Greenhouse Backup Plan

[2Lime]

I'm planning to build a 48v 18650 setup to power my two sump pumps, four season greenhouse ground to air heat transfer system (fans), and things like my fridge during a power outage and I would also like to add solar panels to expand in the future. To that end I've been collecting and testing IR and capacity of mostly laptop cells for the last 5 years. I'm up to about 3000 ICR cells that have an acceptable IR, are within 80% of their original capacity, and are currently between 2000mAh and 2700mAh. I also have about 200 cells at 1900mAh and 200 at 2800-3000mAh. Finally I have another maybe 800 cells that are low IR (high drain) INR cells. Wolf has suggested not combining these chemistries so I will presumably just ignore the 800 INR cells.

So I have two questions:

What I was initially planning was a 120p14s setup with cells in a 15x8 grid in each of the 120 cell packs since that would fit nicely in the locker I have - like others have done. I have a kweld and plan to use batteryhookup's fused nickle strips, unfortunately it looks like the nickle strips don't come in an 8 wide version. Do you folks think it would be OK to use 3 pieces of nickle strip that are 5x8 to cover each side of the 15x8 pack (or perhaps two 15x4 strips)? Obviously those three pieces would be spanned/connected by the soldered on buss bars on each side but I'm not sure if that is OK. Alternatively I could make twice as many (28) 60p14s (15x4) packs and pair them up for the same effect when I connect the buss bars, but that seems like it might make for more complicated wiring (especially if I have even more packs: see below).

My related puzzle is that I have nearly enough cells to make two 120p14s batteries, but the cells I have range from 2000mAh to 3000mAh. Would I be better served in the long run by making one 120p14s battery with say the 2100mAh to 2400mAh (since I have the most in that range) batteries so that the capacity is closer over all (improved longevity)? Or would it make sense to have two 120p14s batteries (a lot more storage!) but with a wide range of capacity cells; say 2000mAh to 3000mAh? One possible advantage of deciding on the second (2x 120p14s) option now is that it would allow me to really spread out the few hundred higher capacity cells that I have across all of the packs. I suppose I could imagine having two batteries with a different number of cells, so that the total capacity is the same, but that seems complex to keep balanced. Another other option could be two independent batteries with different capacities but it seems nice to have one big one!

 

[Korishan]

What I was initially planning was a 120p14s

Let's get this part out of the way first. Based on the information you've presented, you have this designation backwards. It's 14s120p.
14s120p = 14 packs made up of 120 cells. The 120 cells are connected in parallel which these packs are connected in 14 series
120p14s = 120 parallel groups of 14s. This would equate to 120 different batteries that are 14 cells connected in series.

2100mAh

This means that the cell is capable of producing 2.1A for 1 hour, assuming the cell has an amperage output of 1A (in this example, 1A is about 0.5C) (1C = mAh / 1000)
So if you have 120 cells in parallel, this would yield 120 * 2100 = 252,000mAh, or 252Ah.
If your load of the pump is, guessing here, going to be about 10A, then your battery would last 252 / 10 = 25.2 hours. This is of course assuming 100% efficient power transfer, which we know is not possible. We calculate in about 25% losses, and this becomes are more realistic 18.9 hours.

However, we also have to add in surges. Most motors will surge to 3-5x time their running capacity. So if rated for 10A, we can assume that 30A will be pulled on initial startup.

I suppose I could imagine having two batteries with a different number of cells, so that the total capacity is the same, but that seems complex to keep balanced.

Every parallel group of cells needs to be the same capacity, regardless of the individual capacity of the cells.
A very easy method of building these packs is to separate out every cell according to their capacity within 100mAh blocks (50mAh blocks if you want even tighter control; but is ultimately dependent on how many cells in each group you have available).
Then you take 1 cell from group A, 1 cell from group B, etc, etc, looping back as necessary, until you get 120p. Then you repeat this process for each pack, identically.
This should create a basic simple pack from miscellaneous cell capacities. The other important factor is the "number of cells" in each pack needs to be the same as well. Cells in parallel = Amp charge/discharge capacity. So if one pack has 100 cells, and another is 120 cells, then the 100 cell pack will see higher stress/amps going through them.

Another other option could be two independent batteries with different capacities

This could work as well. Just would need to make sure that both batteries have their own dedicated BMS's, or control modules (for instance Batrium has a single main BMS, but has independent modules for each group of cells, regardless of what battery/batteries they are connected to)

Would I be better served in the long run by making one 120p14s battery with say the 2100mAh to 2400mAh (since I have the most in that range) batteries so that the capacity is closer over all (improved longevity)?

