Cheap LiFePO4 BMS?

[MoominPapa]

1 hour ago, nicknorman said:

Well anyway I have now ordered the AD7280 battery monitoring chip, and the AR6000 alternator regulator chip. I’ll design the PCBs to suit shortly. Each chip / PCB requires a microcontroller to talk to /control it, and the AD7280 talks to its microcontroller over SPI (never used it!), whilst the AR6000 talks over LIN (never used it). So a fair bit of development work to get the communications sorted out, before actual application development and  trials. But the comms can be sorted out at home over winter.

Sounds like an interesting and challenging project.

 

MP.

 

[nicknorman]

Question about passive cell top balancing (ie discharging the highest cell(s) through a resistor).

 

As we know, cell voltage disparities become most noticeable at full charge when the fullest cell scampers up the knee. We also know that it is inadvisable to hold the cells in a high voltage as one might do whilst discharging the highest cell and seeing which is next to hit the knee.

 

So how does one calculate how much to discharge specific cells in order to balance the bank, other than by random guesswork?

 

I know that the “plateau” of mid-charge cell voltage is pretty flat, but the chip I’m proposing to use has an alleged voltage measurement accuracy of +-1.6mV. That is probably the relative accuracy rather than absolute, but it is the former that is relevant.

 

Obviously there is a differential cell resistance issue, but if one takes the case where the current in/out of the battery is close to zero, can the cell voltages be used to determine how many AH need be taken out of each of the higher cells when the battery is well charged, but not fully charged?

 

Dr Bob has said that mid range cell voltages shouldn’t be taken too seriously, but is that just because the £30 Chinese BMS isn’t that accurate?

 

In summary, what is the strategy for top balancing the cells, that could be implemented automatically?

[nicknorman]

14 minutes ago, MoominPapa said:

Sounds like an interesting and challenging project.

 

MP.

 

Winter is a’coming! Last winter I had a great glider air data sensor project which now works really well. I learnt about C, about I2C using C and Pic (only ever done it previously using 8051 and assembler) and One-wire bus. So now I need to add new skills, aka SPI and LIN. It will definitely keep me out of trouble!

Edited November 22, 2019 by nicknorman

[MtB]

15 minutes ago, nicknorman said:

 

So how does one calculate how much to discharge specific cells in order to balance the bank, other than by random guesswork?

The most reliable way is to ask Dr Bob. 
 

Hope that helps ... 
 

 

 

[nicknorman]

Just now, Mike the Boilerman said:

The most reliable way is to ask Dr Bob. 
 

Hope that helps ... 
 

The memory of the microcontroller is not that big and doesn’t have the capacity to run an artificial intellgence simulation of Dr Bob’s entire brain. Nearly, but not quite.

[MtB]

5 minutes ago, nicknorman said:

The memory of the microcontroller is not that big and doesn’t have the capacity to run an artificial intellgence simulation of Dr Bob’s entire brain. Nearly, but not quite.

Surely it is bright enough to dial his mobile though, and ask him to dab in some numbers on his fone keypad?!  

[MoominPapa]

1 hour ago, nicknorman said:

Question about passive cell top balancing (ie discharging the highest cell(s) through a resistor).

 

As we know, cell voltage disparities become most noticeable at full charge when the fullest cell scampers up the knee. We also know that it is inadvisable to hold the cells in a high voltage as one might do whilst discharging the highest cell and seeing which is next to hit the knee.

 

So how does one calculate how much to discharge specific cells in order to balance the bank, other than by random guesswork?

 

I know that the “plateau” of mid-charge cell voltage is pretty flat, but the chip I’m proposing to use has an alleged voltage measurement accuracy of +-1.6mV. That is probably the relative accuracy rather than absolute, but it is the former that is relevant.

 

Obviously there is a differential cell resistance issue, but if one takes the case where the current in/out of the battery is close to zero, can the cell voltages be used to determine how many AH need be taken out of each of the higher cells when the battery is well charged, but not fully charged?

 

Dr Bob has said that mid range cell voltages shouldn’t be taken too seriously, but is that just because the £30 Chinese BMS isn’t that accurate?

 

In summary, what is the strategy for top balancing the cells, that could be implemented automatically?

