diy solar

diy solar

3P8S 840 AH build

Bob613

New Member
Joined
Nov 24, 2020
Messages
17
Location
eastern Ontario
The battery is for off-grid storage, replacing a 790AH 24V Crown industrial battery that is actually working quite well but the maintenance is tiresome.

The maximum charge rate is 90 amps from a Midnite Classic controller fed by 3600 watts of panels in two fixed arrays. The maximum discharge rate will be 135 amps (continuous), 250 amps peak during pump starting, through a Magnum MS4024 inverter.

Ordered 24 REPT 280 A cells from Shenzen Continent Technology Co. Leadup to the buy went well, with all my questions answered, shipment arrived nicely packed with no damage 60 days after order. There was a 10 day or so delay because of a backup at the Vancouver port.

As delivered, the cell voltages were uniform, within 20 mv, slightly below the values they sent me before shipment. The internal resistances are all within 2 milliohms of each other. I won’t share the actual values because I measured them with a YR1035+, which according to the manual isn’t accurate at low resistance values, but I trust it as far as cell to cell consistency.

I opted for welded on female terminals. They are 6mm, like the drilled terminals, but are nice and consistent. I went this way instead of the stud type because I want to control the metallurgy, using aluminum bolts and nuts to make sure there is no galvanic action. I want this to last a long time.terminal.jpg

The cells will be connected using 3/16 aluminum bus bars. I wanted 6101 but it is really hard to find so used 6061 and went from 1/8” to 3/16” to account for the higher resistivity of 6061. The bus bars are 7” wide to cover the terminals on three cells; the total bus calculated voltage drop is 10 millivolts at 125 amps.

The terminal is 12.5mm in diameter, with a 7mm hole, so the contact area looks small but is 75 mm^2, between 2/0 and 3/0 cable, more than enough for the current that any individual cell will see. I found a couple of papers discussing the resistance of busbar connections, and they suggested while the area needs to be large enough to carry the current, the joint resistance is greatly affected by the contact pressure.. The target contact pressure is in the 15 to 30 N/mm^2 (2000 to 4000 psi) range. For my terminals, this means a force of 1125 to 2250 N (250 to 500 lbs).

To secure the terminals I am using a series stack of 3 belleville spring washers with a spring rate of 4000 N/mm (900 lb/mm). By compressing the spring stack by 0.50 mm (3 flats on the nut) the joint is loaded to 2000N. I am screwing a nut onto the bolt, hand tightening the bolt into the terminal, then holding the bolt as the nut is tightened 3 flats after it is snug. Experimentation and calculation shows that this load is within the yield strength of the 7075T0 bolts.

I annealed the 6061T6 bus bars to soften the contact area, to give the bus bar and the terminal the best opportunity to become intimate. BTW, an oven self cleaning cycle does a pretty good job annealing aluminum – not perfect but pretty close.
busbar.png bolt.png


The batteries are installed without compression. I considered the various information about compression and decided, after some communication with REPT, and some other research, that given my low C rates, there would be minimal value. On the other hand, by leaving space all around each cell prevents any stress on the battery terminals, which I judged to be a more likely source of unreliability.

A number of sources suggest that a cycle is defined as a cumulative use equal to the batteries' capacity; that is, using 25% one day, 50% the next, and 25% the next equals one cycle. Using that information, if a typical 280 AH battery is good for 3000 cycles, then its life can be calculated as an accumulated 280 * 3000 = 840,000 AH. In my 3P8S battery, that equates to 70,560 kilowatt-hours. Since we use about 6 kw-hr/day, the expected life is 32 years. Affecting that one way or the other by compression is, for me, unnecessary.

Time will tell.

The control system is an Electrodacus SBMSO BMS.

An output turns the Midnite controller on/off through its AUX 2 port (I am abandoning the WB Jr and the ability to absorb to end amps) when the battery reaches 28.4 volts or any cell reaches 3.55. The Midnite will be set to 29 volts, which it will never reach. Instead, it is always in either bulk mode or resting, controlled by the BMS.

Another output turns off the inverter if the voltage is low; this output resets and the inverter will restart if the voltage recovers. The inverter interface is a magnum ME-RSA adapter.

A third output provides an alarm signal if the SOC gets low (likely will set it at about 35%) to alert to turn on the standby generator.