Actually, this would probably be the best option overall. Then all the other cells you could make the 2nd or 3rd battery from. Maybe not 120p, but anything based on what's left. Put your strongest and highest capacity cells together to make the master battery. Then others would be used to make the auxiliary ones.
Longevity is really based on charge/discharge voltage limits (4.0V instead of 4.2V for instance) and how hard you hit the cells in amps during these processes. So if the cells datasheet states that 0.5C is nominal, then anything lower than that will extend the life of the cell a bit.
Obviously the lower the stress is on the cell, the longer it will last.

I'm planning to build a 48v 18650 setup to power my two sump pumps, four season greenhouse ground to air heat transfer system (fans), and things like my fridge during a power outage

Your biggest goal here is to find out how much energy all these devices use over the course of an hour, then you can deduce how much power you need generate/store to run them for a period of time.
Any kind of motor will have surge start up loads. This needs to be included in the calculations. The only somewhat exceptions are if these devices have soft start capability.
Also, are these DC or AC devices? If they are AC and you'll be using an inverter, you need to calculate that in as well. The inverter consumes a certain amount of power while idling. Also has a certain amount of efficiency losses during the conversion process. Generally as a rule of thumb, calculate 25% losses from source to load as this is pretty conservative. There's not many things that are worse than this, and it's always better to over-estimate the losses, than to under-estimate them.

 

[2Lime]

Thanks for the super fast and detailed reply!

Based on the information you've presented, you have this designation backwards. It's 14s120p.

Ha! What a rookie mistake, thanks.

However, we also have to add in surges.

I don't have a clamp style multi meter so unfortunately I don't have a way to properly measure in-rush current. I have been thinking about getting one though if anyone has any suggestions.

I do have a smart outlet that can measure power usage so I dug that out today. I ran my 120v AC refrigerator on it for six hours and it averaged 32 watts per hour. I will try that on my sumps next time it rains.

My two AC greenhouse heating fans (they circulate air underground for heating) are pretty efficient they are 230v at 0.85A and the data sheet says "Shaded pole unit bearing HVAC AC motors have a low starting torque that provides a smooth and quiet startup" so that seems helpful as far as surge goes. I also have been considering upgrading to some more efficient fan motors that I have that are 115v 1.1A (also soft start with a "permanent split capacitor"). The 115v motors would be nice since that would eliminate my only ~220v devices and simplify the inverter I would need (since running a 220v dryer or electric range on battery sounds like a terrible idea). I've delayed this motor upgrade because it will mean that my thermostat controlled outlets for the fans will need to be rewired from 220 to 115. So that's ~400Wh (or ~260Wh if I upgrade) for the two fans.

I could also add a cube heater to the list. If it gets below 20deg F (-7C) outside I have to run a 1500w cube heater. Luckily I only need to run the cube maybe 6 or 12 times a year and very rarely while the power is out. I have a propane heater that did the trick for the one time the power was out and it was that cold.

The two sump pumps are a bit of a trick. They are both 120v AC but only run very intermittently unless it really rains a lot (which of course is when the power is likely to go out) and even then run on a cycle and not constantly. I have the basement sump on a 12v 65Ah lead backup battery that runs another smaller (DC) pump and it was able to cycle many times over a couple of very rainy days without a problem. The greenhouse sump is not vital during a power outage, it just helps keep the bottom 1/4 of the "geothermal" system drained, it normally only runs maybe twice a day for less than 1 minute. All that said I'm sure the surge and running current on the sumps is huge, dang water is heavy. Both sumps are 1/2HP 4.5A 120V, but like you say, I'm sure they have a large initial surge.

I can consider running the small DC pump directly (with a DCDC converter) from this battery but that might take some tricky wiring as they will (of course) be pretty far apart.

So some idealized minimum numbers in summary excluding surge, 32Wh for the fridge, 400Wh for the two fans, skip the scary 1500Wh heater for now and use propane, and some mystery number for the sumps when I can estimate how much they run durring the rain - 540Wh (maybe times two if I include the greenhouse pump) but then reduced by whatever percent of an hour it actually runs. So roughly 1kWh not including surge or inefficencies but running one pump full time.

A very easy method of building these packs is to separate out every cell according to their capacity within 100mAh blocks (50mAh blocks if you want even tighter control; but is ultimately dependent on how many cells in each group you have available).
Then you take 1 cell from group A, 1 cell from group B, etc, etc, looping back as necessary, until you get 120p. Then you repeat this process for each pack, identically.

I like this method and I do have them separated into 100mAh groups.

The other important factor is the "number of cells" in each pack needs to be the same as well. Cells in parallel = Amp charge/discharge capacity. So if one pack has 100 cells, and another is 120 cells, then the 100 cell pack will see higher stress/amps going through them.

Good to know, I will avoid that then.

Actually, this would probably be the best option overall. Then all the other cells you could make the 2nd or 3rd battery from. Maybe not 120p, but anything based on what's left. Put your strongest and highest capacity cells together to make the master battery. Then others would be used to make the auxiliary ones.