There's a difference between achieving balance, and maintaining it. The best way to achieve balance in undoubtedly by paralleling all the cells and charging them all to high end-voltage. Failing that, the best strategy I've found it iterative: note which cell(s) start up the handle of the voltage hockey-stick first during charging, remove some charge from them, relative to the others, and repeat. It took quite a few cycles, but I got my bank into good balance this way.

 

To maintain balance, I invented an algorithm which doesn't rely on accurate voltage measurements, just differences. During charge, a five-minute cycle repeats. For the first minute, the mean voltage of the cells is measured. This has to be done with the balancing resistors/MOSFETS turned off, at least with my hardware, because the same connections are used for balancing (at 200mA ) and voltage measurement. For the remaining four minutes of the cycle, some set of the balancing resistors are turned on, The algorithm to determine which works by ordering the cells by voltage. The cell with the lowest voltage is never discharged, and the cell with the highest is always discharged. The middle two may be discharged if they're closer to the highest cell than to the lowest cell. This cycle continues as long as charging continues. It sees to work at maintaining top balance.

 

MP.

 

 

Edited November 23, 2019 by MoominPapa

[Dr Bob]

1 hour ago, Mike the Boilerman said:

The most reliable way is to ask Dr Bob. 
 

Hope that helps ... 
 

 

 

Well guys, I have just spent the evening at my Aunties 90th birthday party in Scouseland so perhaps now is not the time to come up with any answer tonight. We are driving up to Jockland tomorrow so will try and throw some ideas in later tomorrow when we get there. Basically if you charge up and measure cell voltages and track the Ahrs figure and the Amps going in, you should be able to see the times the highest cell passes through the voltage points 3.30V, 3.35V, 3.40 volts etc and then compare to the times the laggard cells get there. The time difference relates to the Ahrs put in during that time as measured by the constant amps in hence the difference in charge between the cells. In my experience if I then say I need to charge cell X more - I find I needed to put twice that amount in. I have repeated this twice and got the same result. At least its in the ball park. The IPA was very good though.

[nicknorman]

9 hours ago, MoominPapa said:

There's a difference between achieving balance, and maintaining it. The best way to achieve balance in undoubtedly by paralleling all the cells and charging them all to high end-voltage. Failing that, the best strategy I've found it iterative: note which cell(s) start up the handle of the voltage hockey-stick first during charging, remove some charge from them, relative to the others, and repeat. It took quite a few cycles, but I got my bank into good balance this way.

 

To maintain balance, I invented an algorithm which doesn't rely on accurate voltage measurements, just differences. During charge, a five-minute cycle repeats. For the first minute, the mean voltage of the cells is measured. This has to be done with the balancing resistors/MOSFETS turned off, at least with my hardware, because the same connections are used for balancing (at 200mA ) and voltage measurement. For the remaining four minutes of the cycle, some set of the balancing resistors are turned on, The algorithm to determine which works by ordering the cells by voltage. The cell with the lowest voltage is never discharged, and the cell with the highest is always discharged. The middle two may be discharged if they're closer to the highest cell than to the lowest cell. This cycle continues as long as charging continues. It sees to work at maintaining top balance.

 

MP.

 

 

Yes sorry I meant maintenance balancing. I think we all agree that prior to installing a battery, all cells should be balanced by connecting them in parallel and top balancing.

 

Is there a particular reason why you only do it when charging? Would the same strategy not work after charging is finished but while the cells are still in quite a high SoC and under zero or quite low discharge?

 

I’m thinking that in our case with a large alternator and ability to charge say a 400AH battery at 150A, the charge phase isn’t going to last very long.
We have a Travelpower and Mastervolt Combi with 100A charger so I could even charge at say 250A (alternator + Combi). Not sure whether that would be good for battery life though! But it sure would be a good way to stuff in a lot of charge in a short space of time!

[MoominPapa]

22 minutes ago, nicknorman said:

Is there a particular reason why you only do it when charging? Would the same strategy not work after charging is finished but while the cells are still in quite a high SoC and under zero or quite low discharge?

Balancing when under discharge is manual control only. You can tell the BMS to remove x Ah from a cell or cells and it will do it for you.