A fourth is a fault output in case the battery reaches dangerously high or low voltages. This output will trip all the breakers (inverter, solar in, charge controller out, generator). I don’t expect this to ever be used but the system is unattended for part of the year, and I decided this extra safety is worthwhile. It took a while to find shunt trip breakers I was happy with, and in the case of the inverter breaker it involved a bit of invention – I designed a simple add-on to trip the 250 amp breaker.

The generator is not really a standby generator but rather a 24 volt 60 amp propane fueled battery charger. Setting the run timer to 5 hours will take the battery from about 30% to 70 % or so. We don’t expect to run the generator more than a few days a year, if at all, and so decided that this approach would be OK. If it turns out we need to run the generator more, I can easily connect it onto the same charging circuit that controls the Midnite solar controller.

I haven’t done any battery capacity testing; the vendor sent me some data and in any case the battery is large enough that is the cells are a bit off in capacity it doesn’t matter.

The battery cabinet is nearly complete, so I expect that within a week or so I should be ready to swap in the new battery pack. The cell voltages are all close so I am not going to separately top balance. Instead, I’ll keep an eye on the cell voltages on the BMS and if anything looks out of whack I’ll deal with it then. I suspect that in a fairly short while the BMS will take care of any imbalances.

While transporting the batteries, I couldn't help but notice that 24KW fit easily into the frunk; with a little finesse 50KW would fit easily:
20210618_155819.jpg

But I have been warned that is not a priority project :)
 
The battery cabinet is nearly complete, so I expect that within a week or so I should be ready to swap in the new battery pack. The (1) cell voltages are all close so I am not going to separately top balance. Instead, I’ll keep an eye on the cell voltages on the BMS and if anything looks out of whack I’ll deal with it then. I suspect that in a fairly short while the (2) BMS will take care of any imbalances.

Welcome to the forum. Nice work.

The YR1035+ is fine for comparative purposes, and it's as accurate one can reasonably expect from the process.

(1) cell voltages in the as-receive condition are completely meaningless. Upon receipt, my 280Ah Eve cells were within 20mV, but their actual SoC varied by 12%. Low was 28%, high was 40%. That much imbalance would wreak havoc on charge. I had a bunch of cells that had been sitting for a year at 3.300V - they were all >95% SoC.

The flatness of the voltage curve means a huge swath of the SoH range have negligible voltage differences, particularly when extended sit times are involved.

(2) Depends on how you define short. That much imbalance would take 504 hours of continuous balancing above 3.4V to bring them in line. You would need to continually charge them at about 0.2A to make it happen. If you can let them sit there for 500+ hours before you need them, then maybe that will work.

You can find DOZENS of examples of batteries that weren't top balanced causing frustration upon deployment using the same assumptions you are using. They usually end up spending hours fiddling with resistors to try and bleed them down faster, or just give up and top balance them.

I can't say the number is zero, but I don't recall any threads saying how well their not-top-balanced commodity cells are performing. Retail matched cells, sure.

Here's my test data:

 
Bob, if you are not compressing your cells, how are you securing them? Or are you not securing them, given this is a stationary application.

In the past, I have annealed aluminium with a torch, using the colour of a coating of ordinary bar soap as a vague temperature indicator. Do you have any thoughts on that method for this application?
 
Its a stationary application, and, by my analysis, above, the complications of properly constraining the cells isn't attractive.
The cells are not secured, and have a space between them so any "breathing" will not move the cells around. As I said above, time will tell.

Aluminum anneals around 700F; the oven self cleaning cycle hits about 800F. 1/4" pieces left in for the whole cycle including the cooldown, came out much softer than when they started. I decided to not do the torch method because the parts are fairly large - 3" x 7.5" - and I didn't think I could heat them evenly.
 
Nice write up, thanks. Have you done any measurement of internal resistance (and busbar connection resistance)? I would be curious to hear how much your voltage changes under load given all the work you did there.

great choice on the terminals BTW - I got welded studs without really thinking about it and it was a mistake in hindsight.
 