Ok, I will focus on one 14s120P battery for now, then perhaps make other different capacity independent ones later.

So when you say to put the highest capacity cells together into the master battery is this still true if I have dramatically less of the higher capacity cells and will end up with a wide range of mAh? So say I have only 30 3000mAh cells should I include 28 of those putting two in each of the 14 packs and skipping the last two because that would be uneven or skip the whole 30 and move down to 2900mAH cells to first include if I have say 60 (or keep moving down until some critical mass). In other words would packs ranging from (doing some quick stack counts) 2300mAh to 3000mAh capacity cells (higher total capacity, but wider capacity range) be better than packs ranging from 2100mAh to 2400mAh (lower total capacity, but narrower capacity range)? Not sure if it will clarify this or not, but I only have about 300 cells that are 2700mAh or larger.

Longevity is really based on charge/discharge voltage limits (4.0V instead of 4.2V for instance) and how hard you hit the cells in amps during these processes. So if the cells datasheet states that 0.5C is nominal, then anything lower than that will extend the life of the cell a bit.
Obviously the lower the stress is on the cell, the longer it will last.

Absolutely.

Also, are these DC or AC devices? If they are AC and you'll be using an inverter, you need to calculate that in as well. The inverter consumes a certain amount of power while idling. Also has a certain amount of efficiency losses during the conversion process. Generally as a rule of thumb, calculate 25% losses from source to load as this is pretty conservative. There's not many things that are worse than this, and it's always better to over-estimate the losses, than to under-estimate them.

I do also have questions about inverters and research that I need to do. I was pretty lucky and picked up some big ol' server UPS's for free and they have been weighing down my shelves ever since. They are all pure sine wave, and I have 7 of them. They are all 48v (batteries) to 120v out, one is ~1450W, two are ~2000W, and four are ~2700W. I also realize that these are almost certainly not the continuous ratings as UPS's aren't intended for constant use. However, I'm hopeful that I can perhaps use some of them simultaniously? Three of them are DLA3000RM2U's which I see using the original batteries can run for about 1hr 20min at 270W (probably not that helpful but interesting). The efficency graph is interesting too. It drops off so much below about 15% load, but it's pretty great above a 50% load - 95% efficent. Hmm, just noticed in it's manual that this thing is 96v for the batteries not 48. Need to take stock of what I actually have here, that's too high!....

 

[Korishan]

I don't have a clamp style multi meter so unfortunately I don't have a way to properly measure in-rush current. I have been thinking about getting one though if anyone has any suggestions.

You don't generally need an in-rush meter. Just take the ratings on the plate and multiply by 3-5 to get a close approximation. Another thing to note is that most devices won't be starting at the exact same time. You can even set up timers/relays to keep this from happening, as well.
Some devices will actually list their Run Amps and Start Amps. I've seen a label with RT and ST before denoting this, and the ST was about 3x time RT.

My two AC greenhouse heating fans.....as far as surge goes.

Yeah, they are probably pretty low. Designed to not make much noise at start up. They probably have built in soft-start. So don't really need to worry about those, then

Changing from the 187W fans to the 132W fans is a good option. Also means don't need to run as much wire, or give you two circuits now per run (that is, if the runs had 3 conductors of 2 hots 1 neutral; ground is usually always included anyways)

(since running a 220v dryer or electric range on battery sounds like a terrible idea)

not as bad of an idea as you'd think, as long as you have the capacity, both amps and runtime, to do it

The two sump pumps....

Ok, seems like this part is pretty easy. Not much to worry about here. You could set up the delay timer to keep two pumps from kicking on at the same time. This would sense inrush current on the line and disable the 2nd pump until the inrush drops, and then click the relay to run mode and allow the second pump to turn on. Inrush could be configured for anything above X-amps for instance.

Both sumps are 1/2HP 4.5A 120V

Yeah, that's pretty beefy. So if that's the run conditions, or 540W, then you could assume that the startup current would be 3x, or 13.5A, up to 5x, 22.5A. This is acceptable range for these on a 120V 15A breaker. Anything higher would actually require a 20A circuit anyways. But most sumps that are just "Plug it in and let it go" are designed for standard house wiring, which is 15A for most circuits.

running current on the sumps is huge, dang water is heavy

Also gotta take into account the height it has to pump it to. If the head-lift is only a foot, no worries. But the higher the lift, the more current will be required, to a degree, and then flow rate starts to suffer more than anything else.
I wouldn't expect to see anymore than maybe going to 5A under heavy head-lift, tbh.

DC pump directly (with a DCDC converter) from this battery but that might take some tricky wiring as they will (of course) be pretty far apart.