 

The reason I opted to do automatic balancing only on charge is to avoid the doing the equivalent of cutting the legs off a table, a slice at a time, trying to stop it from wobbling. Discharge balance doesn't, in theory, ever have to stop and could therefore completely flatten the battery given enough time. This logic doesn't preclude continuing to balance for a finite time after charging but the method to determine which cells need shunting is completely different in discharge.

 

You could try and come up with procedure that determines an absolute number for the amount of balance required during the charge phase, and then implements that during the subsequent discharge phase.  Since my algorithm is comparative, it doesn't work well for that. It would be an interesting thing to try.  Some combination of "Dr Bob's method" and my "look for the cell climbing the hockey stick" would be a starting point.

 

Most of the balancing boards you can buy seem to have no memory or intelligence at all. The simply work on the instantaneous voltage across a single cell, with no reference to history or the other cells in the string, Almost anything  would be better than that.

22 minutes ago, nicknorman said:

 

I’m thinking that in our case with a large alternator and ability to charge say a 400AH battery at 150A, the charge phase isn’t going to last very long.
We have a Travelpower and Mastervolt Combi with 100A charger so I could even charge at say 250A (alternator + Combi). Not sure whether that would be good for battery life though! But it sure would be a good way to stuff in a lot of charge in a short space of time!

At those charge currents, and that size battery, you would probably hit CV quite early in the charge and start ramping down the current, so not as extreme or as fast it it might seem. Fast charge is good, and I'd certainly experiment, with a fairly conservative charge voltage limit.

 

MP.

 

[peterboat]

2 minutes ago, MoominPapa said:

Balancing when under discharge is manual control only. You can tell the BMS to remove x Ah from a cell or cells and it will do it for you.

 

The reason I opted to do automatic balancing only on charge is to avoid the doing the equivalent of cutting the legs off a table, a slice at a time, trying to stop it from wobbling. Discharge balance doesn't, in theory, ever have to stop and could therefore completely flatten the battery given enough time. This logic doesn't preclude continuing to balance for a finite time after charging but the method to determine which cells need shunting is completely different in discharge.

 

You could try and come up with procedure that determines an absolute number for the amount of balance required during the charge phase, and then implements that during the subsequent discharge phase.  Since my algorithm is comparative, it doesn't work well for that. It would be an interesting thing to try.  Some combination of "Dr Bob's method" and my "look for the cell climbing the hockey stick" would be a starting point.

 

Most of the balancing boards you can buy seem to have no memory or intelligence at all. The simply work on the instantaneous voltage across a single cell, with no reference to history or the other cells in the string, Almost anything  would be better than that.

At those charge currents, and that size battery, you would probably hit CV quite early in the charge and start ramping down the current, so not as extreme or as fast it it might seem. Fast charge is good, and I'd certainly experiment, with a fairly conservative charge voltage limit.

 

MP.

 

I would have thought heat would be an issue,  tesla heat cells up pre charge then cool them during charging, it's how they give them a long life.  Given the size of domestic banks we have and the current available in Nick's case could have issues. 

My drive bank at 36 ish KWHs is always with my solar have a slow and gentle charging experience, so balance between batteries and cells should be good 

[MoominPapa]

1 hour ago, peterboat said:

I would have thought heat would be an issue,  tesla heat cells up pre charge then cool them during charging, it's how they give them a long life.  Given the size of domestic banks we have and the current available in Nick's case could have issues. 

My drive bank at 36 ish KWHs is always with my solar have a slow and gentle charging experience, so balance between batteries and cells should be good 

A quick Google reveals that a Tesla supercharger can charge at up to 150kW and that Tesla batteries have capacity of 60-70 kWh, so charging a Tesla battery involves approx 2C charging rates on a physically much bigger battery which therefore has worse volume to surface area ratio. Nick's proposal is for a physically smaller battery (with better natural cooling properties, therefore) being charged at approx 0.5C. His battery is also unlikely to have been heat-soaked in the Arizona desert or discharged at many-C rates immediately before charging, which a Tesla may have.

 

I find that charging a similar battery at 0.1C or a bit higher all day gives a temperature rise of a couple of degrees. I doubt 0.5C will push the temperature into the danger area unless possibly done in the middle of a heatwave, I'd certainly have temperature sensors and over temperature shutdown, of course, but I'd not expect them to be activated.

 

MP.

