My first LiFePO4 build to achieve a cheap 105ah 24vDC.

[Mattb4]

Battery from scratch.
This is my first ever attempt to build a workable, cheap as possible, LiFePO4 24vDC using 105ah battery cells.

The cells and BMS were bought on Aliexpress with some items coming from Amazon (tape, battery separators, terminal connectors and 8 awg soft silicone wire) plus the tool boxes were from Walmart. On Singles day (11/11) sale from Aliexpress I bought and had shipped 16 Liitokala 105ah LiFePO4 cells. They took until 12/19 to arrive. I also bought 2 BMS’s of 8S 100a configuration at ~$22 ea. The price on the batteries with all the discounts was ~$480/16 or about $30 a piece. The items from Amazon added up to ~ $85, and the 2 tool boxes were ~$27 from Walmart. All told I have ~$650.00.

I did a voltage check on each cell so that during assembly I could match comparable ones together for each battery. 15 of them were real close ranged from 3.296vDC to 3.300vD (mostly 3.298vDC therefore 70% SOC) however there is one low voltage cell at 3.265vDC or about 50% SOC. It will bear watching. I numbered as #1 for record keeping sake.

Construction consisted of taping each two cells together with a separator. The separators I bought had to be cut down slightly for the cell size. At this point 4 double packs with separators taped together was than placed in the tool box for fit. I found that I unfortunately I had to remove the locking tool box top due to battery height, including the terminal rods, was greater than anticipated by the specs given for the battery. Without those bolts it would have fit. Plenty of side space to put all components plus padding. Battery packs was taken out of box for final assembly.

The BMS balance leads were fitted with terminal connectors, plus numbered, and the double 8 awg power leads with connectors were added as well. As an aside, real fine flexible wire is a pain to fit with connectors. Darn strands want to splay out. All the busbars were connected, tightened with the BMS leads connected. The BMS was taped with nylon spacers between it and the cells with the temperature probe downward. At this point it was back in the box and I used the foam shipping to pad the sides.

Below are images of the parts and build.
{Continued on next post}

 

[Mattb4]

My next task was to charge up the batteries which I did with a LiFePO4 3 amp battery charger. Voltage was at 26.36vDC (B1) and at 26.58vDC (B2) before charging. I did this reaching the 29.33vDC of the chargers full voltage after ~12 hours and than let sit for several hours overnight on the charger.

Removed the charger and let the batteries sit for another day/night to settle out. Voltage was 27.31vDC (B1) and 27.01vDC (B2). I noticed that in B1 I had several cells of different voltage ranging from 3.35 to 3.51vDC. B2 almost all of the individual cells were almost equal of about 3.37vDC but one high one at 3.54vDC. Not perfectly balanced after the first charge cycle. My B2 battery did not appear to charge completely.

I used a 150w load tester to run a load test down to 24vDC. My tester was set for Constant power of 150w during the test. The results were a capacity of 90.6ah/2235.8wh for B1. After the test voltage rose back up to 25.2vDC without load so about ~15% capacity remaining. Not too far off battery ah rating.

The results for B2 were 86.3ah/2244wh. The setting voltage after disconnect of tester was 24.61vDC or about 10% remaining capacity therefore below rated. There likely could be a few additional ah if I took the voltage down to low voltage disconnect but I see no reason to do that. It certainly under peformed. I also expect that several more charge/discharges should help clear up some cell balance issues.

My conclusions are that at the bottom end of LiFePO4 pricing you probably are better to simply buy ready made. Doing that you have a more standard setup battery to connect to. Plus the amount of time and effort to actually assemble everything has to be thought about. Can not say it worked out entirely as I would want. I built a means of covering the top of each battery with scrap wood from my shop. It is not pretty but it is functional and easily pulled off for access. The cell terminals are just below the wood top so there is no electrical path.

One reason I did a battery build instead of buying a ready made 24vDC battery is to be able to reconfigure my setup in the future if I go to a 48vDC AIO. I would just have to replace the two 24vDC BMS with one 16S BMS.

At any rate it has been an interesting 5 day endeavor.

 

[sunshine_eggo]

Nice write up.