Have you done any measurement of internal resistance (and busbar connection resistance)?
Based on the IR of the individual cells, the calculated IR of the battery is 0.671 milliohms.
By measurement, using the same YR1035 as I used to measure the cells, the Neg to Pos IR is 1.15, a difference from calculated of 0.479, presumably due to the connections.
I measure fairly consistent cell terminal to the top of the busbar resistance of 0.06 milliohms; there are 16 connections in each string of 8, so that would total 0.96 milliohms. There is a discrepancy with respect to the 0.479; that is likely just measurement error.
These measurements are outside of the YR1035 accuracy range, although I believe that the numbers will be consistent measurement to measurement. The meter has an offset of about 0.02 at zero, so if the 0.06 measured was instead 0.04, then the measurements align fairly closely.

So, it looks to me that the bus to battery connections are pretty good. It wasn't mentioned above, but I cleaned the bus bars using a Dremel fine flap wheel soaked in NoAlOx, then wiped it off, and recoated with NoAlOx. I cleaned the battery terminals using a scotchbrite pad and NoAlOx, and left enough NoAlOx in place that the sqeezeout formed a little sealing ring around the terminal. I think that is good to help keep air out of the joint.

I'd be interested in hearing if anyone else has measured connection resistance, for comparison.
 
I'd be interested in hearing if anyone else has measured connection resistance, for comparison.
I’d be reasonably skeptical of those numbers without doing a test with a real load of 100A or more and measuring the dV between 0A and 100A with a multimeter.

The connections do look good though, I’m about to a second pass on mine with a little more NoAlOx on one side of the connection that I don’t think had enough earlier and the addition of Belleville washers.
 
Any guess what torque is on the nut to get the 1/2 turn? Mine have the welded nuts also.
 
Hey Bob, i'm looking at doing something similar to you in the next few months. Is there any particular reason you don't want the midnite classic to control the charging via the wizbang? Are your cells going to have a chance to balance with charge turned off as soon as the first cell hits 3.55v? If your battery becomes full earlier in the day the inverter will be pulling power from the battery rather than the midnite.

Can you eleborate some more on how you are setting up the 4th output? This is how i want my battery to disconnect with a high or low cell voltage which may never actually happen. I trust the midnite should prevent overcharging and the generator under.

Thanks.
 
Hi Bob, I have been checking out Shenzhen Continent Technology Co on the exact same batteries.
May I ask what the price for each cell is? Did you have a chance to perform capacity test yet?
Thanks.
 
Hi Bob, I have been checking out Shenzhen Continent Technology Co on the exact same batteries.
May I ask what the price for each cell is? Did you have a chance to perform capacity test yet?
Thanks.
Probably best to get a quote from them - prices from a few months ago might not be current. I paid right around $100/cell, delivered to Canada.
I dealt with Elodie Liu, who answered everything I asked, provided me with the REPTA spec for the cells, and accepted the changes I requested in the pro forma invoice (specifying intact QR codes, IR maximum and range, send me data for approval before shipping).
They also sent a capacity test from the batch of cells.
Of course, all of the data could be made up, but my sense from the communications is that it was all OK.
The cells were very well packed, and they provided good tracking info, and updates when I asked.
On arrival, the cells looked nice, and my IR tests duplicated theirs, keeping in mind that a YR1035 isn't particularly accurate at these low IR's. I was looking for cell to cell consistency and that was OK.
I have not done a capacity test and am not going to bother. If the capacity is off it doesn't really matter to me because the battery is somewhat bigger than I need.
Hope that helps
 
Any guess what torque is on the nut to get the 1/2 turn? Mine have the welded nuts also.
With the belleville spring washers, torque doesn't matter. Based on the spring rate and the force I wanted at the connection, I needed to compress the spring 0.5mm; 1/2 turn on a M6-1.0 bolt. The spring compression method is much more accurate than torque.
 
Hey Bob, i'm looking at doing something similar to you in the next few months. Is there any particular reason you don't want the midnite classic to control the charging via the wizbang? Are your cells going to have a chance to balance with charge turned off as soon as the first cell hits 3.55v? If your battery becomes full earlier in the day the inverter will be pulling power from the battery rather than the midnite.

Can you eleborate some more on how you are setting up the 4th output? This is how i want my battery to disconnect with a high or low cell voltage which may never actually happen. I trust the midnite should prevent overcharging and the generator under.

