connected. Once I have a handle on how the loads are handled, I will decide if I need to add a load shed relay. From my calculation, I should have enough power to keep the loads up since I am not moving over the A/C or any of my shop tools. My worst case load is the Furnace blower and controls, which can hit 8 amps at 120 volt. And my old refrigerator which runs at about 5 amps at 120 volt. Beyond those, I have my main PC along with a small NAS, internet modem, router, and 2 switches. Total about 300 watts if I am not playing a graphic game. And I don't have a very serious graphics card in this machine. My second PC, which will be on the inverter does have an nvidia 660 card, is sitting idle most of the time, acting as a file store and printer server, but is used as an entertainment machine if we know we have the power to run it. And then I have one flat screen TV and the Dish Network Hopper box so we can still get new etc. if the internet goes down here. Dish coming from a satellite will keep running even if all local services quit. So until I get a solid check on the power use, I "think" I can run from 7 pm to 9 am and have some power left, but it may be tight. My current plan is to set the low battery cut off at 47 volts. If I get down there, it will cut, but then I can change it down to 44 volts, and get the system back up when there is sun. I do have a local UPS on the main PC, and I will set up a low battery alarm so I have some warning before the inverter cuts off. By my rough math, it looks like my chosen loads are going to only need about 12 KWH's and my battery should be able to supply up to 17 KWH's. To get that much power though, when I go off grid, I do need to raise my full charge voltage up to 58.5, almost 4.2 per cell, and my low batt cutoff down to 42 volts which is 3.0 volts per cell. The LG data sheet says they can go down to 2.8 volts without damage, but that is more than I want to risk.

It "Should" start up and charge from grid as long as the battery voltage is about 40 even. Worst case, I do have a manual generator and a 10 amp 14S charger I can hook up to get the battery up a bit. Once the XW-Pro will turn on, I also installed the generator socket on the XW-Pro AC2 input. I could then charge from the generator at up to 4,500 watts. That would be in the 90 amp range which is still less than 0.3C and the cells are rated to 1C continuous charge rate. The generator could go from dead to full charge in under 4 hours. But that will take nearly 4 gallons of gasoline. Any solar power I can get will help that out.

During an average day, my solar should make about 20 KWH's. Even yesterday it topped 24 KWH's. While the sun is up, my gear will consume 8 KWH's while the batteries get 12 KWH's of charge. Then during the 14 hours of dark, my gear will pull 11 KWH's. Yes, it is close if I don't shut anything down. I am looking at adding 4 panels on the garage roof. Still torn on if I should DC couple them. The Schneider MPPT units are stupid expensive though. Their smaller unit MPPT 60 150 is about $550. If I use 4 385 watt panels, I could run them in series parallel and get about 30 amps or charge current. Figure about $0.60 per watt for the panels, so $1,000 for the panels. Then add in racking and cabling, and what do I need to do about rapid shut down? I am not worried about it enough yet until I see how it works with my loads moved over and Enphase gives me a firmware update to correct the lock out issue when it goes off grid.
 
A few small updates.

Schneider is returning my e-mails!!!
They have not been able to connect into my gateway yet, so no changes were made, nothing to explain why it started dropping off network, but he took not that it has not gone off network since I switched it to a wires Ethernet connection. A but a 5 port switch on the board near the XW, and I will also connect the Enphase Envoy and my PLC to it, leaving one free port for a PC when I service it, or if I have to make a change with the power out.

It is so nice to be getting good sun again, 2 days in a row, my solar has topped 24 KWH's again. And the panels are still very dirty. I will try and get up to hose them off again when the sun goes down tonight.

Walked by my SCE power meter just now, and my power consumption was 0.0 KW. My solar is already making almost 2 KW at 9:30 am and I have the battery charging at 1,100 watts, and the house is using 900 so nothing going back out on the grid. The LCD segments that normally simulate the spinning disk are all blank. I have never seen it sit at zero power before. But now 10 minutes later, a bit more sun, and I am going negative. I have drawn up the flow chart and I am working on the code for my PLC. I am going to have it check the grid watts and adjust the charge power and grid sell to keep the grid watts close to zero. With my normal house load all on the main panel right now, I actually have to "grid sell" about 900 watts to cover the PC's and lights in the main panel. My plan is once every 5 minutes, it will check the grid power. Only if it is 500 watts too high or too low will it make an adjustment. Until I add a true grid power meter input, I may just add 1,000 or 1,500 watts when the A/C compressor is running. That obviously will not match the 3,000 watts it really takes, but it will sure help smooth out the swinging from 0 to 3,000 watts. I found a fairly cheap current sense relay that I can just put around the lead from the 40 amp A/C breaker. It closes the contact at the set current. The PLC can see that to make the adjustment to the inverter.


