LiFePO4 Batteries: Discussion Thread for Those Using Them as House Banks

[careka]

Thanks again Frank.
OK. then it looks like this is something i can do to.
will follow how it goes for you "starters". then when my Leads starts to fail working out. i will change to a bigger charger, and new LiFePO4 Batts.

[Lagoon4us]

The cells were delivered less than 2 months old, they were NEW build sequential and precisely charged to match each other.
I never installed them till we got till Greece as we had sailing to do and some guests on board i never wanted to disrupt etc...

Wait and see how it works out is the go.....

Cheers

[roetter]

I just purchased 8 250 Ah LFP cells from Balqon configured as two 1P4S batteries. I am planning on installing one battery each into two boats. Currently they are sitting on a desk in my garage for balancing and testing.

I am still waiting for my bench power supply.

Here are pictures of the test setup.

[deckofficer]

roetter, welcome to the "no sag" club. I'll bet that 1500 watt inverter pulling 100+ amps won't drop the bank's voltage more than 0.2 volts under load from resting voltage.

[Lagoon4us]

Rolf are the voltages of each cell identical when delivered to you?
I ask because mine were precise and i never had to balance at all, in reality now if a cell goes slightly over a load instantly brings all cells to identical voltages.
I'm thinking cells from China may be computer charged prior to delivery, there attitude is install and use.
Beats chasing cats!
Cheers interested to know the voltages and build dates.

[roetter]

I should mention that the 8 cell I purchased are used cells from the "specials" section in the Balqon store.

This is what I got when they arrived:
Bat1
Cell# V
...92 3.45
...91 3.45
...72 3.45
...71 3.45

Bat 2
Cell# V
...73 3.45
...74 3.44
...188 3.41
...89 3.45
Except for cell #188, they are all pretty even. This cold mean:
- different SOC unlikely
- different batch and with that different resting voltage, maybe

I will keep an eye on this cell when I do the top balancing and then discharge to the lower knee. My bench power supply will come in handy for that.

The batteries have an unknown SOC, but are expected to be at about 50% SOC as that is usually how they are shipped.

Here are two logs from the Junsi Celllog.

The first log shows the the cell voltage for four cells without any load over a period of about 14 hours. Only the Junsi Cellog is attached. I measured one of the cell cables and it had 20 micro amps flowing. So the total amps flowing to the Junsi should be extremely low, maybe 100mA. The voltage drops about 40 mV per cell in this 14 hour period. That seems pretty high. Has anybody else measured this? Am I missing something?


The second log is the first load test with the inverter and a fan-heater. I loaded in several stages. All for just a short time.
1 - No Load (1A inverter idle), standby load of the inverter is about 0.25A
2 - Just the fan - 2.8A
3 - Power setting 1 - 46A
4 - Power setting 2 - 76A
5 - Power setting 4 - 124 A

At no load the "yellow" cell has the highest voltage, but under load it drops to the lowest voltage. This should mean that it has the highest internal resistance.
The "yellow" cell drops about 35mV more than the best cell ("purple").

Can anyone comment on this? Is this a normal variation between cells.

[roetter]

Originally Posted by deckofficer

roetter, welcome to the "no sag" club. I'll bet that 1500 watt inverter pulling 100+ amps won't drop the bank's voltage more than 0.2 volts under load from resting voltage.

Sadly they dropped about 0.6V with 130A load, and then another 0.15 after 5 minutes under this load. At that point the tangent is close to horizontal and the further drop would be about 0.1V for every 15 minutes of almost 130A load. That is 0.5C.

However, that is much less than the same load on 50% charged LA of 840Ah capacity. I could hear all the alarms beeping, starting with the battery monitor and ending to the autopilot.

[Lagoon4us]

Try this if you like connect the cells up as they are intended ie batt1 and batt2 and on the uneven ones put a load on it, i believe they will equalise to the same voltage.

This top and bottom balance for new cells is a crock it's only meant for rehabilitation of the cells after an incident.

Your number one battery is ready for life it needs nothing other than usage it's ready to accept load or charge.

[T1 Terry]

Originally Posted by roetter

I just purchased 8 250 Ah LFP cells from Balqon configured as two 1P4S batteries. I am planning on installing one battery each into two boats. Currently they are sitting on a desk in my garage for balancing and testing.

I am still waiting for my bench power supply.

Here are pictures of the test setup.

Danger, Danger Will Robinson The wires to the Junsi cell logger alarm, the black wire is connected to the logger chassis and the negative pin of the logger, the red wire is switched to negative or links the black wire to the red wire. reading your wiring diagram, you have one of those wires connected to a he battery positive, if you haven't already destroyed the little transistor in the logger, the first alarm signal will fry it.

Next, you show 2 Junsi loggers in the diagram, one for low cell cut and one for high cell cut, yet there is only one logger in the photo One logger can't do both jobs, it can sound an alarm when a preset threshold has been exceeded, but it would switch a relay with any thing that would set off an alarm, so the reaction would be the same for high cell or low cell voltage. You can turn on or off any of the alarm functions to send the signal to the alarm port within the logger, so 2 loggers would be required, one set for high cell cut and one for low cell cut.

Next, nothing other than the logger can be connected to the logger wiring, it will affect the readings, you can have the 2 loggers run off the one wiring run though, but that's all.

Next, you need to fuse each of the logger wires at the cell connection, a short will melt all 5 cables into one lump real quick and that equals a dead short across a 12v lithium battery, quite capable of delivering an instant 3,000 plus amps, sure to spoil your day and your wiring harness.

You already have the low voltage cut so you may as well make it work, but it isn't really needed for the inverter, it has it's own built in, you can't drag one cell low enough to do any damage and still have more than 10.5v at the terminals under load. I would run a cable to the lighting and emergency equipment that bypasses the low voltage cut, you don't want to be trying to sort that stuff out in an emergency without any lights.

The Junsi cell logger alarm is only activated when an alarm threshold is exceeded, as soon as the voltage is back within spec, the alarm stops, you need either a latching relay you need to manually reset or a timer circuit that maintains the alarm signal for a set period, we use 2 mins, seems to work fine with that, 2 mins is enough for the load to pull a high cell voltage down before it reconnects the charging circuit.

Be aware, the alarm circuit in the logger isn't designed for high loads, 500mA I think is the limit, so be very careful switching mechanical relays just using the alarm port.

Sorry to rain on your parade, but better to fid these things out before they are in the boat.

T1 Terry

[T1 Terry]

Originally Posted by roetter

I should mention that the 8 cell I purchased are used cells from the "specials" section in the Balqon store.

This is what I got when they arrived:
Bat1
Cell# V
...92 3.45
...91 3.45
...72 3.45
...71 3.45

Bat 2
Cell# V
...73 3.45
...74 3.44
...188 3.41
...89 3.45
Except for cell #188, they are all pretty even. This cold mean:
- different SOC unlikely
- different batch and with that different resting voltage, maybe

I will keep an eye on this cell when I do the top balancing and then discharge to the lower knee. My bench power supply will come in handy for that.

The batteries have an unknown SOC, but are expected to be at about 50% SOC as that is usually how they are shipped.

Here are two logs from the Junsi Celllog.

The first log shows the the cell voltage for four cells without any load over a period of about 14 hours. Only the Junsi Cellog is attached. I measured one of the cell cables and it had 20 micro amps flowing. So the total amps flowing to the Junsi should be extremely low, maybe 100mA. The voltage drops about 40 mV per cell in this 14 hour period. That seems pretty high. Has anybody else measured this? Am I missing something?


The second log is the first load test with the inverter and a fan-heater. I loaded in several stages. All for just a short time.
1 - No Load (1A inverter idle), standby load of the inverter is about 0.25A
2 - Just the fan - 2.8A
3 - Power setting 1 - 46A
4 - Power setting 2 - 76A
5 - Power setting 4 - 124 A

At no load the "yellow" cell has the highest voltage, but under load it drops to the lowest voltage. This should mean that it has the highest internal resistance.
The "yellow" cell drops about 35mV more than the best cell ("purple").

Can anyone comment on this? Is this a normal variation between cells.

If the resting voltages are higher than 3.4v the cells are as fully charged, anything over that is surface voltage. Measure voltage drop below 13.4v or 3.35v under even the slightest load, there is a min. 0.05v difference between rested and discharge and rested and charging, so as soon as a decent load was applied the cell voltage would have dropped to 3.35v almost instantly, even with the cells fully charged, just the nature of the beast. Once the cells drop to 3.3v, the discharge curve is nearly flat, you can see it at the 3rd decimal point but remember, that's thousandths of a volt.
The cell that reads high when charging and low under load has had a high cell voltage incident, it's not a huge variation so no serious damage has been done, but that is the tell tale sign, internal resistance climbs.

I'll have another look over your logs and see if anything else stands out, but it all looks well within acceptable limits, those thousandth of a volt readings can be very misleading, looking like a problem that really isn't there.

T1 Terry

[T1 Terry]

With that first chart showing the voltage drop with just the logger connected, had the cells just come off a charger or is that how they wer delivered?
You do realise these cells aren't new or near new or as new, they have done some serious work, they will be fine as house batteries, but their days as traction batteries are over.

T1 Terry

[Lagoon4us]

Terry can you draw the correct junsi diagram for four cells please?
cheers Frank

[roetter]

Originally Posted by T1 Terry

Danger, Danger Will Robinson The wires to the Junsi cell logger alarm, the black wire is connected to the logger chassis and the negative pin of the logger, the red wire is switched to negative or links the black wire to the red wire. reading your wiring diagram, you have one of those wires connected to a he battery positive, if you haven't already destroyed the little transistor in the logger, the first alarm signal will fry it.

The logger has no chassis wire. The chassis is plastic. Please explain.

You are right I should put the relay connection into the red-wire portion of the branch to make it work, but that is all. There is nothing to fry. The relay board has transistors driving the relays from a a 12V supply voltage.

Originally Posted by T1 Terry

Next, you show 2 Junsi loggers in the diagram, one for low cell cut and one for high cell cut, yet there is only one logger in the photo One logger can't do both jobs, it can sound an alarm when a preset threshold has been exceeded, but it would switch a relay with any thing that would set off an alarm, so the reaction would be the same for high cell or low cell voltage. You can turn on or off any of the alarm functions to send the signal to the alarm port within the logger, so 2 loggers would be required, one set for high cell cut and one for low cell cut.

Yes I know that I need two loggers, that is why I have two in the diagram. I do not have a charger yet, so this portion is not completely wired. You may see that the relay board portion for the HVC does not have anything connected either.

Originally Posted by T1 Terry

Next, nothing other than the logger can be connected to the logger wiring, it will affect the readings, you can have the 2 loggers run off the one wiring run though, but that's all.

That is what I have done.

Originally Posted by T1 Terry

Next, you need to fuse each of the logger wires at the cell connection, a short will melt all 5 cables into one lump real quick and that equals a dead short across a 12v lithium battery, quite capable of delivering an instant 3,000 plus amps, sure to spoil your day and your wiring harness.

These are AWG 24 wires. There will not be 3000 plus amps. But, yes fuses will be the sensible thing to do.

Originally Posted by T1 Terry

You already have the low voltage cut so you may as well make it work, but it isn't really needed for the inverter, it has it's own built in, you can't drag one cell low enough to do any damage and still have more than 10.5v at the terminals under load. I would run a cable to the lighting and emergency equipment that bypasses the low voltage cut, you don't want to be trying to sort that stuff out in an emergency without any lights.

The inverter is only a test load, nothing else. I rather search for my flash light when I come back on board after several months than come back to murdered batteries.

Originally Posted by T1 Terry

The Junsi cell logger alarm is only activated when an alarm threshold is exceeded, as soon as the voltage is back within spec, the alarm stops, you need either a latching relay you need to manually reset or a timer circuit that maintains the alarm signal for a set period, we use 2 mins, seems to work fine with that, 2 mins is enough for the load to pull a high cell voltage down before it reconnects the charging circuit.

That will be part of the testing. The Junsi manual does not say anything about that. It would be strange if you would not have to clear the alarm manually. It would not work otherwise for the large RC crowd.

If i have to, I will try and see if I can do the time delay with a large capacitor and resistors on the base of the the relay transistor. However, I will have a second test bed with an Arduino micro controller. So I may not go this route at all. So I will only cross this bridge when I have to.

Originally Posted by T1 Terry

Be aware, the alarm circuit in the logger isn't designed for high loads, 500mA I think is the limit, so be very careful switching mechanical relays just using the alarm port.


I know that. As I said before, the alarm leads to the base of a transistor on the relay board. The draw will be in micro amps.

Sorry to rain on your parade, but better to fid these things out before they are in the boat.

T1 Terry

Thanks for the feedback.

[roetter]

Originally Posted by T1 Terry

With that first chart showing the voltage drop with just the logger connected, had the cells just come off a charger or is that how they wer delivered?
You do realise these cells aren't new or near new or as new, they have done some serious work, they will be fine as house batteries, but their days as traction batteries are over.

T1 Terry

I stated in my earlier posts that I don't know the state of charge. My bench top power supply is not here yet, so I can not charge. So I can only assume that they were shipped at about SOC50%, which is standard procedure.

Yes, I also stated before that they are from "specials / used section" in the Balqon store. That is why I want to test them before they go into the boat.

[T1 Terry]

The logger has no chassis wire. The chassis is plastic. Please explain.

You are right I should put the relay connection into the red-wire portion of the branch to make it work, but that is all. There is nothing to fry. The relay board has transistors driving the relays from a a 12V supply voltage.

The manual with the cell logger is very deceiving in it's explanation of how the alarm port functions. The case is plastic but the metal frame around the mini USB port makes up part of the chassis, this and the black wire in the alarm port are all connected to the negative pin on the logger so ultimately connected to the battery negative terminal. By your wiring diagram and photos you have the red wire of these 2 wires connected to the battery positive, when the alarm is activated there will be a dead short across these 2 wires. The red wire on the alarm port is the wire that is switch to battery negative as the transistor links the black wire to the red wire. If the device to be switched is powered by the same battery as the logger then this red wire is the negative leg of what ever you want to switch.

This bit is not for you in your circuit, just general information for those reading this post.
If you want to switch a smoke alarm, it is powered by it's own 9v battery, so in that case the black alarm port wire would go to one side of the test button switch and the red wire would go to the other side, then when the transistor linked the black and red wire of the alarm port it would be the same as pressing the test button on the smoke alarm.

These are AWG 24 wires. There will not be 3000 plus amps. But, yes fuses will be the sensible thing to do.

I know the little wires won't carry 3,000 amps, that was just to say there is enough current available to make them glow red hot and melt the insulation around all the wires they touch creating a possible fire in the wiring loom.

That will be part of the testing. The Junsi manual does not say anything about that. It would be strange if you would not have to clear the alarm manually. It would not work otherwise for the large RC crowd.

There is no latching within the cell logger for the alarm port, the RC modellers only use this port to trigger a counter or similar circuit to let them know if an alarm condition has occurred and how often it occurred, then they down load the log file to see what happened to trigger the alarm.

I stated in my earlier posts that I don't know the state of charge. My bench top power supply is not here yet, so I can not charge. So I can only assume that they were shipped at about SOC50%, which is standard procedure.

The first log shows the cell voltage for four cells without any load over a period of about 14 hours. Only the Junsi Cellog is attached. I measured one of the cell cables and it had 20 micro amps flowing. So the total amps flowing to the Junsi should be extremely low, maybe 100mA. The voltage drops about 40 mV per cell in this 14 hour period. That seems pretty high. Has anybody else measured this? Am I missing something?

Judging by the cell voltages they were shipped fully charged, I have no explanation as to why a tiny load from the logger pulled the battery voltage down like that if that was the only load attached, that appears to be quite a large drain down of the cells.
That first log file is over 60hrs, not 14hrs, each 4 vertical bars represents 22hrs 13 mins and 20 sec so I'm assuming the logging was every 20 secs? The drain down is in steps, hard to tell accurately but the load was off for approx 4hrs, then on for approx 1hr, then off again fir 4hrs and so on, these are not the usual patterns for a cell logger load as it would be constant draw down.
The reason I asked if they had just been charged was to discount the possibility it was just surface charge absorbing into the cells, if this is how they arrived from Balqon then they would have had plenty of time to absorb any surface charge if they were going to, so the drop in cell voltage from 3.45v down to 3.35v appears to be the result of a switching load, not the cell logger load.

The second graph with the inverter load test has 5 lines, I'm guessing the blue line is accumulated cell voltage, the total value of the other 4 lines adds up to the blue line voltage on the left scale. These cells had already been partially drained before the test commenced, but they still show significant voltage drop for a 0.5CA load but the whole test only dropped the rested voltage from 13.4v to 13.2v, so the capacity still seems to be ok over the length of that test anyway, just the internal resistance appears to be a little higher than expected.

T1 Terry