Thanks.
The BMS I chose has two levels of control, one for controlling the charging and discharging within the normal battery parameters, and another for an emergency shutdown in case things (voltage, temperature) fall into a battery damaging area.
The BMS balances anytime the voltage is below 3.55 and the cell to cell delta is more than 10mv and there is more than 300mA of charging current. So essentially anytime there is charging. Because of this, I don't need to rely on an absorb period to accomplish balancing, and rather than complicate things I decided to "dumb-down" the Midnite charge controller and let the BMS have control.
Btw, all of the parameters are programmable in the BMS.

By the 4th output, I assume you mean the fault or emergency shutdown. This output is on if all is well, and opens on fault (the parameters are programmable but essentially are a hi-hi and lo-lo voltage). The output holds on a small relay; when the relay is de-energized, it sends 24 volts to the shunt trip on the breakers for the PV in, controller out, generator in, and inverter. Tripping these breakers effects a complete electrical shutdown.
I am using these breakers and shunt trip accessory:

1625750733097.png1625750792415.png
The shunt trip has auxiliary contacts which I use to cut the trip circuit once the shunt has tripped the breaker.
I don't expect that this circuit should ever trip unless something goes haywire but decided the added complication and expense is worht having for peace of mind. The system is unattended for part of the year.
Hope that helps.
 
The BMS I chose has two levels of control, one for controlling the charging and discharging within the normal battery parameters, and another for an emergency shutdown in case things (voltage, temperature) fall into a battery damaging area.
The BMS balances anytime the voltage is below 3.55 and the cell to cell delta is more than 10mv and there is more than 300mA of charging current. So essentially anytime there is charging. Because of this, I don't need to rely on an absorb period to accomplish balancing, and rather than complicate things I decided to "dumb-down" the Midnite charge controller and let the BMS have control.
Btw, all of the parameters are programmable in the BMS.

By the 4th output, I assume you mean the fault or emergency shutdown. This output is on if all is well, and opens on fault (the parameters are programmable but essentially are a hi-hi and lo-lo voltage). The output holds on a small relay; when the relay is de-energized, it sends 24 volts to the shunt trip on the breakers for the PV in, controller out, generator in, and inverter. Tripping these breakers effects a complete electrical shutdown.
I am using these breakers and shunt trip accessory:

View attachment 55424View attachment 55425
The shunt trip has auxiliary contacts which I use to cut the trip circuit once the shunt has tripped the breaker.
I don't expect that this circuit should ever trip unless something goes haywire but decided the added complication and expense is worht having for peace of mind. The system is unattended for part of the year.
Hope that helps.
Is there an issue with suddenly disconecting the Midnite from the battery?
 
The breakers trip in a sequence, first disconnecting the panels, although the time gap is a couple milliseconds.
In any case, according to a discussion on the Midnite forum, it isn't an issue.
I have tested the shutdown a few times without any ill effect.
Fwiw
 
I’d be reasonably skeptical of those numbers without doing a test with a real load of 100A or more and measuring the dV between 0A and 100A with a multimeter.
Battery was on charge today at 80 amps. I checked the delta V between battery terminals, across the bus and diagonally.
It is really hard to repeatably measure such low voltages, but by using both a YR1035 and a multimeter, I came up with a set of measurements in the range of 0.2 and 0.9 millivolts. While that is a pretty large spread, the actual values are low enough to be inconsequential, I think. The YR1035 accuracy is about 0.2mv. My multimeter is more like 1mv, so I looked at those measurements for repeatability but discarded the values.
Its probably useful to have these values as a basis for future checking, to see if there are any changes.

As far as the resistance, .5mv at 80/3=27.5 A gives a resistance of .02 milliohms, so in the same ballpark as measured.
I'm pretty satisfied with the connections and the busbars.
 
Just a followup now that the battery has been operating for a couple of weeks.
THe following is the charging cycle, it is consistent from cycle to cycle.
The cell to cell delta V is under 10 millivolts until about 5 minutes before the end when one cell hits 3.55 volts.
During that 5 minutes or so, the cell to cell delta V starts to increased, and cell 8 consistently hits 3.55 first, with the others achieving 3.45.
Then, once the charging stops, voltages start dropping, reaching 3.35 volts after about 40 minutes, at which time the delta V is back to under 10 mv, where it stays until the next charging cycle.
The BMS does passive balancing during the charging cycle, but when discharging.