You run full system voltage to the pump, and that's where you drop the voltage down to 12V. Otherwise you'll have too much voltage drop and when the pump kicks in, it will try to compensate voltage with amps.
For example: If the pump is rated 12V@10A, that's 120W. If the voltage drops to 10V, the pump will still try to pull 120W, so it will up the amp draw to 12A.
So by keeping system voltage along the main line and then dropping at the load, you reduce the chance of voltage dropping below what the pump needs. And tbh, you could probably set the voltage to 14V and the pump will actually run cooler, and slightly faster, than at 12-13V

I have a propane heater that did the trick for the one time the power was out and it was that cold.

Propane heat is my friend. I have a propane lantern (Coleman style) and a cooker/heater type that I use if it gets too chilly. Can bring the temps up really quickly, replenish the humidity so it's not so dry, and still stay very comfortable.

So when you say to put the highest capacity cells together into the master battery is this still true if I have dramatically less of the higher capacity cells and will end up with a wide range of mAh? So say I have only 30 3000mAh cells should I include 28 of those putting two in each of the 14 packs and skipping the last two because that would be uneven or skip the whole 30 and move down to 2900mAH cells to first include if I have say 60 (or keep moving down until some critical mass). In other words would packs ranging from (doing some quick stack counts) 2300mAh to 3000mAh capacity cells (higher total capacity, but wider capacity range) be better than packs ranging from 2100mAh to 2400mAh (lower total capacity, but narrower capacity range)? Not sure if it will clarify this or not, but I only have about 300 cells that are 2700mAh or larger.

If you only have 30x 3000mAh, then you would put 2x in each parallel pack and you'd have 2 left over. Then you do the next balancing for the next group of capacity cells. If you don't have enough cells to make at least 1 in each pack, do not put those in any of the packs of that battery because you can't (really safely and consistently and balanced) make up for it by adding an extra lower capacity cell (for instance using 2x 1500mAh cells to balance with the 1x 3000mAh cell. You now have an extra cell in the pack with the 1500's and this will lead to imbalancing issues down the road, sooner rather than later)

I would recommend starting with the higher capacity cells and work your way down for your first/primary battery. This will yield the best battery with the most capacity and the highest current capacity as well. Then take the remaining cells and make your secondary/backup battery.
Then you connect these in parallel at the absolute Pos/Neg ends. Each battery would have its own BMS hardware independent of the others.

I was pretty lucky and picked up some big ol' server UPS's for free and they have been weighing down my shelves ever since. They are all pure sine wave, and I have 7 of them. They are all 48v (batteries) to 120v out, one is ~1450W, two are ~2000W, and four are ~2700W. I also realize that these are almost certainly not the continuous ratings as UPS's aren't intended for constant use.

Hahah You'd be surprised. Server grade UPS's are designed to run at high usage (don't recall the load requirements, but >50% capacity) for 24Hrs without issues. They have internal fans to keep things cool, and they have their components arranged in such a way to make sure everything gets plenty of air flow. So running them long term is fine.
The #1 killer of a UPS is lack of airflow, not runtime. I actually ran a 550VA "consumer" UPS (looks like a long beige box, weighs 20lbs max) for over a week almost continuous one year. I had the case pulled off to help with air flow. Ran it off 12V lead acid, which I recharged with my truck when I went into town. Unfortunately though, they are not designed for long run times like that, and I did eventually kill it as the components didn't have heatsinks on the important bits, just the FETs.
I have an SMT3000RM2U that I run my server rack plus my livingroom/bedroom lights/tvs/computer/fans on as well. I'm about 20% of load capacity. I recently upgraded the battery to a 14s1p SPIM08HP's from BatteryHookup. Works pretty good and I have 16Ah instead of 6Ah now.

However, I'm hopeful that I can perhaps use some of them simultaniously? Three of them are DLA3000RM2U's

Oooo, these are nice, actually. With some fiddling, you can actually get 220VAC out from a pair of these. But you have to set them up correctly to do this and may have to do some soldering to trick the "slave" unit to work correctly.
Check out knurlgnar24's videos: https://www.youtube.com/@knurlgnar24 if you're interested in tackling such a task.

just noticed in it's manual that this thing is 96v for the batteries not 48

Ooph, that complicates things a little bit there But still doable! You could make 2x 14s60p batteries and then when ready to connect to the UPS, you throw the switch that connects them together in series (I would not recommend working with a 96V battery, and in some places it's actually against regulations to do so directly). So a heavy breaker switch connecting them together (DC Rated of course, and possibly 2 or 3 in series for extra safety)

 

[2Lime]

You don't generally need an in-rush meter.

Darn, thought I had a good excuse to buy one.

Another thing to note is that most devices won't be starting at the exact same time. You can even set up timers/relays to keep this from happening, as well.

That's a good idea. I will have to ponder how that control will actually work but that's a problem for future me.