 

[peterboat]

45 minutes ago, MoominPapa said:

A quick Google reveals that a Tesla supercharger can charge at up to 150kW and that Tesla batteries have capacity of 60-70 kWh, so charging a Tesla battery involves approx 2C charging rates on a physically much bigger battery which therefore has worse volume to surface area ratio. Nick's proposal is for a physically smaller battery (with better natural cooling properties, therefore) being charged at approx 0.5C. His battery is also unlikely to have been heat-soaked in the Arizona desert or discharged at many-C rates immediately before charging, which a Tesla may have.

 

I find that charging a similar battery at 0.1C or a bit higher all day gives a temperature rise of a couple of degrees. I doubt 0.5C will push the temperature into the danger area unless possibly done in the middle of a heatwave, I'd certainly have temperature sensors and over temperature shutdown, of course, but I'd not expect them to be activated.

 

MP.

 

True however Nissan leads and Renault does don't have this protection and consequently do suffer battery degradation through repeated fast charging,  it's just a thought and something that with Nick's much bigger charging ability has to be taken into consideration 

[IanD]

1 hour ago, peterboat said:

True however Nissan leads and Renault does don't have this protection and consequently do suffer battery degradation through repeated fast charging,  it's just a thought and something that with Nick's much bigger charging ability has to be taken into consideration 

Also IIRC Tesla batteries have a complex liquid-cooling system which is why they can charge really fast...

[WotEver]

8 minutes ago, IanD said:

Also IIRC Tesla batteries have a complex liquid-cooling system which is why they can charge really fast...

Like post #511 then..

[peterboat]

41 minutes ago, IanD said:

Also IIRC Tesla batteries have a complex liquid-cooling system which is why they can charge really fast...

They aren't listening to us Ian  Tesla's have a fantastic heating and cooling system to allow that rapid charging so battery degradation is minimal, Nissan and Renault haven't, so restrict charging speeds yet still suffer degradation far worse than Tesla. I will stick with my safe rates of charging and will have good batteries to will to my kids ?

 

[Dr Bob]

5 hours ago, nicknorman said:

 

 

Is there a particular reason why you only do it when charging? Would the same strategy not work after charging is finished but while the cells are still in quite a high SoC and under zero or quite low discharge?

 

 

I can only do the balancing on mine when you can see one or more of the cells have a big delta over the others so it cant be done on the plateau. Normally that means doing it when charging. As soon as you finish charging and disconnect the charger, the voltage drops back to circa 13.35V quickly and you are back into the plateau. Balancing will take a while if you do it at say 0.5A to 1A and you need to take 10Ahrs out of an individual cell. You dont want to leave the batteries at 100% for 24hrs to balance. Last week, I could see one cell was in the bottom knee so it was possible to balance it then which I did by charging it for 2.5hrs at 5A. Once again you cant balance it on the plateau as once cell can be lower than the rest but could still get to the upper knee first. My cell 1 is like that now. Since Nick has been saying about inaccuracy of voltage measurement, I have been checking all my cell voltages for the last week on both my multimeter and the BG8S (including the excursion down to the bottom knee) and the variation in cell voltage from the BG8S matches the multimeter well.

 

You asked before about how to estimate charge to remove/put in. I am not on the boat this weekend so cant show some real data but the way I have done it is to monitor Cell V, total Ahr and A with time. So the data might look like this, assuming a 30A charge:

 

 Time              Cell 1 Voltage      Cell 2   Voltage      Cell V delta

1400Z                   3.370                  3.400                      30mV

1500Z                   3.400                  3.550                     150mV

(these are just guesses but show the theory)

You can see that cell 2 has gone up by 150mV in 60 mins and the inbalance is showing up with an unacceptable cell delta. You know therefore that 30Ahrs has gone into the bank in that 60 minute period. Cell 1 at 1500Z is at the voltage that cell 2 was at 60 mins ago therefore Cell 1 needs that amount of capacity added or cell 2 needs it removed. Now, where I got it wrong further up this thread, I assumed I would divide the 30Ahrs by 4 as there are 4 cells and therefore I needed to take out 30/4Ahrs = 7.5AHrs of cell 2 to balance.