About 1/10 the cost of the WattNode
 
Interesting idea with the current sense relay. The AC draw should be pretty stable. You could probably get close to matching it to guestimate bit out to 0. But, you're already programming the PLC, why not source a different meter and add it to your device?

On the ashy panels: I cleaned mine yesterday and the only output difference was in the first hour after cleaning. Saturday and today we're within 0.6kwh of Sunday when I hosed them down.
 
The real power meters run abut $400 to be able to tell me both power going in and out. I can find a current only, but I won't know if it is going in or out of the house. I did find one super cheap, about $40 but it is only a single phase, I would need 2 of the for the split phase and then add the 2 wattage readings. Still not bad under $100 total, but I have not found their data read details. They come with a PC app to graph the power data. One of them says it is Modbus, but it has an odd interface. TTL voltage with separate send and receive lines. One is balanced bidirectional RS-485 data and it might work. It did list a few of the data registers, and I am looking into it. 2 of these is still 1/4 the cost of the WattNode. The first add I found did not come with CT's. Then I found with a solid CT, I would have to disconnect the main wired from the meter, No Thanks!. I did finally find it with clamp on CT's. for a few dollars more. My coding skill is a bit off, I am still working on the automated routines. The communication is quite slow so I keep getting a read error. If I manually send the connect, wait a few seconds, then check the connection, I get a <connected ok> response, then read the data and it works. But when I have the PLC run the commands together it is too quick and errors without a reply. But I ask again, and the connection is good, finally. 2 seconds it a long time in a computer. I can't just make it wait, it has to keep checking the main loop and keep looking for the status reply message.

I hit over 24 KWH Sat and Sun, but today only 23 KWH. Not sure why it was down, the sky was clear and it was about the same air temp. I want to clean my panels again, by the looks, I am guessing I'll gain about 8-10 % with this thin uneven dust layer. So if I do it this week, my guess is 26 KWH on a clear day. I can't compare off of last year as I still had the palm trees killing the lower panels. but looking at a few panels on the upper roof that did not get shade from the palm trees, they were doing about 1.4 KWH's per panel for the same day last year. This year those same panels topped 1.5 KWH's, so I am guessing it was a bit overcast as the panels were still close to new this date last year.
 
I just disconnected my Sense meter and will be selling it on Ebay. I love my Neurio because it interfaces with IFTTT which is the kind of programming that I can handle. The Neurio used to be the current sensing device for the Tesla Powerwall but now is part of Generac who continues to support it as PWR view.
 
Do you have info on the Sense meter? Could my PLC read the data? How much do you want for it?
 
We have still been hitting 90's here, but it has been getting cool overnight so I can "pre cool" the house and not need near as much A/C during the day. Here are my previous 2 days of usage.
SCEusage9-21.PNG SCEusage9-22.PNG
I am cheating a bit. Since I am home a lot, I have been manually tweaking the charge current and "grid sell" power a little to get an idea of how I want the PLC to do it. It will actually do a much better job as my code should be able to make an adjustment every 2 minutes. I also did order the current sense switch so it will be able to adjust when the A/C is running or off. So it if can do as good as I am doing here, you can see the high time of use cost is very small now. And with better tracking, I am sure I can do even better. I am still only using 8-10 KWH's of the battery bank to do this.

When I have less than an hour of sun left tonight, I may try to do another power fail test. But this time, I want to have some more load on the inverter. I think the voltage took a healthy jump last time with only about 100 watts of load on it. I thought the fan I had in there was more, but I was wrong.

