diy solar

diy solar

Adding storage to my Enphase system

I am almost done building up the 2 new 14S banks of Chevy Bolt modules. I was able to cut one of the 10S blocks into a pair of 4S blocks, and a spare 2S brick. They are mechanically bolted back together. I made up the connecting buss bars to electrically join the 10S and 4S sections. I made the clamps to attach the buss bar to the tab, but I still need to buy some more stainless steel 6-32 nut and bolts to affix the clamps. But with just the nut holding the buss bar on the 10S brick, it has some natural tension against the tab on the 4S block. So I was able to verify it is all wired correctly. I measured all of the cells and the total pack voltage. Every cell group measures dead on 4.12 volts. Because it is over 4 volts, my meter can't show the third digit. I may have to break out 5 digit Fluke 8060A, but I don't think it is completely necessary. Calculating 4.12 x 14 = 57.68 volts, and sure enough, both bricks measure 57.7 volts. Again, can't see that 4th digit, but I know it has to be very close. For giggles, I put a clip lead across the negative end of both packs. And when I measure from positive to positive, I will see the difference between the packs. When I first connected it, I was still in the normal "volts" range, and it displayed 0.000 not even 1 millivolt difference. I put the meter in the millivolt range. Have to be careful, as this can only measure to 300 mv. And I can finally see a slight imbalance. 0.6 mv or 0.0006 volts difference between the two 14S bricks. I will call that dead on balanced. IMG_9372.JPG
Total weight for 18 KWH of battery is right at 200 pounds. but now I need to find another enclosure to put these in. I still think I want to make these into 2 separate packs. Each one with have it's own BMS. I think I will get a case large enough to fit a 2,000 watt sine wave inverter and my 600 watt charger. That way, it will basically be a stand alone mega power bank. And then I will put a 150 amp Anderson connector for tying them into my main system. They will only add capacity. I will not expect my system to supply any more current as these will have a bit more resistance in the line to the inverter. When the current draw is low, they will still self balance to the same voltage. Unlike LFP cells, these do have a decent slope to the voltage, so they will still share the load fairly well and balance to very close to the same state of charge. I do need to get these puled down a bit as I never charge my current system all the way up to 57.7 volts (4.12 per cell).

Now to charge all of this battery, I need to get those 9 more 300 watt solar panels and a charge controller.
 
^^^^Amazing, beautiful work, Sir!!!

How did you go about determining the wiring/connection schematic for that battery pack?
There can't be any factory documentation about, did you have to probe around w/ the DMM to
make the schematic?
 
Last edited:
The most difficult part is figuring out how to cut the larger block down. I have to give credit to 400bird for showing me his cuts. Each pair of cells is in a plastic carrier with an aluminum plate for the cooling system in the car. But since the batteries are actually wired in 3P groups, it adds another later to figure out which 2 cell carriers can be separated. On my first pack I built last ear, I "cheated" a bit. Each Chevy Bolt actually has 8 10S blocks and 2 8S blocks. I was able to get 2 10s and 1 8 so I only cut the buss bas to electrically split the 8S into 2 4S blocks. Then just had to wire them together. But this time, I was only able to buy 10S blocks. Since there are more of them, they are a little easier to find. I could have done the same game, and only cut buss bars etc., but then what to do with the extra cells? Just leaving them in the pack unused is not a good idea. So I decided to try the cut.

You can see the pics earlier in my thread, and there are some over on 400bird's thread as well. Using an oscillating tool with a fine saw blade, you only need to cut the buss bar, the balance lead PC board, and then the plastic insert that holds all the buss bars and the PC board in place. There are two scary things about this operation. The first one is you will make a spark. I really tried to avoid it, but since the cell leads are on that PC board, any metal saw blade is bound to hit at least 2 of the traces. Luckily they are fused, so it just makes a little spark as it hits each trace. The second scare it knowing you are plunging a saw towards Li NMC "Flammable!" pouch cells. Thanks to his pictures, I had a very good idea about how much room there is, but you still can't help but be a little nervous. Work slow, let the tool do the cut, don't push. And you feel it pop out the back side of the plastic insert, and you are good. You have over 1/4 inch before you hit the cell, but if you are pushing, look out.