Changing from the 187W fans to the 132W fans is a good option. Also means don't need to run as much wire, or give you two circuits now per run (that is, if the runs had 3 conductors of 2 hots 1 neutral; ground is usually always included anyways)

Yeah, I got really lucky with those new fans I found them for super cheap at a ReStore and they are new! Of course it was just a few days after I bought and replaced one of the old ones that burned out after 3 years (it was salvaged from a grocery store and old when I got it) but at least that meant I knew the specs. The new ones are super fancy they can actually run on 115 or 230 so that could work out if I end up with an inverter setup capable of 230. They also are reversible which may be another efficiency improvement for the 'geothermal' setup. I've read that pulling air is more efficient than pushing it and the old fans are going the 'wrong' way and pushing. My 230 wiring is exactly as you describe. Wish I had put an extra neutral in the conduit. And yes it is definitely grounded. In fact the whole greenhouse is on a 50A 220v GFCI breaker to a sub panel out there because GFCI outlets don't work well in the high humidity. But I digress.

Yeah, that's pretty beefy. So if that's the run conditions, or 540W, then you could assume that the startup current would be 3x, or 13.5A, up to 5x, 22.5A. This is acceptable range for these on a 120V 15A breaker. Anything higher would actually require a 20A circuit anyways. But most sumps that are just "Plug it in and let it go" are designed for standard house wiring, which is 15A for most circuits.

It's conveniently been raining quite a lot today so I put the smart plug on the basement 120V sump. I got new larger gutters fairly recently and I think that helped, I'm just sitting around staring at my sump and it's filling incredibly slowly. After an hour and a half it was nearly full so I tripped it manually. It ran for less than 10 seconds at a constant 7.1A and about 740W according to my smart plug. Water is being lifted 9ft (3m) in 1.5in (4cm) PVC and it has check valves so that it doesn't backflow. It is on a 20A breaker. So it may be seeing surges of 21A to 35A - wowee.

You run full system voltage to the pump, and that's where you drop the voltage down to 12V. Otherwise you'll have too much voltage drop and when the pump kicks in, it will try to compensate voltage with amps.
For example: If the pump is rated 12V@10A, that's 120W. If the voltage drops to 10V, the pump will still try to pull 120W, so it will up the amp draw to 12A.
So by keeping system voltage along the main line and then dropping at the load, you reduce the chance of voltage dropping below what the pump needs. And tbh, you could probably set the voltage to 14V and the pump will actually run cooler, and slightly faster, than at 12-13V

OK, looks like the 12V DC pump is rated at 6.5A (some of this stuff I'm just putting here so I don't have to dig it up again later). Makes sense to do the DCDC conversion near the pump if I go this route. The 14V suggestion is interesting, is that because it would be higher V lower A?

If you only have 30x 3000mAh, then you would put 2x in each parallel pack and you'd have 2 left over...

This is really clear and I appreciate you taking the time to explain this in detail! This is great news.

I would recommend starting with the higher capacity cells and work your way down for your first/primary battery. This will yield the best battery with the most capacity and the highest current capacity as well. Then take the remaining cells and make your secondary/backup battery.
Then you connect these in parallel at the absolute Pos/Neg ends. Each battery would have its own BMS hardware independent of the others.

AH. I see.

Server grade UPS's are designed to run at high usage

This is also great news. It's cool that you got such use out of one of those classic beige UPS's sounds like it was a life saver.

Check out knurlgnar24's videos

Thanks for the recommendation, his videos look great! I've dug out my UPS boat anchor collection and it looks like the three I though were the same are actually different. I have:

UPS model	data ports	Watts	Max input Amps	VA	Battery Voltage	Max battery A
DLA3000RM2U	Many	2700W	24A	2880VA	48v	Not listed
SUA3000R2X428	Many	2700W	24A	2880VA	48V	Not listed
SUA3000R2X428	Many	2700W	24A	2880VA	48V	Not listed
PR2200LCDRTXL2U	Many	2150W	17.2A	2150VA	48V	70A
SMT3000RM2U	Serial only	2700W	24A	2880VA	48V	Not listed
GXT2-2000RT120	Serial only	1400W	17.4A	2000VA	48V	34A
R1500XRNA	Serial only	1340W	12A	1440VA	48V	40A

 

[2Lime]

Ah shoot, posted that before I got my table updated. I will fill it out later.

 

[Korishan]

The 14V suggestion is interesting, is that because it would be higher V lower A?

Yup. You generally always want your amps to be the lowest possible as that's what heats the wires up.

This is also great news. It's cool that you got such use out of one of those classic beige UPS's sounds like it was a life saver.

Yeah, we were out of power for almost 2 weeks. It's when hurricane Irma came through basically right up the center of Florida in 2017.

Ah shoot, posted that before I got my table updated. I will fill it out later.

