Jack's Off-grid 198Ah 24V LiFePO4 Powerwall

I live at the end of 8 miles of dirt road, and that'show far away the nearest utility line is. This December, my four-year-old Deka GC-15 12V 230Ahbattery bank wasn't cutting it near the solstice, and I looked at my options, determined not to be in the dark next winter.

For starters,I upgraded from the Morningstar TS-45 (PWM)without the optional RTS, which I never bought, becauseof themismatchbetween the two24V GT panels and the 12Vbattery bank,to the TS-MPPT-60 with RTS and remote Battery Sense. I figure the Lead Weight could last a few months or even more than a year, but even withthe new charge controller assembly, it's obvious that it's days are numbered.

When I asked my sister's boyfriend for some feedback about my strategy, he suggested a DIY powerwall and offered that he knows someone who could get me salvaged 18650s at no cost. I wasn't comfortable with using used laptop batteries, due to both the chemistryand the headache of matching used cells. After doing some digging, learning everything that I could about LiFePO4 and finding some discounted new, old stock,5500mAh 32650 cells from a liquidation company on ebay, I decided to go for it.

UPS was very confused about my address, because it wasn't in their system, so after the packages had been in town for almost two days, I drove through the mud to pick up the 320 cells in town.

The battery bank will be 72p4s until I get around to shopping for a 24V inverter, at which point I'll rewire it for 36p8s.

I've tested a sample of cells for capacity individually and in parallel, and I'm pleased with the results. I'm going to run one more capacity test at C/20, to confirm a theory that I have about the Peukert exponent of these cells vs. their rated capacity.After I run to town for some hardware and order fuses and contacts, itwill then be time to put it together.

I'll use the Lead Weight as a buffer for the Powerlab 8. I could charge directly from the TS-MPPT-60. However, sunlight isn't constant, and since I already have the Powerlab for testing the cells, I might as well use it for charging.

I have tested the electrical contact resistance of Keystone Electronics 1016-1 gold-plated battery contacts in my test bed, and I'm comfortable that the resistance can be balanced well enough to use them for the battery assembly. Having removable cells will make me more comfortable, by making it easier to tinker with the battery.

Here's my 10p test rig.




This photo shows the balanced parallel wiring, which is something that I haven't seen or discussed regarding DIY battery packs.




The battery bank will be wired for at least 125A. I'm going to experiment with a mostly solderless assembly, because sordering almost 600 battery contacts to bus bars doesn't sound like fun to me. It will have fuses onindividual cells, parallel modules and the pack. It will also have a Trimetric TM-2030 battery monitor to monitor the charge. I bought the TM-2030 when I bought the TS-MPPT-60, so I already have that, too.

I only use about 30Ah at 12V daily and don't have any high-power electric appliances, so I'm considering charging the battery intermittently and using the solar power for other things, such as a hot water heater.

When I was shopping for the new charge controller, I didn't imagine that I would be building anything like this, but I'm very excited about it.

Welcome to the club. And thanks for bringing us along with your build

I like that your going the solderless (on the cells) route seeing as how that's my path I'm wanting to take as well. Altho, i was looking at it from the standpoint of easier to replace a cell if it goes down for what ever reason.

Look forward to the updates.

Korishan said:

Welcome to the club. And thanks for bringing us along with your build

I like that your going the solderless (on the cells) route seeing as how that's my path I'm wanting to take as well. Altho, i was looking at it from the standpoint of easier to replace a cell if it goes down for what ever reason.

Look forward to the updates.

Click to expand...

Thank ye. That was exactly my thought about replacing bad cells and part of what I meant by tinkering. After looking at the minimal voltage drop, I'm comfortable with the electrical contact resistance. The loss is so small that soldering or welding seems completely unnecessary for a stationary application .

I was thinking about how to arrange the cells in standard electrical enclosure of an available size. I came up with the Battery Sandwich.





The envelopes represent plastic sheets. Theclothespins represent bolts, which can be fastened with nuts, to hold the cells in place with the contacts under uniform compression.





(Source:http://www.powerstream.com/BPD.htm)

Staggering the meat between the bread maximizes space.

With n=18, p=4 and d=32mm, there are 72 cells and the footprint is 592mm x 115.136mm (23.31" x 4.53").

If height of the cell is 70mm, risers are 17.88mm and dividers (sheets) are 6.35mm, with 4 stacked arrays, height is 454.75mm (17.9").

So it fits in an enclosure of 24" x 24" x 6", which is an available size of electrical enclosure.

There's 72p4s or36p8s in a standard 24x24x6 electrical enclosure. The sandwich can be assembled on the wall, with enough room in the box to fasten the top slice of bread.

Comments?

---

Note: The height of the contact under compression was included in the height of the riser.

Hahah, nice. If you're going to keep your packs small, then being packed tight probably won't be a big issue. However, it is possible that you could generate some excess heat with that tight of a packing. Since you're not soldering it all together, it'll at least be easy to redesign if something starts to get a little hairy along the way.

Korishan said:

Hahah, nice. If you're going to keep your packs small, then being packed tight probably won't be a big issue. However, it is possible that you could generate some excess heat with that tight of a packing. Since you're not soldering it all together, it'll at least be easy to redesign if something starts to get a little hairy along the way.

Click to expand...

That might be a problem with hot cells, but these cells are cool. I haven't tested them at more than 0.2C, but I don't have a need for more than that.

Also, in reading about building battery packs, I found a theory that tightly packed cells will conduct heat more quickly than loosely packed cells.

Also, in building a heat shield for my wood stove, I found a design that transfers heat very efficiently and complies with Code. The spaces between the bricks allows cooler air to be drawn up as heat rises. The wood paneling behind the shield stays cool. Surely, with some lateral thinking, I can take something from this to adequately ventilate the minimal heat that will be generated by the battery. There will be enough space between the cells and between the cells and the risers to transfer plenty of heat up and out.