For me, the wiring is the easy part. The cells go across the width of the pack. Each group of 3 cells alternate with positive on one side, and the negative on the other. The factory buss bars join 3 positive tabs to the next group of 3 negative tabs. Due to how the packs are installed in the car, some of the packs have the negative end output stud on the left end, while others have it on the right end. So you do need to measure with a meter to see which way they are. The two 10S that I kept complete this time are opposite direction, with the mounting plate on the left end, one is negative left, the other is positive left. This actually worked out though. When I cut the 3rd 10S block, I took the left end that was positive, and just flipped it over and stacked it on the end of the positive right 10S block. The right end was negative, but it also had the mounting plate on that end. So the cell holders are actually facing the opposite direction from my other complete 10S block, even thought, they are both negative right. Again, I got a little lucky. This is the orientation that they sit in the car. They use a plastic spacer plate that sits in the middle and reverses the cell orientation. I still had two of those plastic spacers from my original cell blocks. But I did find out something interesting. My old blocks were LG Vista 2.0 type, while the new ones are LG N2.2 type. Even up close, they look identical, but... When I went to use my old spacer plate on the new cells, it is about 1/16th inch, or about 1 mm shorter in length than the newer cells. I had to trim out the square locating dowel sockets to make the spacer fit on the newer cell carriers. Now that all the cells fit together with all the polarity going the right directions, I just had to cut threaded rods to the right length. Or so I though.

The block with the spacer plate in the middle, and the GM/LG steel mounting plate at each end worked perfectly. Slide in the threaded rods, and torque down the nuts, all down, except for the one missing buss bar. The other one is a different story. The end cell away from the mounting bracket would normally go into the plastic separator. Luckily, I had one more. But it won't be going to another bank of cells. I got a heavy embossed steel plate to put on the end, and I cut out a piece of a thick plastic cutting board that is the size of a cell pouch to fill in the gap. It all looks good and the cells are all under proper compression.

And now I made the little buss bars to couple the output post at the end of the 10S bricks to the first tab on the 4S block. So the output is on factory 6mm studs on both packs, but on one of them, the negative stud ended up facing down instead of up. It is far enough from the bottom, it won't hit, even if the pack is set on a flat surface, but I plan to have it elevated a bit anyways.

My next step is soldering on all of the balance lead wires. And then I have to decide what BMS I want to get. I think I am just going to go with the basic 100 amp DALY for these two.
 
Looking good, glad the cells were so we'll balanced.

I agree in the nerve racking cut to split the pack. Slow and steady. I use a vacuum while performing the cut to help keep the mess and debris away from the cells.

I forgot to mention to cut the traces right next to your cut (on the cell side). I think there is one trace on the bottom of the board. But, if you cut all the traces on top. Too late, sorry


On a different subject: I just installed my battery temp sensor from the XW. I was going to set the temperature compensation to 0 as I believe that is appropriate. Have you installed the inverter's battery temp sensor? If so, what did you set the compensation?
 
I have the temp compensation at zero as well. I have not been able to find anything that says any different for Li NMC cells.

Well, my system got an off grid test today while I was at work though, so I did not get see it happen. When I got home, it had the log entry for grid volts too low. It happened at 3:21 PM so the solar was still putting out nicely. It looks like most of the Enphase inverters stayed running this time as it went to charging at nearly 23 amps while also powering everything in the backup loads panel. And it looks like frequency shift worked. The battery was very near full, and it topped up to absorb volts again, and then the charge current dropped to just 5 amps. And a few minutes later, with the battery topped out at my 57.4 volt setting, the charge current dropped all the way to zero. The solar was running all of the loads in the house all the way to 5:27 PM with no battery current. I do not have any log entry to tell when the grid came back, but it looks like it started exporting grid power around 6:43 PM. So the grid probably came back 5 minutes before that.

In any case, all my PC's are all still up and running, and the battery held up great. My only issue is the house is a little warm because I think the digital thermostat reset somehow. It was in cool mode, but at 78 degree F. I don't remember having it that high when I left for work.
 
I finished up the clamps for the buss bars. I did a light load check at just 4 amps and the voltage drop is less than 1/10 mv different than the other buss bars, so I think it's good. I will do a higher current test to make sure before it goes into full service. Here are a couple pics.
IMG_9377.JPGIMG_9376.JPGIMG_9375.JPG
The new copper bar is clamped to the factory copper bar, so the aluminum is not actually carrying any current. I will also coat the contact points in "Ox Guard" before final assembly. Still not sure what I want to use for a case for the two new packs. I may end up making a frame out of angle iron and screwing on panels. The shape is odd, so any pre made case big enough, is way too big in other directions.
 