In practice, what  I found was that when I estimate I needed to take out I actually needed to double it. Not a clue why. The first time I balanced I calculated I needed to remove 40Ahrs from my cell 4 to bring it into the pack and it actually took 80Ahr (in a 480Ahr bank). Current flow was checked with a clamp meter and was exactly as predicted by ohms law. I repeated that with my cell 3 that I estimated was 20Ahr out and it took 40Ahr to get it right. At least I was in the right ball park. At the time I did this, my cell delta before the current began to decay (ie still at 30A) so maybe 95-98% charged was 200mV so quite unbalanced but not as bad as Mikes.

Nick has said dont divide by 4, but I would then be twice the actual needed. The difference is likely to be that the 3.400V on cell 1 at 1500z is still in the plateau so a large degree of guesswork as you dont really know how far the  voltage is behind the other cell. At least this method gives you a guide.

One interesting thing that I really dont understand - how much charge needs to be put in if you charge up a laggard cell to balance. At the same time I did my balancing in July, I had to bring my Cell 1 up by about 50mV. I guessed from the discharging I would need to add maybe 20Ahrs to cell 1. It only took 5Ahr  ie 3.6V/5A charge from a bench power supply for 1 hr - checked using clamp meter. When I charged my laggard cell last week, I was 100mV low on cell 2 and I brought it back into the pack with 12.5Ahrs when I thought I may need far more than that. Again the power supply showed 5A going in as did the clamp meter. This suggests to me that putting charge in requires far less charge than when you try and take it out ...which is pure bo**ox!!!

What you cant do is relate the amount of charge you need in or out to the mV delta - it is related to the time taken to get the voltages the same ie it is related to current/time.

If anyone wants to see the actual data I used to calculate the above I can post it later next week.

MPs system is far better but you have to be clever to do that.

I tried the module Tom suggested to balance and didnt find it much use. Maximum current being transferred was 200mA so to transfer 10 A was going to take 50 hours.

 

[Dr Bob]

47 minutes ago, peterboat said:

They aren't listening to us Ian  Tesla's have a fantastic heating and cooling system to allow that rapid charging so battery degradation is minimal, Nissan and Renault haven't, so restrict charging speeds yet still suffer degradation far worse than Tesla. I will stick with my safe rates of charging and will have good batteries to will to my kids ?

 

The new Model 3 Tesla uses hundreds of panasonic 'small' cells with a honeycomb of heating/cooling running between them. The battery pack looks rather neat...even to the point of 'fantastic'!. There are some good videos about them somwhere on the internet....I looked at them last week but cant for the life of me find them again. It is interesting that Tesla really push the 'only charge to 80%' even on the slower home charging which is at a much lower rate.

[peterboat]

12 minutes ago, Dr Bob said:

The new Model 3 Tesla uses hundreds of panasonic 'small' cells with a honeycomb of heating/cooling running between them. The battery pack looks rather neat...even to the point of 'fantastic'!. There are some good videos about them somwhere on the internet....I looked at them last week but cant for the life of me find them again. It is interesting that Tesla really push the 'only charge to 80%' even on the slower home charging which is at a much lower rate.

Exactly and that's why i charge mine to 80% it maximizes the batteries lifes, same as to much rapid charging reduces battery life,I would rather follow the market leader than destr9a valuable asset 

[nicknorman]

Probably not the right place to post this, sorry, but just took the lid of the Iskra 175A alternator to see how easy it would be to connect an external regulator. Interesting to see the big diodes and the teensy (relatively) field diodes. The brushes and regulator are in one module and look to be brazed/silver soldered onto the regulator chip. I think it will be safest to buy a new module that I can butcher and attach flying leads to, keeping this one intact for “backup”! (ie cockup!). They are about £15 - £20 on eBay. Also, interesting to note that it has fans front and rear and apparently little or no flow right through. Field resistance is about 3 ohms so the max field current limit of 12A on my chosen smart regulator should be plenty.

 

Edited November 23, 2019 by nicknorman

[peterboat]

8 minutes ago, nicknorman said:

Probably not the right place to post this, sorry, but just took the lid of the Iskra 175A alternator to see how easy it would be to connect an external regulator. Interesting to see the big diodes and the teensy (relatively) field diodes. The brushes and regulator are in one module and look to be brazed/silver soldered onto the regulator chip. I think it will be safest to buy a new module that I can butcher and attach flying leads to, keeping this one intact for “backup”! (ie cockup!). They are about £15 - £20 on eBay. Also, interesting to note that it has fans front and rear and apparently little or no flow right through. Field resistance is about 3 ohms so the max field current limit of 12A on my chosen smart regulator should be plenty.