You can edit your previous post to add them

 

[2Lime]

UPS model ... Table

I had some difficulty locating the battery voltage of the big ups's, especially the DLA3000RM2U. Unfortunately all of the original batteries are long recycled since they were mad swollen, but I remember past me only keeping UPS's that had 48V packs. As far as I can tell from more research the DLA3000RM2U uses 8 12V cells but half of them must be in parallel to increase the amps from 5 to 10 resulting in a 48V pack (like this one). This ebay ad agrees so looks like it really is 48. It's funny how hard APC makes this information to find when a model is discontinued.

I really have a lot of inverters to pick from. I think I will probably send on the bottom two since they are pretty old and relatively small wattage (and have proportionally dinky power cords)

This reminds me, my friends at the electronics recycling company do get 220V UPS's in fairly often, I had been just ignoring those since most of my stuff is 120V, but maybe I should reconsider if they have respectable wattage since I could presumably use one hot leg to get back to 120V just like from my main breaker box. I will have to see what voltage batteries they are using too.

I have an SMT3000RM2U that I run my server rack plus my livingroom/bedroom lights/tvs/computer/fans on as well. I'm about 20% of load capacity. I recently upgraded the battery to a 14s1p SPIM08HP's from BatteryHookup. Works pretty good and I have 16Ah instead of 6Ah now.

Crazy that one of the UPS's I have is the same as you! Those SPIM08HP's seem pretty reasonably priced compared to an 18650 and that's an interesting upgrade. Is the goal of that setup to avoid brownouts and short power outages? Seems like it couldn't run super long with that many devices. I love the relative simplicity of the 14s1p though.

Yeah, we were out of power for almost 2 weeks. It's when hurricane Irma came through basically right up the center of Florida in 2017.

Man, that's a long time! I think the longest I've every been out for was 5 days from a Derecho and it was mostly just waiting for them to fix all the other lines to get to our small branch outages.

 

[Korishan]

but maybe I should reconsider if they have respectable wattage since I could presumably use one hot leg to get back to 120V just like from my main breaker box.

Nah, this isn't really a good idea, tbh. It'll make the inverter become imbalanced on the leds and will most likely trigger a safety feature. The other thing is, you'd have no neutral anyways. The outputs are 2 hots and ground normally. Unless they have a NEMA Twist lock that has 4 lugs on it, you won't have access to a neutral leg.
You could use a transformer that has a center tap. But you definitely need to make sure you keep the loads as balanced as possible, or you'll work the inverter of the UPS harder.

I would just run loads that require straight up 220V, like your dryer, water heater, oven, fans, etc. Keep it on a separate grid, as it were.

Is the goal of that setup to avoid brownouts and short power outages? Seems like it couldn't run super long with that many devices.

Pretty much, yeah. Currently the UPS says I have about 45mins of runtime with my current load out. This would at least allow me enough time to be able to get the generator online and get things switched over before power goes completely out.
Eventually I'm going to have solar/wind to help keep the battery(ies) charged and the UPS would at that point run indefinitely. Solar would be for our clear days, and wind would be for the storm days. We get a lot of winds during storms.
Getting a little OT here, but I'm also planning on doing some low grade hydro to help out as well. We have a ditch that runs down the middle of the property that I've cleaned out and widened last year. It can flow a lot of water during heavy rains, and that water would last quite some time as the head of the ditch is at the base of a hill at the road front. But that's a long time project.

just waiting for them to fix all the other lines to get to our small branch outages.

Yup, that's us too. We're a leg off a leg off a leg, as it were. There's only 3 homes on this leg. And that storm knocked out quite a few mains lines that needed to be cleared and repaired before they could re-energize it. We're sort of in the middle of 3 junctions, so power can be fed from 3 different directions. But with that kind of damage, that don't mess with "Looks good to me!" kind of mentality. I waited patiently for the repairs, and took very quick cold showers

 

[DG98]

I have an SMT3000RM2U that I run my server rack plus my livingroom/bedroom lights/tvs/computer/fans on as well. I'm about 20% of load capacity. I recently upgraded the battery to a 14s1p SPIM08HP's from BatteryHookup. Works pretty good and I have 16Ah instead of 6Ah now.

@Korishan, do you have a thread about reconfiguring your UPS to use the SPIM08 cells? I have an older APC rackmount / server UPS as well, which serves all the Cisco networking equipment in my home. I've replaced the lead-acid batteries in it once, and it's been awhile, so I'm just waiting for it to need another replacement. I don't want to derail this thread but would be interested in seeing how you set it up. Did you just leave all the new cells outside the case with a BMS, and then run a connector to the APC wire for the additional shelf battery?

Thanks --
Cheers, John

 

[2Lime]

...do you have a thread about reconfiguring your UPS to use the SPIM08 cells?

I would be interested in this as well.

Nah, this isn't really a good idea, tbh.

Good to know! I will keep an eye out for a 220 UPS for the separate grid option, that is an interesting option.

Unless they have a NEMA Twist lock that has 4 lugs on it, you won't have access to a neutral leg.