I think this solidly demonstrates the need to top balance the cells. Cells should not ship at > 30% charge to meet hazmat shipping requirements though it's pretty common to find them up to 50%. Mine were 28-41% as received (I discharged them before charging them). 3.298V does not represent 70% when cells have been sitting this long.

The voltage matching was better than nothing, but it clearly still left a fair amount on the table.

I'm going to assume a $22 BMS doesn't include active balancing or possibly even low temp charge protection. Most passive balance BMS only do so at about 30-70mA, so it will take 24-48h of continuous balancing PER Ah to bring them in line... that's 10-28 days of continuous balancing. I doubt it would be resolved in "several more cycles."

 

[Mattb4]

Nice write up.

I think this solidly demonstrates the need to top balance the cells. Cells should not ship at > 30% charge to meet hazmat shipping requirements though it's pretty common to find them up to 50%. Mine were 28-41% as received (I discharged them before charging them). 3.298V does not represent 70% when cells have been sitting this long.

The voltage matching was better than nothing, but it clearly still left a fair amount on the table.

I'm going to assume a $22 BMS doesn't include active balancing or possibly even low temp charge protection. Most passive balance BMS only do so at about 30-70mA, so it will take 24-48h of continuous balancing PER Ah to bring them in line... that's 10-28 days of continuous balancing. I doubt it would be resolved in "several more cycles."

The BMS does have a low temperature protection but it is a bargain basement unit without any documentation. I agree that proper cell balancing is going to take awhile. My original LiFePO4 battery bank that was already built as batteries seem to operate erratic until after several week in my PV system. As a note of interest during my rebuild stage all 5 of those paralleled batteries have perfectly matched voltage when I disconnected them. I would not have been surprised to find a big mismatch.

ETA: Regarding initial capacity. The 70% based on cell voltage was in line with the charge time at 3a to full. 70% of 105ah=73.5ah or 31.4ah to recharge.

 

[Steve_S]

Looks Good and a nice setup.
If you had charged the cells to 3.400 and then used a Tester (like a YR-1035+) to match the cell IR's up and built your packs with cells that matched they would be far more balanced and behaved. Top Balancing high to 3.600-3.650 and watching them settle will also indicate which cells are weaker as they'll drop faster.

 

[Mattb4]

Looks Good and a nice setup.
If you had charged the cells to 3.400 and then used a Tester (like a YR-1035+) to match the cell IR's up and built your packs with cells that matched they would be far more balanced and behaved. Top Balancing high to 3.600-3.650 and watching them settle will also indicate which cells are weaker as they'll drop faster.

Yes, Individually charging the cells would give better results but the time involved would have been a lot. Building packs with properly matched cells by resistance would have required not only the test equipment but a lot more cells than the 16 I bought. Buying cheap as I did I am just pleased that they seem to be working OK for now. The nice thing about low expectations is it does not take much to be happy.

Yesterday I integrated them into my PV setup and went through a full charge (had to use grid charging due to weather conditions). Noticed on the top end a temporary high voltage spike (jumped briefly to 29.7vDC from the 28.6 Boost voltage) before ending charge. Seen this before back when I first setup my first battery bank (factory built units) for the PV system.

 

[Skypower]

You might want to figure out what caused that voltage spike. On a regular basis, that’s going to take a toll. 27.8 to 28 is a better range for absorption voltage. Strange that the BMS didn’t disconnect

 

[Steve_S]

If you operate the batteries within the 27.4 (3.425) down to 22.V (2.75-) and have the BMS shutdown at 2.600 (you never brick the BMS and it will always allow charge that way, you'll likely never notice any amiss and will work perfectly fine. When people try to push over 3.450 those cells start to run and next thing you see is Overvolt disconnects.

There is a nasty gotcha with bans which I have had to "cope" with as well.
Example : Take 4 Pack in parallel, charging to full and about 80% there. Now 1 cell in 1 pack is a dud and won't go over 3.350, the other cells do and of course, they will go over the 3.450 so that the pack itself as a whole can reach say 27.2V (3.400 Vpc) and one or more of these may run and hit OVP. Anyways, you finish charging, all packs have reached 27.2 and the settling starts... now Pack 1 with the low cell will cause the other cells within that pack to drop & the other packs in the bank will attempt to backfill that energy. Ultimately, that entire bank will drop to "meet" the lowest cell.