I lightly sand the coper faces, and then put a little dielectric grease to keep out moisture as I assemble it. The aluminum clamps get a light coating of the Ox Guard. That is how I did my first pack, and the clamps still look like new after nearly a year, and they run stone cold.
 
When you talking about timing issues, is it that the gateway/XW times out if you don't send messages every 8 seconds or less?

Initially (and currently) I am sending charge wattage commands manually and only when I want to update the charge output. It will give a few different error messages while it "wakes up" the error messages make it sound like the gateway is opening the port and then readying to receive the command.

If I resend the same charge command over and over again, (at 10 second intervals it occasionally errors out) every 8 seconds I don't see any errors.

Once I finally get this all set up, I will likely bring the timer down from 8 seconds. But, with it slow moving like this I am more likely have the time to catch a cascading problem.
 
I lied, I still get time outs occasionally at 8 seconds. I switched to 5 second intervals and didn't see any errors for the rest of the afternoon.
 
I have not had time to work on it lately. On my PLC I have 4 custom functions I wrote in their T-Basic language. One opens the Modbus TCP channel. The next checks the connection status, the third reads the battery voltage location, and the last one closes the Modbus channel.

If I manually trigger each function, waiting a few second between, it works every time. I ask to connect, WAIT, check the connection, get the <connected> response. Ask for the battery volts, and I get a valid and correct response, an integer 1,000 times the actual voltage. The last digit stays a zero, so it is out to 10 mv resolution. And during the charge, I have checked it and it matches the data in the Insight web viewer, so I know it is reading the correct port. But if I try to read again, it just keeps returning the old value. I have to close the port, then open and check it again, then rad it again, and then I get the current new value.

Any time I try to put it into an automatic timed loop, it seems to fail. I think the Gateway and XW-Pro are just very slow, and even poling for the good connection response is causing a failure. Next time I get into it, I am going to try and just make it get the battery voltage every minute, waiting like 10 seconds between each step. If I get a bad connection result, I have to close and re-open the channel, or I get no data.

Hopefully I can work on it again this weekend.
 
So Cal Edison finally got their usage data working again. With my A/C cranking on these 100F+ days, I am consuming about 14 KWHs a day. My existing panels have been producing about 28-30 KWHs a day. So that puts my "sun hours" at 5.8 to 6.2 a day. To cover that extra 14 KWHs I would need about 14 / 5.8 = 2.4 more KW of solar panels. 9 x 300 = 2,700 so that should do it.

The extra panels on the garage and the extra battery capacity, I may be able to zero my bill even with the A/C running.
 
One of my co workers was asking me about solar power, knowing I have a system. I popped up my Enphase Enlighten app, and I noticed the 12:45 to 1PM bar was way down. Hmmm. The rest of the bars for today are right on the curve from yesterday. I figured maybe we had another power glitch. When I got home from work, I checked the Schneider XW-Pro event log, and it shows nothing wrong. Ok. I went into the Enphase Toolkit, and sure enough, it shows most of the iQ7 microinverters went offline for 5 minutes due to "Grid Instability". This is the first time the Enphase iQ7's have shut down without the Schneider XW-Pro showing any error at all. My guess is that the frequency error was smaller than the "ride through" range of the Rule-21 settings in the XW-Pro, so it stayed running just fine, but the iQ7's are still set in "mainland alternate" grid code, so they did not ride through, they disconnected. The weather is totally clear, but there have been a few smaller fires, and it is hot, so many people have their A/C running hard.
 
Work has gotten pretty busy, so progress on my additional battery bank has been a bit slow. I finally got the two Daly BMS units ordered and I completed the mounting frame to support the batteries in the new cabinet. I am working outside, and it is HOT!! Looking at my SCE usage, I definitely need the additional solar panels. With these crazy temps, solar production is down and the A/C compressor is being a total pig. I have been purchasing 20 to 30 KWHs from SCE each day to keep the house cool. That is on top of the 26 KWH my solar is producing. I am managing to keep the peak rate consumption down to under 2,000 watts by having the battery push back to the main panel, but to push any more, I would have to charge with grid power, and that is technically illegal. I am doing my best to only charge with excess solar, and that allows me to move 8 KWHs of solar to the peak rate time of use. With the extra panels I plan to add, I could nearly double that.
 
Hope the weather improves soon!
My heat pump compressor is also a big hit on the watts.
The funny thing is, after the initial start inrush, the compressor and air handler only need 2-3kw to run.

The LRA for the Copeland scroll compressor I measured @ 132A.
I thought by adding a second Schneider XW Pro 6848, I could start the compressor.
After reviewing a new document:
XW Pro Multi-unit Design Guide
990-91373A
April 2021

It is apparent the second inverter only adds a small amount of capacity, and I can’t find a spec on the inrush surge capacity for two inverters.

So plan “B”, I bought one of the soft start devices you mentioned.
SS1B16-32SN (230V, 60/50Hz, 16-32 FLA)

I contacted HyperEngineering, and they said the device would allow the compressor to start w/ the single inverter capacity of 52A.
They told me to remove the "hard start" kit I installed when I put in the new heat pump in 2017.
It was completly destroyed & non-functional anyway from the LRA load, and I now understand the difference between "hard start" and "soft start".
I have not installed the HyperEngineering device yet, not sure I completely understand the wiring diagram, and if I toast the heat pump compressor during hot weather, the wife will be unhappy.

So I’m waiting until the weather cools off to do the install. Might be in the 70’s here by December-January.
surestart_schematic copy.png
surestart_goodman_instructions copy.png
 
Last edited:
That is a bit of a confusing schematic. I assume "CM" means compressor motor. It only shows the compressor having GND, L1, and L2 as high voltage connections, and then 3 more low voltage wires going to the control board. The low voltage wires are only labeled as WH, YL, and BL. It looks like the capacitor may be inside of the CM housing. What is the make and model of the unit? Maybe there is a better diagram available.

My Carrier outdoor A/C unit is very simple, just a contactor, dual run capacitor, the compressor motor, and the fan motor.
 
I miss the simplicity of our old R-22 heat pump!
But after I had repaired/rebuilt it over 30+ years, it was time to move on.
The new compressor is a two stage, 4 ton Goodman DSZ16048.
Much more complicated!
Here is a photo of the schematic inside the unit, w/ call-outs.
goodman compressor schematic1 copy 2.jpg
 
I guess it is some good news that I have not updated here in a while. The reason is that my work is picking back up as the Covid-19 situation is improving a bit.

So my new battery modules have been sitting a lot, and they are still over 90% charged. I checked them again, and all of the cells are still sitting at 4.11 volts. I got the two Daly 100 amp 14S BMS units, a pair of CNN 100 amp fuses (these are meant for fork lifts, and they are properly rated for 80 volts DC. Almost as good as Class T at half the price), Two 200 amp disconnect switches, various buss bars and a lot of wire and connectors. And I found a decent Craftsman brand cabinet that is just large enough to hold it all. I built a frame and brackets to actually support the weight of the cells. The two strings of 3P14S cells weight in at just under 100 pounds each. The total weight of this beast is going to end up about 240 pounds. That's not too bad for 18 kilowatt hours of battery storage. It is far more compact than my first battery bank. It will be about 1/2 the total volume. But that smaller size is making assembly a bit more difficult. This setup is two completely separate battery banks. After the issue I had with my bad balance wire, I figured redundancy is a good thing. So each 14S string has it's own shut off switch, fuse, and BMS. Each string should take 100 amps of charge or discharge current on it's own if needed. The two systems don't join until they hit a pair of output buss bars to the 2/0 cable that goes to the inverter/charger.
Here is a look as it sits now.
NewBattPic1.JPG
The cabinet just fits around the frame. It has all been test fit, but I will need to glue the nuts on the inside in several places, because there will be no way to get hands in there to hold the nuts in place. All of the balance wires are connected to the cells. The lower pack wires are all tested, and I am about to test the upper pack now. I took a break after soldering all the leads. The cabinet is just going to be a cover. The frame holds all of the weight and the casters will bolt through the bottom of the cabinet. There will be about an inch around all sides of the batteries, and just over 1/2 inch between the two banks. I made quite a few changes from the design of my first battery bank. I am not leaving a lot of ventilation in this setup. My first battery runs dead cold, so I did not think air flow was that important. I can easily make an opening and add a fan if it does seem to warm up, but with both battery banks, the current to each one is going to be half. So I really do not think heat will be an issue at all.

Hopefully in a few days, it will all be in the cabinet, and I can post a final assembly pic.
 
Back
Top