Here are a few more pics of the battery and BMS. These show the holes I drilled for the Bluetooth, and my splice from the 2 #7 wires to my 2/0 cable. And also the cleaned up BMS wiring on the battery bank and the inverter cabling.
IMG_8502.JPG IMG_8503.JPG IMG_8504.JPG IMG_8505.JPG
Here is the BMS screen from my phone as the battery was almost topped out. 98% 355.6 Amp Hours remaining.
IMG_8495.PNG
Absorption charge still going at just 1.4 amps. The BMS is seeing it as just 57.1 volts, so I can raise it a bit more, but I just don't need to. I am only using abut 50% each night, so I am doing fine. My only problems are getting the PLC to control it all, and I need more solar.

I did do a quick rinse off of the panels last night. They are certainly not super clean, but it did raise my output about 9% over yesterday's numbers. It might hit 25 KWH's which is very good for only 16 x 300 watt panels at this time of year here.
 
So the last few days, I have been around my PC a bit as I am working on the PLC code. It is not automated at all yet, and I am still just looking in an adjusting values manually. I was trying to think what I would like the PLC to do. Obviously, it can take a peak fairly often and make trims, I only made a change about every hour, but my goal was to zero out my usage where possible. I did go out for lunch, so I was not here to adjust the charge rate when the A/C came on between 1 and 2 PM so there was some extra power used there as it was still charging at 34 amps (at 56 volts is 1900 watts) along with the A/C compressor (14 amps at 240 = 3360 watts) and the solar in the same time was only putting out 3400 watts. So here is what the battery current and voltage graph looks like.
XWbatt9-23.PNG
And here is what my usage from SCE was for the same day
SCEusage9-23.PNG
The A/C ran pretty consistent from 1 PM until about 6 30 PM. I pushed out a little too much power in the 5 PM slot, and at 7 pm I reduced the grid sell, but as the A/C ran less I still pushed out a bit more, but other than the 5PM slot I kept the export under 1,000 watts.

I think I will have the PLC watch my power and make adjustments about every 5 minutes. It should be able to do a lot better than my few number of changes on this day. Having to buy only 11 KWH's is great, and even better, they have to discount me for about 3 KWH's at the higher rate.

I am not getting something for nothing. If I didn't have the XW-Pro and the battery bank, I would have purchased even less power from SCE, but it would have been mostly the high rate power, and it would have been selling them a lot more power at the cheap rate. If I figure in the round trip efficiency, I lost about 1KWH to the efficiency of running the XW, charging the battery bank, and then inverting that back and pushing it to the grid. And to be completely honest, this particular day was a little bit of a cheat. I did export a little more from the battery than I charged into it. So I did steal a bit of energy from the day before, and I had to put it back the next day. I will try to make it limit the sell to about what it charged, but that is a bit tougher to keep track of on the fly. For this particular day, I only charged 8.836 KWH's into the battery. But it had to stop as it was then full. Well, full to me 57.6 set voltage. I could push in a bit more but I am being easy on the batteries. But then I sold out 9.132 KWH's and the ending battery voltage was just 50.29 volts. The starting voltage was 50.71 volts before it started charging that morning. Those numbers are after the battery was allowed to rest for 2 hours, so I should be able to get a decent state of charge from it. The ending SoC looks like it was close to 53% while the beginning SoC was more like 56%. Only 3% different?? It looks worse than that on the graph, but it makes sense by the math. 3% of 18 KWH is about 540 watt hours, that I took from the day before. So far, I have had no problem getting the battery to 99% SoC on less than my solar production before 4 PM every day. If I end up using a bit more than the day's charge cycle in the evening, it just means a little more on the charge cycle the next morning to get the battery topped back up. I just want to keep a solid 50% in the battery in case we lose power overnight and I have to run on battery until the sun comes up. To be truly stable, with all of my resting house loads, I do need about 10 KWH from 9 PM until 7 AM as the solar starts to make power. So 50% SoC won't cut it, unless I lose a little more load. My old refrigerator is a PIG! That compressor never seems to shut off. I am going to put a Kill-a-Watt on it and see if it is as bad as I think.
 
...unless I lose a little more load. My old refrigerator is a PIG! That compressor never seems to shut off. I am going to put a Kill-a-Watt on it and see if it is as bad as I think....
Should it come down to it, you can sort the energy star gov site data by various criteria to find the hyper-efficient models.
 