I did not do any pre assembly balancing - just assembled the battery and let the BMS control the charging.
It seems to me that the battery is working pretty well, given that the cell voltages are very close except for the short period at the end of the charging cycle.
Capture.JPG
 
i like your setup. i have a 4P8S 24-volt battery 32 cells (really 25.2 volts nominal) that has been functioning for about 2months. i also have 2 other 2P8S 24-volt battery builds (16 cells each) that have been functioning for about 11 months. each battery has its own Electrodacus SBMS0 and none were top balanced. all cells were bought new from 2 different vendors on Alibaba. i used Lishen 272Ah cells for those 3 battery builds. i ordered another 32 280Ah Eve cells for an upcoming build another 4P8S 24-volt battery build. all are being charged with solar panels each day. i do not have the inverter cut-off set up yet with the SBMS0 but monitor my usage.
I use the Electrodacus DSSR20's for the solar charge controller. the SBMS0 shuts them off when any cell group of the battery reaches 3.55volts.
Keep up the good work,
I am in the process of building solar panel array ground mount structures for more solar panels.
cheers from South Dakota - winter storm today ? very windy with snow!
 
Well, my battery has been in operation for a few months now and the only remarkable thing is that there have been no hassles at all.
I have been monitoring it pretty closely, looking for issues, using the SBMSO on-board data log as well as some additional logging I can monitor remotely. The cell to cell delta V stays under 10 millivolts other than a few minutes as the charge is completed. While the balancing is limited to 150 milliamps, it operates anytime the cells are above 3.2 volts and are charging. As a result, there is a lot of balancing time and that seems to keep things in order.
As mentioned above, I have the Midnite Classic 150 charge controller set to transition to Absorb at 29 volts. The SBMSO cuts off charging when a cell hits 3.55, so the charge controller never leaves the bulk phase. I have calculated capacity a couple of times based on data and am within a couple percent of the cell's specified 280AH.
So all in all, I am very satisfied.
Well, except for deciding to add another 2500 watts of solar panels on the roof of my garage. Got a deal on some 330 watt panels and will connect them to the system using DSSR20's. There might be some interaction with the charge controller but I don't think it will be of any consequence, and I don't want to rewire the existing arrays.
 
The BMS I chose has two levels of control, one for controlling the charging and discharging within the normal battery parameters, and another for an emergency shutdown in case things (voltage, temperature) fall into a battery damaging area.
The BMS balances anytime the voltage is below 3.55 and the cell to cell delta is more than 10mv and there is more than 300mA of charging current. So essentially anytime there is charging. Because of this, I don't need to rely on an absorb period to accomplish balancing, and rather than complicate things I decided to "dumb-down" the Midnite charge controller and let the BMS have control.
Btw, all of the parameters are programmable in the BMS.

By the 4th output, I assume you mean the fault or emergency shutdown. This output is on if all is well, and opens on fault (the parameters are programmable but essentially are a hi-hi and lo-lo voltage). The output holds on a small relay; when the relay is de-energized, it sends 24 volts to the shunt trip on the breakers for the PV in, controller out, generator in, and inverter. Tripping these breakers effects a complete electrical shutdown.
I am using these breakers and shunt trip accessory:

View attachment 55424View attachment 55425
The shunt trip has auxiliary contacts which I use to cut the trip circuit once the shunt has tripped the breaker.
I don't expect that this circuit should ever trip unless something goes haywire but decided the added complication and expense is worht having for peace of mind. The system is unattended for part of the year.
Hope that helps.
Bob,

really like your system and attention to detail.

I have EVE cells in a mobile application using the SBMS0 as well. Worse thing about the cells is the busbar connections. Looks like you have that sorted.

I also set up a "type 5 " emergency cut off - but only on the charge side of my system. I used a 220amp Victron Battery Protect to cut charging, which is not ideal. Some voltage drop through the SSR, even at 60-70 amps. Also, terrible lug connectors.

I had previously purchased a Midnight solar breaker (rebranded Carling ) with a built in trip shunt for the emergency cut off, but it turns out the trip shunt does not de-energize itself after tripping - so I gave up. Would be a better solution and avoid the voltage drop if I could make it work.

I would like to set up some alarms in the system. I like your idea of an alarm at SOC of 35%. I assume you are using a NC SSR for this?
 
Back
Top