Simplifying the Battery Sandwich reduces materials and improves ventilation. If the lengths are closeenough, the nuts and bolts won't be necessary either. Break out the sandpaper.

Am I guessing correctly that the cells will be standing up in that sandwich? If so, are you only have 2 rows, or will there be more frames?

I'm wondering if I should stick with wood for the scaffolding or use another material. Wood is easy to find. Some other things require ordering online, which racks up the shipping costs. What can be found at Depot or Ace?

---

Korishan said:

Am I guessing correctly that the cells will be standing up in that sandwich? If so, are you only have 2 rows, or will there be more frames?

Click to expand...

Yes! The cells should be upright. It's not a big deal, due to minimal risk, but the vents really should be up.

Each leVelda will be 18x4 cells. The mockup isn't too scale. I'm just playing with ideas, and it isn't practical to build it all up to develop and convey the concept.

---

LeVelda = level. This Android keyboard is such a hassle in forums.

LOL but i like LeVelda


Do a google on LeVelda and ya get some interesting results.

Korishan said:

LOL but i like LeVelda


Do a google on LeVelda and ya get some interesting results.

Click to expand...

Velda is a friend. Autouncorrect squealed on me.

This weekend, I had to beef up the wiring, before it warmed up too much. When I upgraded the controller and moved everything inside from the porch, I realized that I didn't buy enough #4 wire or any #4 ring terminals. I used #10. I knew it was wrong, but I knew the limits and used what I had. Now it's all wired properly, and after four years without circuit protection I finally have circuit breakers.

Any more news on these LifePo4 cells? I have a set that I am testing and so far so good. I have used A123s in the past and they have been great. I want try these cells on a 1k watt off grid solar to power my garage and possibly run my pool pump.

I'm back! It was a crazy year. After a road trip to New Jersey, lots of yard work and chaos in the neighborhood, it's time to get back to this project. A few weeks ago, I killed my FLA bank by neglecting towaterit for too long. The capacity has been diminished and probably won't get me through the winter, so the PowerWall has become a priority.

I've done some research and calculations on fusible link wire and need to experiment with that.

Alright, I kludge together a 4-wire system and the PL8. Unfortunately, I couldn't manage to wedge the leads between the cell and the contacts, so I wedged them between the contacts and the frame of the test bed. Though this isn't ideal, it's the best that I could manage.

I'm currently running a capacity test on a single cell. PL8 is reporting internal resistance at about 22.5m?, whichis far less than my DC measurement of 90m? across the contacts. Previous reading about PL8 Ri tests results had prepared me to see this kind of difference between the DC test and the PL8 measurements.

Yesterday, I ran the first ccapacity test with the 4-wire Kelvin system on the PL8. On a single cell, from 3.65V to 2.5V at 1A, it gave me 5093mAh.

My records show that I didn't do a test for that range on a single cell, because I was testing between the knees. For 10p, from 3.65V to 2.5V at 11A, I got 49029mAh, with only the banana plugs connected.

I found the answer to how PL8 calculates IR.

The FMA PL8 measures IR during charge only. It does this by taking a rest every 50 mS to check on charge progress. It does the same while discharging as well but does not measure IR during a discharge. During the rest period according to FMA, IR is measured with the voltage rise difference between full charge current and zero amps. The current is turned off for 300uS and the at rest battery voltage is compared to the charging battery voltage on a cell by cell level.

Source:FMA viarcgroups

Hi there.

The50m? internal resistance figure seems rather high for that type of cell. I would have expected it to be closer to single digits.

What I really want to know is: where is that sweet spot that's 8 miles away from your closest grid connection? Are there any restrictions with your county/state over living in a building that's off-grid? Most counties will not recognize you for certain services, even if they were available, if you're not on the electric grid. Things like public water works, postal service, etc could be limited.

Back on topic:I, too, ordered a box of those LiFePo4 cells off of eBay with the intent of running some tests. If all goes well, I plan to buy quite a few of them so I can be as much 'off-grid' as I can.

Regards,
JR

JRoque said:

Hi there.

The50m? internal resistance figure seems rather high for that type of cell. I would have expected it to be closer to single digits.

What I really want to know is: where is that sweet spot that's 8 miles away from your closest grid connection? Are there any restrictions with your county/state over living in a building that's off-grid? Most counties will not recognize you for certain services, even if they were available, if you're not on the electric grid. Things like public water works, postal service, etc could be limited.

Back on topic:I, too, ordered a box of those LiFePo4 cells off of eBay with the intent of running some tests. If all goes well, I plan to buy quite a few of them so I can be as much 'off-grid' as I can.

Regards,
JR

Click to expand...

Hi,

I measured the DC internal resistance. Are you thinking of the 1000Hz (or is it kHz) AC impedance? PL8 measured only 22m?, using the method that I found documented and mentioned in a previous update.

County doesn't grade the roads. County doesn't pick up trash. County doesn't really bother us. And County doesn't even want to come out here. We don't get much support from County, but the other day my neighbor and I watched a Sheriff's deputy and a dog catcher chase two bulls on her fully fenced property.

Hi. 50m? isn't bad at all, especially forthis type of application. Newer variations of the chemistry and manufacturing processes of LiFePO4 have lower internal resistance. But again, not really all that useful in powerwall applications.

For this lowish C-rate discharge reason, you might be better off with a higher voltage pack. At 48V, instead of the planned 24V - if you can find an inverter for it - discharge rate with halved. It can alsosave you on connectors, wiring and busbars and typically have lower losses. In fact, I've been giving some thought on building a ~350V battery bank and using a standard string inverter.

So... what does the county say about you being able to chase people offyour property?

JR