Also check into refrigerant recharge kits.
I plan that for my upright freezer than never reaches its setpoint.
(I'll take reliability over efficiency any day.)
 
I'll decide on the fridge based on what the Kill-O-Watt tells me. I know I have a 20% coupon here somewhere.

I got my thermal conductive double sided tape today, so I slapped the heat sink on the BMS.
IMG_8517.JPG
Even with no fan on it, it dropped the mosfet temp by 5 degrees C in a few minutes with it charging at 40 amps. In the rack cabinet, this should place the heat pipe fins right under one of the fans in the cabinet.

I also received my 24 volt converter, so I am going to have a panel with sockets for 12 volt up to 25 amps, 24 volt up to 10 amps, and battery direct 48 (50-58) volts fused at 30 amps. I will use the 12 and 24 to power the Gateway, the network switch, and the PLC, so even if the inverter has to shut down, they will have power unless a cell drops all the way to 3.1 volts. The only other thing I was thinking of adding to my battery bank chassis was a small 48 volt input 120 volt inverter. But there are not many small 48 volt inverters that are decent build quality and not stupid expensive. I am just thinking it would be nice to have a bit of 120 AC available to charge my laptop if the inverter has to shut down. Has anybody used this brand? Is it total junk?
Or do I bite the bullet and get this
 
...Even with no fan on it, it dropped the mosfet temp by 5 degrees C in a few minutes with it charging at 40 amps....
I saw a theory once that if you let the heat build up eventually it'll get so hot no more heat can't transfer into it. At that point it'll be more efficient as there will be no lost energy being converted into heat. ? It's wrong of course, but thought you might appreciate the humor. You're really chipping away at this... it's inspiring me to stop being a slug and do some stuff on my laundry list... or at least go ?
 
Did you get that heat sink from an old computer? I saw you say something about that before.

I am also interested in a cheap inverter that would not be used much.

I'll decide on the fridge based on what the Kill-O-Watt tells me. I know I have a 20% coupon here somewhere.

I got my thermal conductive double sided tape today, so I slapped the heat sink on the BMS.
View attachment 23710
Even with no fan on it, it dropped the mosfet temp by 5 degrees C in a few minutes with it charging at 40 amps. In the rack cabinet, this should place the heat pipe fins right under one of the fans in the cabinet.

I also received my 24 volt converter, so I am going to have a panel with sockets for 12 volt up to 25 amps, 24 volt up to 10 amps, and battery direct 48 (50-58) volts fused at 30 amps. I will use the 12 and 24 to power the Gateway, the network switch, and the PLC, so even if the inverter has to shut down, they will have power unless a cell drops all the way to 3.1 volts. The only other thing I was thinking of adding to my battery bank chassis was a small 48 volt input 120 volt inverter. But there are not many small 48 volt inverters that are decent build quality and not stupid expensive. I am just thinking it would be nice to have a bit of 120 AC available to charge my laptop if the inverter has to shut down. Has anybody used this brand? Is it total junk?
Or do I bite the bullet and get this
 
am also interested in a cheap inverter that would not be used much.
I bought a 700 Watt Renogy for less than $200. My main reason for picking the Renogy was not price but the fact that it could handle 16 volts because I wanted to use it with some Nissan Leaf modules which can only be incremented in steps of 7.5-8.0 volts. It ran an inverter driven refrig in my garage just fine.
 
The only other thing I was thinking of adding to my battery bank chassis was a small 48 volt input 120 volt inverter. But there are not many small 48 volt inverters that are decent build quality and not stupid expensive. I am just thinking it would be nice to have a bit of 120 AC available to charge my laptop if the inverter has to shut down. Has anybody used this brand? Is it total junk?

I haven't used that brand, but they do have a smaller 1000 watt inverter listed also. Should still plenty to power the laptop.

If you just need to power a laptop, what about using a 12vdc to 110vac off the buck converter you already have? It's not remotely efficient to convert the voltage twice. But a 200 watt inverter is cheap and plenty to power a laptop.


I agree with Svetz, your inspiring me to get off myass and start into this project.
 
Back
Top