Passive Balancing = No Help as it only burns off voltage from a High Cell wasting it.
Active Balancing = Not much use either, it will move the energy from Higher Cells to the Lower one but if it does not go over 3.35, it gets wasted and the other cells are "forced down" by the active Balancer, then of course the other packs will absolutely try to backfill.
The only real fix is to replace the Limited Cell with an identical one.
IF the "weak Cell" is just "slower" (due to higher IR) but can still reach 3.400 then an Active Balancer could do the trick with proper timing.

Honestly, there is no point in charging higher than 3.500, even better is to charge to 3.450 and float at 3.425 - 3.410 as long as you can reach EndAmps/Tailcurrent which for 100AH is 5A.

ALSO it's quite important to correct for Voltage Losses and dop... My SCC (Midnite) has offsets in the programming to correct for that... I have a 0.2V correction for charging because the SCC would see 27.2V but at Batt Terminals they see 27.0. Sadly they don't have an offset for when no SCC power is incoming, so my software displays it as 0.2V off when SCC at rest. Similar situation @ Inverter/Charger, the offset when charging is NOT the same when discharging only, so to ensure it's charging is correct, I had to adjust & compensate so that when the Inverter reads 27.4 going into the batts that it is actually 27.2 at the batt terminals. ALSO MORE IMPORTANTLY, the Low Voltage Cutoff !!! I cutoff at 21.6V (2.7Vpc) (at batt terminal voltage) and of course this will happen when "not charging" but the bias is inverted so on the Inverter I set it to 21.4 due to the 0.2V drop from batt to Inverter.

As an aside, I've been running with a mix of batteries, 105AH. 174AH & 280AH in my bank. I've gotten a couple of batches of the EVE 105 (A Matched & Batched) and they are quite impressive, they all work so nicely together and all put out 110AH... I am now in final reconfig (finally) and so I gave away my 174's to a deserving neighbour (along with Midnite SCC, Inverter & Panels), my 105's are headed to the bunkhouse and my bank will now be all 280AH packs... Because I am doing this which includes new Steel Cases, JKBMS - Inverter version, I will be extracting my existing cells to transfer them to new cases & BMS. This means it will free up JKBMS' with 2A Active Balancers - If you are interested for a low cost BMS and willing to pay for S&H (small & light enough to go through mail even) send me a PM. These are 150A with 2A Active Balancing. If memory serves, I also have a 150A & 200A New in box units as well, so I'd want a bit more for those... I have the new ones at hand, the ones in use right now will not be available until after my next shipment with the cases cells & BMS' which I expect in Mid Feb.

Hope it helps, Good Luck
Steve

 

[Mattb4]

You might want to figure out what caused that voltage spike. On a regular basis, that’s going to take a toll. 27.8 to 28 is a better range for absorption voltage. Strange that the BMS didn’t disconnect

Pretty sure it is a cell spiking high. As Steve mentioned in his post what can help is to reduce Boost voltage for awhile. I am dropping mine down to 28.2vDC.

 

[Mattb4]

If you operate the batteries within the 27.4 (3.425) down to 22.V (2.75-) and have the BMS shutdown at 2.600 (you never brick the BMS and it will always allow charge that way, you'll likely never notice any amiss and will work perfectly fine. When people try to push over 3.450 those cells start to run and next thing you see is Overvolt disconnects.

There is a nasty gotcha with bans which I have had to "cope" with as well.
Example : Take 4 Pack in parallel, charging to full and about 80% there. Now 1 cell in 1 pack is a dud and won't go over 3.350, the other cells do and of course, they will go over the 3.450 so that the pack itself as a whole can reach say 27.2V (3.400 Vpc) and one or more of these may run and hit OVP. Anyways, you finish charging, all packs have reached 27.2 and the settling starts... now Pack 1 with the low cell will cause the other cells within that pack to drop & the other packs in the bank will attempt to backfill that energy. Ultimately, that entire bank will drop to "meet" the lowest cell.

Passive Balancing = No Help as it only burns off voltage from a High Cell wasting it.
Active Balancing = Not much use either, it will move the energy from Higher Cells to the Lower one but if it does not go over 3.35, it gets wasted and the other cells are "forced down" by the active Balancer, then of course the other packs will absolutely try to backfill.
The only real fix is to replace the Limited Cell with an identical one.
IF the "weak Cell" is just "slower" (due to higher IR) but can still reach 3.400 then an Active Balancer could do the trick with proper timing.

Honestly, there is no point in charging higher than 3.500, even better is to charge to 3.450 and float at 3.425 - 3.410 as long as you can reach EndAmps/Tailcurrent which for 100AH is 5A.

ALSO it's quite important to correct for Voltage Losses and dop... My SCC (Midnite) has offsets in the programming to correct for that... I have a 0.2V correction for charging because the SCC would see 27.2V but at Batt Terminals they see 27.0. Sadly they don't have an offset for when no SCC power is incoming, so my software displays it as 0.2V off when SCC at rest. Similar situation @ Inverter/Charger, the offset when charging is NOT the same when discharging only, so to ensure it's charging is correct, I had to adjust & compensate so that when the Inverter reads 27.4 going into the batts that it is actually 27.2 at the batt terminals. ALSO MORE IMPORTANTLY, the Low Voltage Cutoff !!! I cutoff at 21.6V (2.7Vpc) (at batt terminal voltage) and of course this will happen when "not charging" but the bias is inverted so on the Inverter I set it to 21.4 due to the 0.2V drop from batt to Inverter.

As an aside, I've been running with a mix of batteries, 105AH. 174AH & 280AH in my bank. I've gotten a couple of batches of the EVE 105 (A Matched & Batched) and they are quite impressive, they all work so nicely together and all put out 110AH... I am now in final reconfig (finally) and so I gave away my 174's to a deserving neighbour (along with Midnite SCC, Inverter & Panels), my 105's are headed to the bunkhouse and my bank will now be all 280AH packs... Because I am doing this which includes new Steel Cases, JKBMS - Inverter version, I will be extracting my existing cells to transfer them to new cases & BMS. This means it will free up JKBMS' with 2A Active Balancers - If you are interested for a low cost BMS and willing to pay for S&H (small & light enough to go through mail even) send me a PM. These are 150A with 2A Active Balancing. If memory serves, I also have a 150A & 200A New in box units as well, so I'd want a bit more for those... I have the new ones at hand, the ones in use right now will not be available until after my next shipment with the cases cells & BMS' which I expect in Mid Feb.

Hope it helps, Good Luck
Steve

Good post! We all have to work within the constraints of our individual systems.

Sounds like nice BMS's. I would be interested if I had any spare money. Your cases sounds interesting. One thing I found when trying to come up with a battery case is how little of pre-made cases were available other than top end designs and those for 48vDC mostly. I just wanted a typical plastic case with two top bolt on connections that would end up looking just like a typical battery.

 

[Steve_S]

Matt, you have a PM.
Even 28.2V is pushing over the edge. Really, no point going any higher than 3.500 / 28.0V and even then you WILL have a runner or two... It's all about the Working Range... Battery Companies ONLY test Capacity to ensure that the stated AH is delivered FROM the Working Voltage Range and not the full allowable range. This is the way it is and always has been.

 

[Mattb4]

Matt, you have a PM.
Even 28.2V is pushing over the edge. Really, no point going any higher than 3.500 / 28.0V and even then you WILL have a runner or two... It's all about the Working Range... Battery Companies ONLY test Capacity to ensure that the stated AH is delivered FROM the Working Voltage Range and not the full allowable range. This is the way it is and always has been.

As you mentioned the voltage at my AIO is a bit higher than the voltage my battery sees.

Regarding battery capacity. Frankly knowing that a battery is a chemical device I take capacity as a variable. Depending on environmental and loading conditions capacity will change from cycle to cycle. It is not a volume measurement. Companies tend to have either the desire to state the best case or if they are a bit more honest give a more conservative value. I like to achieve 80% of ratings.

 

[Steve_S]

As you mentioned the voltage at my AIO is a bit higher than the voltage my battery sees.

Regarding battery capacity. Frankly knowing that a battery is a chemical device I take capacity as a variable. Depending on environmental and loading conditions capacity will change from cycle to cycle. It is not a volume measurement. Companies tend to have either the desire to state the best case or if they are a bit more honest give a more conservative value. I like to achieve 80% of ratings.

I wrote this earlier this morning while nursing Coffee #1 - Hell I just started coffee 2, MAN I'M SLOW TODAY ! Thank Goodness no Family Circus to add to it.

This post in this thread is one thing for you to do as well... Everyone needs to do it but far too many don't and then hit walls and won't realize a simple issue can be fixed.

Schneider LiFePo4 Build

Thanks again to everyone on the forum so far in the design help with our system. Due to my rather public problems with our supplier, I've had to put things together a bit at a time and it will take a while before completion. I have a Schneider XWPro a Conext MPPT 100 charge controller and a...

diysolarforum.com

 

[Mattb4]

I wrote this earlier this morning while nursing Coffee #1 - Hell I just started coffee 2, MAN I'M SLOW TODAY ! Thank Goodness no Family Circus to add to it.

This post in this thread is one thing for you to do as well... Everyone needs to do it but far too many don't and then hit walls and won't realize a simple issue can be fixed.

Schneider LiFePo4 Build

Thanks again to everyone on the forum so far in the design help with our system. Due to my rather public problems with our supplier, I've had to put things together a bit at a time and it will take a while before completion. I have a Schneider XWPro a Conext MPPT 100 charge controller and a...

diysolarforum.com

Oh yes, voltage readings done for every point of a PV system (not just AIO/SCC/charger to battery) is really important. It not only allows for corrections but abnormal values can help discover problem areas.

ETA: Went back to the site for purchasing the BMS and found the specs.:
Style 2
Voltage: DC24V
Discharge current: 100A
Charging current (same port): ≤ 50A
Main circuit conduction internal resistance: ≤ 5MQ
Discharge overcurrent protection: 600+10A
Product size: 145 * 65 * 11+1MM
Overcharge protection
Detection voltage: 3.65+0.025V
Detection delay: 1000+500MS
Release voltage: 3.5+0.05V
Overdischarge protection
Detection voltage: 2.35+0.1V
Detection delay: 1000+500MS
Release voltage: 2.7+0.1V
Equilibrium parameters
Detection voltage: 3.5+0.025V
Detection delay: 100+50MS
Equalizing current: 70+5MA
Overcurrent protection
Detection voltage: 50+15MV
Detection delay: 1000+500MS
Current protection: 600+100A
Short circuit protection
Trigger condition: external short circuit
Detection delay: ≤ 350US
Release condition: disconnect the load
Temperature control protection
Charging high temperature protection detection temperature: 75 ° C
Discharge high-temperature protection detection temperature: 75 ° C
Release condition: 65 ° C
Self consumption
Working power consumption: ≤ 40UA
Sleep over discharge power consumption: ≤ 10UA
Working temperature range: -20/+80C

 

[Mattb4]

Now that I have had a few days of operation it appears the 28.2vDC Boost setting is working out good. No more high voltage spikes at end of Boost cycle. It charges up to the 28.2vDC and than attempts to hold it for an absorb period (my unit has a 120 minute absorb default setting) by decreasing and raising charge voltage/current. During this the voltage drifts down to 27.9vDC and than back up to 28.2vDC.

The setup is these two new 105ah batteries plus an older 80ah in parallel. They are wired to my PowMr 3kW AIO. This unit serves my office and bedroom that includes 2 deep freezers.

The last few days solar has been almost none existent so I've been using grid/battery to power the rooms. I charge up each day off grid at 30a and than transfer back to battery/inverter once charged. The cycle battery voltage is down to 26.3vDC overnight and then I transfer back to grid to recharge. Yes the unit could do this automatically but I have been interested in how charging the new batteries was going to work out.

Seems good.

 

[Steve_S]

LFP safe temps are not to exceed 50C/122F - NOT 75C/167F

 

[Mattb4]

LFP safe temps are not to exceed 50C/122F - NOT 75C/167F

So not the best thermal protection from the BMS. Thanks for pointing it out. Not being familiar with all the proper specifications I would never have noticed it.

At some point I likely will redo this. As a first attempt it was a learning experience and time passer.