This is helpful too, that's an easy quick thing to look for. I was wondering how the wiring would work otherwise.

solar/wind ... hydro

So many options! I wish I had a water option like that that's excellent.

...unfortunately it looks like the nickle strips don't come in an 8 wide version.

I was thinking about this initial question I asked and I figured it out. The way the bus bars are laid out on a 8x15 (120p) pack like this make it so that two parallel 4x15 nickle strips would be connected to the batteries the same as how folks are doing the glass fuses. See this photo from KGE's thread.

 

[Korishan]

do you have a thread about reconfiguring your UPS to use the SPIM08 cells?

Not yet.

If you were going to do this design, you wouldn't be able to incorporate the fuse from the buss-bar to the cells. All the cells would still be connected in parallel.
Fusing at the cell level is not for protection of overcurrent of loads. It is protection of a cell(s) going short circuit internally creating a basic dead short and pulling the energy from the adjacent cells into self, which this causes that cell(s) to heat up and go thermal run away if there's enough energy available.
The best way to incorporate this design with nickel sheets is to use the kind that have fuses built into the tabbing, like these:

Image from AveRageJoe's video on the topic:

View: https://youtu.be/BAPHF3Sq2t8


This still keeps the cells isolated from each other, but you can still put a buss-bar on it and get balanced current flow. Which, btw, I would put a buss between groupings like the first image, so there's still a single cell on either side of each buss that runs the length. This will minimize any voltage/current issues of imbalancing during heavy charge/discharge

 

[2Lime]

This still keeps the cells isolated from each other, but you can still put a buss-bar on it and get balanced current flow. Which, btw, I would put a buss between groupings like the first image, so there's still a single cell on either side of each buss that runs the length. This will minimize any voltage/current issues of imbalancing during heavy charge/discharge

Cool. Yes, this is exactly what I meant, I definitely want to use the fused nickle strips with the K-weld and avoid all the soldering of glass fuses. My only hold up was that battery hookup doesn't carry 8 cell wide fused strips, so I was trying to figure out if using 2 4x15 fused strips on the positive side and 2 4x15 fused strips on the negative side would be OK and it clearly is with this bus layout. Even four, 2x15 fused strips on each side of a pack would be effectively the same. It sure is difficult to explain things in text sometimes.

 

[Korishan]

o I was trying to figure out if using 2 4x15 fused strips on the positive side and 2 4x15 fused strips on the negative side would be OK and it clearly is with this bus layout. Even four, 2x15 fused strips on each side of a pack would be effectively the same. It sure is difficult to explain things in text sometimes.

Yeah, would be fine. To combine them together you'd spot weld the two "halves" with regular nickel strips so that it becomes a single unit.
You'd use a strip long enough to pass by several cells on both sides, and probably want to put a strip every 3rd cell line, I would think. Every 4th might be fine, tbh, as well. But if you have a buss-bar on both halves that combines, that's going to help out a lot too. But definitely want to bond both halves together at the sheet level.

 

[2Lime]

You'd use a strip long enough to pass by several cells on both sides, and probably want to put a strip every 3rd cell line... But definitely want to bond both halves together at the sheet level.

Great, this sounds straight forward enough. I will go ahead and order some 4 cell wide strips and get things laid out. Appreciate the help!

Another question. When building a 14S120P battery I've seen some folks build a 15th pack at the same time to serve as a backup unit. Is this generally advised? I'm wondering how the spare pack is maintained - is it just kept at a storage charge or is it cycled out with the other packs in some rotation so that it experiences the same level of wear?

 

[Korishan]

seen some folks build a 15th pack at the same time to serve as a backup unit. Is this generally advised? I'm wondering how the spare pack is maintained - is it just kept at a storage charge

yes, yes, yes

A hot spare basically, in the advent you need to bring a pack out of service for cell replacement, then you can take the spare and pop it in and let your system come back online while you work on the other one. At this point depending on how long it's been since construction and putting into service, you may want to keep it as a spare as it won't age at the same rate as the other packs.

Another option some have done is built what is referred to as X+1 systems. You build the number of strings required for your loads and runtime. Then, you build 1 more string to serve as a backup in case you need to bring a string out of service. This way you don't have "packs" being mismatched based on age. You just let the replacement string stay in service, as all the packs in the string will age at the same rate at that point.

cycled out with the other packs in some rotation so that it experiences the same level of wear?