 

It looks a substantial and well built unit

[nicknorman]

7 hours ago, MoominPapa said:

Balancing when under discharge is manual control only. You can tell the BMS to remove x Ah from a cell or cells and it will do it for you.

 

The reason I opted to do automatic balancing only on charge is to avoid the doing the equivalent of cutting the legs off a table, a slice at a time, trying to stop it from wobbling. Discharge balance doesn't, in theory, ever have to stop and could therefore completely flatten the battery given enough time. This logic doesn't preclude continuing to balance for a finite time after charging but the method to determine which cells need shunting is completely different in discharge.

 

You could try and come up with procedure that determines an absolute number for the amount of balance required during the charge phase, and then implements that during the subsequent discharge phase.  Since my algorithm is comparative, it doesn't work well for that. It would be an interesting thing to try.  Some combination of "Dr Bob's method" and my "look for the cell climbing the hockey stick" would be a starting point.

 

Most of the balancing boards you can buy seem to have no memory or intelligence at all. The simply work on the instantaneous voltage across a single cell, with no reference to history or the other cells in the string, Almost anything  would be better than that.

At those charge currents, and that size battery, you would probably hit CV quite early in the charge and start ramping down the current, so not as extreme or as fast it it might seem. Fast charge is good, and I'd certainly experiment, with a fairly conservative charge voltage limit.

 

MP.

 

The AD7280A has a discharge timing thing, in other words you can tell it to turn on one or more discharge MOSFETs, and specify an auto-off time. Unfortunately max time is 36.9 minutes. But then again, the microcontroller can always do it “manually”. I’m thinking of using a PIC that has a built in RTC so it shouldn’t lose track of time even if it reboots. The last thing one wants is a balancing resistor stuck on!

 

Fast charge: yes bound to hit CV fairly soon, but if one can say stuff in 100AH in 24 mins, that is surely good for a brief but effective top up charge.

Edited November 23, 2019 by nicknorman

[nicknorman]

7 hours ago, peterboat said:

I would have thought heat would be an issue,  tesla heat cells up pre charge then cool them during charging, it's how they give them a long life.  Given the size of domestic banks we have and the current available in Nick's case could have issues. 

My drive bank at 36 ish KWHs is always with my solar have a slow and gentle charging experience, so balance between batteries and cells should be good 

The system will definitely have temperature monitoring of the batteries, both to limit alternator voltage to whatever the battery voltage currently is if temperature <0, but also to reduce charge current if batteries get too hot. Our batteries are in the under-boards engine bay at the back of the cabin area and thus, after a day’s summer cruising, are at around 35C. However if desirous of a fast charge, it’s likely that the engine bay and hence batteries would be cold. No need to fast charge at the end of a day’s cruising. It remains to be seen whether there is significant temperature rise charging at say 1/3C for 1/2 hr, but the system will certainly be monitoring it.

 

I should perhaps reinforce that in the event of loss of sensor data or communications between the battery monitoring module and the alternator controller module, the latter will default to a low safe charge voltage, 13.6v or the like.

Edited November 23, 2019 by nicknorman

[nicknorman]

1 hour ago, Dr Bob said:

I can only do the balancing on mine when you can see one or more of the cells have a big delta over the others so it cant be done on the plateau. Normally that means doing it when charging. As soon as you finish charging and disconnect the charger, the voltage drops back to circa 13.35V quickly and you are back into the plateau. Balancing will take a while if you do it at say 0.5A to 1A and you need to take 10Ahrs out of an individual cell. You dont want to leave the batteries at 100% for 24hrs to balance. Last week, I could see one cell was in the bottom knee so it was possible to balance it then which I did by charging it for 2.5hrs at 5A. Once again you cant balance it on the plateau as once cell can be lower than the rest but could still get to the upper knee first. My cell 1 is like that now. Since Nick has been saying about inaccuracy of voltage measurement, I have been checking all my cell voltages for the last week on both my multimeter and the BG8S (including the excursion down to the bottom knee) and the variation in cell voltage from the BG8S matches the multimeter well.

 

You asked before about how to estimate charge to remove/put in. I am not on the boat this weekend so cant show some real data but the way I have done it is to monitor Cell V, total Ahr and A with time. So the data might look like this, assuming a 30A charge:

 

 Time              Cell 1 Voltage      Cell 2   Voltage      Cell V delta

1400Z                   3.370                  3.400                      30mV

1500Z                   3.400                  3.550                     150mV

(these are just guesses but show the theory)

You can see that cell 2 has gone up by 150mV in 60 mins and the inbalance is showing up with an unacceptable cell delta. You know therefore that 30Ahrs has gone into the bank in that 60 minute period. Cell 1 at 1500Z is at the voltage that cell 2 was at 60 mins ago therefore Cell 1 needs that amount of capacity added or cell 2 needs it removed. Now, where I got it wrong further up this thread, I assumed I would divide the 30Ahrs by 4 as there are 4 cells and therefore I needed to take out 30/4Ahrs = 7.5AHrs of cell 2 to balance.

In practice, what  I found was that when I estimate I needed to take out I actually needed to double it. Not a clue why. The first time I balanced I calculated I needed to remove 40Ahrs from my cell 4 to bring it into the pack and it actually took 80Ahr (in a 480Ahr bank). Current flow was checked with a clamp meter and was exactly as predicted by ohms law. I repeated that with my cell 3 that I estimated was 20Ahr out and it took 40Ahr to get it right. At least I was in the right ball park. At the time I did this, my cell delta before the current began to decay (ie still at 30A) so maybe 95-98% charged was 200mV so quite unbalanced but not as bad as Mikes.

Nick has said dont divide by 4, but I would then be twice the actual needed. The difference is likely to be that the 3.400V on cell 1 at 1500z is still in the plateau so a large degree of guesswork as you dont really know how far the  voltage is behind the other cell. At least this method gives you a guide.

One interesting thing that I really dont understand - how much charge needs to be put in if you charge up a laggard cell to balance. At the same time I did my balancing in July, I had to bring my Cell 1 up by about 50mV. I guessed from the discharging I would need to add maybe 20Ahrs to cell 1. It only took 5Ahr  ie 3.6V/5A charge from a bench power supply for 1 hr - checked using clamp meter. When I charged my laggard cell last week, I was 100mV low on cell 2 and I brought it back into the pack with 12.5Ahrs when I thought I may need far more than that. Again the power supply showed 5A going in as did the clamp meter. This suggests to me that putting charge in requires far less charge than when you try and take it out ...which is pure bo**ox!!!

What you cant do is relate the amount of charge you need in or out to the mV delta - it is related to the time taken to get the voltages the same ie it is related to current/time.

If anyone wants to see the actual data I used to calculate the above I can post it later next week.

MPs system is far better but you have to be clever to do that.

I tried the module Tom suggested to balance and didnt find it much use. Maximum current being transferred was 200mA so to transfer 10 A was going to take 50 hours.

 

Interesting, thanks. My difficulty might be that I wasn’t planning to log current in very accurately, but perhaps rather than using the Hall effect device I was thinking of (pretty inaccurate) I could measure the voltage drop in the existing Mastershunt shunt. I guess I’ll have to see how accurate a Hall effect device might be.

 

Otherwise, what I take away from your post is that I note the highest cell voltage, then measure how much charge has to be put in to get the next highest cell to the value the highest cell was when previously noted. Then discharge the top cell by that amount of AH.

 

The trouble with that might be that at high charge currents, differences in cell resistance might make comparing cell voltages rather inaccurate. I’ll have to experiment. I plan to experiment with some very small LiFePO4 cells so it will be easy to create charge currents in proportion to those for the real system.

[Dr Bob]

3 minutes ago, nicknorman said:

 

 

Otherwise, what I take away from your post is that I note the highest cell voltage, then measure how much charge has to be put in to get the next highest cell to the value the highest cell was when previously noted. Then discharge the top cell by that amount of AH.

 

 

Yes, but I only think that works if you are doing it manually and it is only useful to get the ball park figure for how much to drain/fill. If you have a 'brain' in your module trying to work it out then MPs approach shirley is much better?