You could do this, but overall probably wouldn't happen more than a few times in application. We forget. Something comes up and "I'll do next weekend....next month.....in a few days...." and then before you know it, it's almost a year later and never got rotated. I personally don't think rotating in/out the packs would be worth it in the long run. Just make another string and use it as a full unit instead of peace-meal

 

[2Lime]

In the sorting process I came across some 18650's that I wanted to ask about. I don't have very many of them so I can just leave them out if needed. They are LG and are labeled INR18650B4 and +ASO93DO38A1- (marking the + and - end, that's nice). Dark gray with a hint of green, and a white ring. So since they are INR I would expect them to be high drain with a low IR, however they consistently test right around 51mOhm and 2730mAh without getting hot. I don't see them in the main cell database (although I do see some pretty similar one's in Wolf's cell database). I found a datasheet here that says that they were renamed from ICR to INR which is pretty funny to me, it also lists them as less than 70mOhms on the IR test. If I'm reading this correctly then these should be fine to use with my other 'low' drain cells, it's more about not having too wide of an IR range.

 

[OffGridInTheCity]

In the sorting process I came across some 18650's that I wanted to ask about. I don't have very many of them so I can just leave them out if needed. They are LG and are labeled INR18650B4 and +ASO93DO38A1- (marking the + and - end, that's nice). Dark gray with a hint of green, and a white ring. So since they are INR I would expect them to be high drain with a low IR, however they consistently test right around 51mOhm and 2730mAh without getting hot. I don't see them in the main cell database (although I do see some pretty similar one's in Wolf's cell database). I found a datasheet here that says that they were renamed from ICR to INR which is pretty funny to me, it also lists them as less than 70mOhms on the IR test. If I'm reading this correctly then these should be fine to use with my other 'low' drain cells, it's more about not having too wide of an IR range.

I've had good luck with mixed IR / mixed cells in 1 battery. The key (in my experience) is to evenly distribute the mix ratio thru each of the 14 individual packs. Last year I did a 48v @ 260ah battery #9 (14s107p) for my Powerwall of my 'left-overs' from several years. Here's a visual of the cells.....

Here's the exact makeup of each 107p pack. My packs are targeted to be 260ah (+ a bit).

*The Sony G5 green cells (10 cells out of 107 per pack) have an IR of 200+! The rest are normal range.
** Also notice the worst cell tested 83% of original capacity and most are 90%+ - hopefully they won't degrade quickly.

Battery #9 is working smoothly along with the other 8 batteries in my powerwall - no balance for several months.
Note: I typically run 40-80mv max difference thru all the 126 packs so that's not a function of having a mixed cell battery.

In general, I'm in favor of making batteries (14sXXp) of all the same cell type and good IR for better performance, but this Frankenstein shows that it's possible to do a mix and still have a perfectly useful battery.

 

[2Lime]

The Sony G5 green cells (10 cells out of 107 per pack) have an IR of 200+

Wowsers that's an enormous IR, crazy to me that those still work!

Pretty cool that you were able to build a battery like that with such a mix of cells! I appreciate all of the visuals. It took me a while to realize just how much storage you have! Your Batrium bar graph is getting skinny bars .

Note: I typically run 40-80mv max difference thru all the 126 packs

So is the idea here that a tiny bit of voltage difference doesn't matter much and keeps them from having to charge / discharge quite as much? Or is this just a standard value.

I was thinking more about my INR18650B4 cells and I only have like 12 cells (not even enough for one per pack) so I think I will err on the side of caution and just not use them. Maybe I will use them in my 9th battery build...

 

[OffGridInTheCity]

Wowsers that's an enormous IR, crazy to me that those still work!

Pretty cool that you were able to build a battery like that with such a mix of cells! I appreciate all of the visuals. It took me a while to realize just how much storage you have! Your Batrium bar graph is getting skinny bars .

It's 121kwh

So is the idea here that a tiny bit of voltage difference doesn't matter much and keeps them from having to charge / discharge quite as much? Or is this just a standard value.

I can balance them at the top (4.15v) or the bottom (3.4v) or in the middle 3.7-3.8v but over the operating voltage range they diverge as much as 40mv no matter what I do. I believe it's because I have different packs - e.g. the cells are different manufacturers and have slightly different charge/discharge curves - but I don't actually know.

@Wolf has shown in his own powerwall that the same cell, hi original capacity, and good IR can reduce this variability. I don't even measure IR (as an example) except loosely with the OPUS BT-C3100. Maybe he'll chime in here

Background / more detail....
Overall, I operate between 3.5v and 4.0v but in spring and summer it's 4.0v to 3.7v type of thing and in winter I'm at 3.5v -> 3.7v a lot of the time and in general all over the range as I don't use the available capacity in the 3.5v to 4.0v range. Overall I'm running a 36.7% DOD. So, I don't even try to balance on a regular basis and it turns out that healthy packs don't need it as they dift slowly and as long as I'm limited to 4.0v I have 200mv to play with before risking overcharge.

I'll tighten the balance every 6-9 months (or adding a battery) to 30-40mv max difference using Batrium Auto-Level. Then, over the months, with no active balance, the max difference will drift. I started tracking the min/max mv difference between highest and lowest pack per month 2 years ago and it looks like this: