Jump to content

Another Lithium battery thread


Dr Bob

Featured Posts

Looks like it's going well for both of you. Yes the cells do behave slightly differently from each other, don't know if brand new cells would be the same or not?

 

Current does drop reasonably quickly at the end of charge with voltage at 14v. 3.5v per cell is easily enough to charge to 100%. It's surprising just how quickly a cell's voltage can rise at the end of charge, although those figures do look quite well balanced to me.

 

As to one cell's voltage dropping slightly, that's not unexpected - the rising voltage on one cell has to come from somewhere if charging at constant voltage. It demonstrates clearly how having a high voltage disconnect must be done on cell level, and just relying on pack level would allow an individual cell to climb too high. 

 

Just a word of warning about using lithium settings on charge sources - these are generally set to much too high a voltage. I presume they are only suitable for the propriety ready made packs from the likes of Victron with their own built in BMS, not for charging bare cells directly. 

 

I would advise manual balancing if any more is required. Don't rely on the balancing function on the monitor - these small devices are designed for monitoring and balancing small rc models, not huge capacity leisure banks!

Link to comment
Share on other sites

7 hours ago, Tom and Bex said:

Looks like it's going well for both of you. Yes the cells do behave slightly differently from each other, don't know if brand new cells would be the same or not?

 

Current does drop reasonably quickly at the end of charge with voltage at 14v. 3.5v per cell is easily enough to charge to 100%. It's surprising just how quickly a cell's voltage can rise at the end of charge, although those figures do look quite well balanced to me.

 

As to one cell's voltage dropping slightly, that's not unexpected - the rising voltage on one cell has to come from somewhere if charging at constant voltage. It demonstrates clearly how having a high voltage disconnect must be done on cell level, and just relying on pack level would allow an individual cell to climb too high. 

 

Just a word of warning about using lithium settings on charge sources - these are generally set to much too high a voltage. I presume they are only suitable for the propriety ready made packs from the likes of Victron with their own built in BMS, not for charging bare cells directly. 

 

I would advise manual balancing if any more is required. Don't rely on the balancing function on the monitor - these small devices are designed for monitoring and balancing small rc models, not huge capacity leisure banks!

Thanks Tom, fully agree 3.5v is fine for 100%. Plan is only ever to go there every 4-8 weeks to check balance and re-set the BMV.

 

Agree also about the lithium settings on the charge sources. Hopefully the solar mppt works without intervention but the charger will be terminated via the BMV total voltage relay opening at 13.8V (at least that is what I will start with).

 

You and MP think they are reasonably well balanced so I will leave them at this for now and see how it goes. I now need to rethink the top end alarm and relay disconnect total bank voltages as my planned numbers are too high. Maybe audible alarm at 13.7v and disconnect at 13.8v.  Also need to think about the voltage to reset as the bank voltage will drop from 13.9v on charge to 13.7v without charge straight away but not sure if this goes down to say 13.5v with no load - so the reconnect voltage would have to be lower than 13.5v otherwise the bank is just being reconnected to the high voltage charge source again.

 

 

 

Link to comment
Share on other sites

6 minutes ago, Sea Dog said:

Well, I can see there are undoubted electrical supply advantages to having Lithium batteries aboard, but surely their biggest draw is that they're a fabulous opportunity for those in need of a new, all consuming hobby!  I think you've probably also invented a new category of neglected wife - the Lithium battery widow! :D

 

You got it! Mrs Bob wanted to get a puppy but we've now got these 4 cells. We are calling the first one Jeremy because under normal load he is the worst and lags behind every one. Usain is the second as he is streets ahead of the others. The third is Teresa as he just seems to follow the leader but a bit behind and the fourth, Boris, as he is up fighting with the rest during normal operation but totally disappears into the background when the charging gets high. A puppy would probably be less work. 

  • Haha 2
Link to comment
Share on other sites

I still think charging to 13.8v and then terminating charge may not get them very well charged, but that depends on charge rate. I feel charge to 14.0v then cut charging will work better, at least on our setup it seems to work well - if we stop charging at 14.0v that's approx 80%.

 

9 hours ago, Sea Dog said:

Well, I can see there are undoubted electrical supply advantages to having Lithium batteries aboard, but surely their biggest draw is that they're a fabulous opportunity for those in need of a new, all consuming hobby!  I think you've probably also invented a new category of neglected wife - the Lithium battery widow! :D

 

That might be the case initially! However having had ours installed since September, I rarely think about them now. Just run engine when convenient and I feel like it for an hour or 2 every now and then, and occasionally glance at BMV to see what percentage they are at.

 

If looking a bit low after couple loads of laundry, soup maker etc, might run engine a bit longer. Now we've got used to them they seem to be performing very consistently, and if they're getting towards 20% (not often now) then it's time to charge for a bit longer. Far less care and monitoring than the old lead acid batteries!

 

The other improvement noticed is the fact our lights don't dim or flicker at all when flushing toilet, running waking machine, starting Eberspacher etc like they used to. 

  • Greenie 1
Link to comment
Share on other sites

5 minutes ago, Tom and Bex said:

I still think charging to 13.8v and then terminating charge may not get them very well charged, but that depends on charge rate. I feel charge to 14.0v then cut charging will work better, at least on our setup it seems to work well - if we stop charging at 14.0v that's approx 80%.

 

 

I repeated the charge today to check out the high voltage alarms and relay disconnect. I set the relay for 13.8v and it did the auto disconnect at about 95% charge. The tail current was starting to fall and was around 4% (using the IP22 30A charger). I will keep an eye on this and see what it is like charging with the alternator and solar. I checked the voltage measurement and on 3 different meters and they are all in the same ball park and similar to MP's which I checked my multimeter against, so not sure why I am looking at a lower voltage.

I'll report back after the weekend once I get to run the engine.

 

Link to comment
Share on other sites

19 hours ago, Dr Bob said:

Do you think they are well balanced? I was worrried one shot off too quickly. If I stop charging at 13.8-9v then they do look ok.

 

Just from the one charge cycle, I can see my 4 cells are different. They behave very differently if under charge or not, i.e. their rate of change is different. I think I will give them names. I read them " the hungry caterpillar" tonight. The duck went looking for one.

 

Yes, I was surprised the Amps dropped that fast, it was using the Victron IP22 30A charger, yet 3.55v seems quite low for 100% charge. I'll try to take 1Ahr out of cell 2 in the morning and then recharge to see if I can balance a bit better then try and run for 3-4 days at 30-80% SoC. 

Ive set the solar up for absorption at 13.9v with a float of 13.2v (can't turn it off) and that seemed fine today. The victron IP22 was in lithium mode and was merrily still in absorption when I turned it off. I must get the disconnect to work tomorrow.

Thanks for the help MP.

 

 

eta... just re-read your post and 3 of your cells sound exactly like mine! Maybe this is a new hobby....classifying LiFePo4 cells.

Hi Bob the reason we decided on 13.8 - 9 volts, is because the batteries do seem to stay balanced up to that point without any external aids. Today has been a dull day with little solar so I have fired up the whispergen, it is still on bulk charging at 37 amps [24 volt system] and has been for over an hour, last time it went into float quickly so clearly the batteries were well charged then.

The midnite classic seems to ramp down very fast as well when getting towards full, one second its putting in 60 amps the next its 1 amp! with my old LAs it would wind down very slowly

Link to comment
Share on other sites

2 hours ago, Tom and Bex said:

 

That might be the case initially! However having had ours installed since September, I rarely think about them now. Just run engine when convenient and I feel like it for an hour or 2 every now and then, and occasionally glance at BMV to see what percentage they are at.

 

 

That's what I am hoping for and I think it will be like that here, once I understand the limits of what they can do.

It's just a steep learning curve when you first install them to see how they behave and learn the voltage settings of alarms and disconnects. Before I instlalled them, I wrote down the settings I was going to use, but that has been now ripped up and I am choosing the settings with a bit of experience. Once I know how they respond to the different charge sources and know the voltage limits, then they will look after themselves.....until something breaks ....like a solar controller and presents them with 20V+.....in which case the alarms and disconnect must take over. For now, its bedtime stories. By the end of spring, it should be dead easy.

 

2 hours ago, peterboat said:

Hi Bob the reason we decided on 13.8 - 9 volts, is because the batteries do seem to stay balanced up to that point without any external aids.

 

 

Yes, all my cells were below 3.500v at 13.8v charge today so happy now they are reasonably well balanced - so they should be fine cycled between 20 and 80% SoC.

Really pleased with them so far. I am sure I made the right decision to put them in.

Amazed that I have been able to wire it all in, do all the crimping of 50mm cable connectors (thanks to MP for pointing me in the right tool to use), fit the BMM8 and BG-8S cell monitoring and then get the BMV linked in to get the BEP motorised switch to work. The only snags are that the BMV 712 meter says it is in 'buzzer mode' but doesnt buzz....will try and investigate tomorrow....and the BG-8S battery monitor is great, but doesnt have a high voltage alarm! T&B used the BC-8S monitor which does have the high voltage alarm but it went out of production last year and the makers replaced it with a new all singing all dancing model (the BG-8S). I managed to track down the BC model ...out of stock in most places.....and it should arrive tomorrow. I will have both in the circuit as the display on the BG model is better.

Big thanks to MP, T&B and Peter for all the help.

Link to comment
Share on other sites

4 minutes ago, Dr Bob said:

That's what I am hoping for and I think it will be like that here, once I understand the limits of what they can do.

It's just a steep learning curve when you first install them to see how they behave and learn the voltage settings of alarms and disconnects. Before I instlalled them, I wrote down the settings I was going to use, but that has been now ripped up and I am choosing the settings with a bit of experience. Once I know how they respond to the different charge sources and know the voltage limits, then they will look after themselves.....until something breaks ....like a solar controller and presents them with 20V+.....in which case the alarms and disconnect must take over. For now, its bedtime stories. By the end of spring, it should be dead easy.

 

Yes, all my cells were below 3.500v at 13.8v charge today so happy now they are reasonably well balanced - so they should be fine cycled between 20 and 80% SoC.

Really pleased with them so far. I am sure I made the right decision to put them in.

Amazed that I have been able to wire it all in, do all the crimping of 50mm cable connectors (thanks to MP for pointing me in the right tool to use), fit the BMM8 and BG-8S cell monitoring and then get the BMV linked in to get the BEP motorised switch to work. The only snags are that the BMV 712 meter says it is in 'buzzer mode' but doesnt buzz....will try and investigate tomorrow....and the BG-8S battery monitor is great, but doesnt have a high voltage alarm! T&B used the BC-8S monitor which does have the high voltage alarm but it went out of production last year and the makers replaced it with a new all singing all dancing model (the BG-8S). I managed to track down the BC model ...out of stock in most places.....and it should arrive tomorrow. I will have both in the circuit as the display on the BG model is better.

Big thanks to MP, T&B and Peter for all the help.

Enjoy them for what they are Bob, a reliable, easy and fast to charge battery that under normal circumstances will last us out:)

  • Greenie 1
Link to comment
Share on other sites

I think voltage settings very much depend on charge rate. Charging a 480ah bank with a 30A charger then not really surprising you are getting to higher state of charge at lower voltage. We charge our 320ah bank at 90-100A+ and that gets them to 80% charged by the time the voltage climbs to 14v.

 

The cells certainly stay very well balanced up to the very end stages of charging to 100% when they can start to drift if pushed too hard. Don't know if this would occur if only charging at 13 8v or not, suspect it would if charging was continued. 

 

By only charging at 30a you're missing out on one of the big advantages of lithium and that is their ability to accept high charge rates - up to 240A with your setup!

 

Agree about learning the correct voltage settings to use with experience. We had to vary ours a few times from my initial settings to get the best compromise between looking after the cells and avoiding false alarms. 

 

If only cycling them from 20-80% then from what I've read they should be good for 10000 cycles plus! I guess only (a very long!) time will tell how likely it is to get close to that figure. We do take ours to 100% every few weeks though to sync the BMV, but so far it seems to be staying in sync remarkably well. That's where I do like MP's setup with his arduino based BMS as I think he said it keeps track of any drift and self corrects. 

  • Greenie 2
Link to comment
Share on other sites

5 minutes ago, Tom and Bex said:

I think voltage settings very much depend on charge rate. Charging a 480ah bank with a 30A charger then not really surprising you are getting to higher state of charge at lower voltage. We charge our 320ah bank at 90-100A+ and that gets them to 80% charged by the time the voltage climbs to 14v.

 

The cells certainly stay very well balanced up to the very end stages of charging to 100% when they can start to drift if pushed too hard. Don't know if this would occur if only charging at 13 8v or not, suspect it would if charging was continued. 

 

By only charging at 30a you're missing out on one of the big advantages of lithium and that is their ability to accept high charge rates - up to 240A with your setup!

 

Agree about learning the correct voltage settings to use with experience. We had to vary ours a few times from my initial settings to get the best compromise between looking after the cells and avoiding false alarms. 

 

If only cycling them from 20-80% then from what I've read they should be good for 10000 cycles plus! I guess only (a very long!) time will tell how likely it is to get close to that figure. We do take ours to 100% every few weeks though to sync the BMV, but so far it seems to be staying in sync remarkably well. That's where I do like MP's setup with his arduino based BMS as I think he said it keeps track of any drift and self corrects. 

I agree on the higher charging rates. I looked at some of MP's data tonight and he was using 30-40A and was reaching 90-95% at 13.8V also.

....but no, I'm not missing out.  I reconfigured my electrics last year to have separates and got the IP22 30A charger as it was cheap (<£200) and is great for LA's ......when its only used as you get back into a marina when the LA's are likely fully charged anyway (different if you are using a suitcase genny!). However I can still change a plug over and use the Combi to charge, which I did yesterday to get to full from 230Ahrs down. The Li's just lapped it up at 80A so it was nearly full after 3 hours and I then switched to the 30A charger so I had more control.

Next step is to try it with the alternator to see what voltage I need to disconnect. I have a 90A alternator so maybe it will be putting it in at 60A. It's all a learning curve.

Tom, have you had any problems setting the buzzer on your BMV? Not sure if mine has a problem.

Link to comment
Share on other sites

55 minutes ago, Tom and Bex said:

That's where I do like MP's setup with his arduino based BMS as I think he said it keeps track of any drift and self corrects. 

Ah. That's very much an ambition at this point. I'm keeping all my logs, and plan to use them to build a suitable algorithm.

 

Goes off, muttering, to Google Kalman filters.

 

MP.

 

 

  • Greenie 1
Link to comment
Share on other sites

20 minutes ago, MoominPapa said:

Ah. That's very much an ambition at this point. I'm keeping all my logs, and plan to use them to build a suitable algorithm.

 

Goes off, muttering, to Google Kalman filters.

 

MP.

 

 

I sold my Kalman filter and bought an ecofan instead.

 

eta, wish I knew what an arduino was?

Edited by Dr Bob
Link to comment
Share on other sites

7 hours ago, MoominPapa said:

Ah. That's very much an ambition at this point. I'm keeping all my logs, and plan to use them to build a suitable algorithm.

 

Goes off, muttering, to Google Kalman filters.

 

MP.

 

Good luck with that! I recently tried to get my head around them but failed miserably! Neural net maybe?

Link to comment
Share on other sites

I think a neural net is what you want. In summary, during the learning phase you show it the answers you want, along with multiple parameters of data that were extant at the time the answers were valid. It learns the correlations between the data and the answers, even though at first these may not be obvious. After the learning phase, it can then work out the current “answer” from live data sets.

 

 

Link to comment
Share on other sites

17 hours ago, Tom and Bex said:

I think voltage settings very much depend on charge rate. Charging a 480ah bank with a 30A charger then not really surprising you are getting to higher state of charge at lower voltage. We charge our 320ah bank at 90-100A+ and that gets them to 80% charged by the time the voltage climbs to 14v.

 

 

I've just pulled some data off the vrm.victron site for our Lithiums for anyone who is interested.

The first plot is the first charge cycle I did  - done on wednesday. Bank was 50% SoC (ish). Charged at 80A for 2 hours, then at 30A. BMV connected to the victron site at 1300hrs.

Voltage in blue rises and current is  steady at just under 30A to point A. The bank is now nearing full and current starts to drop to a point a B ...where the current is now 12A (which is 3% of original  battery capacity - 480Ahrs). At this point I would estimate we are at 100% SoC - terminal voltage 14.03V. The charger was turned off and the volts/current dropped. At point C, the inverter was turned on so the current dropped more.

It is interesting that as the voltage was rising to 14.0V, the voltage on the LA bank (dont forget it is in parallel) was approaching 14.1V which is the absorbtion voltage of the IP22 in lithium mode, to that is why the blue trace does not start shooting up.

So the following day, I decided to auto terminate the charge at 13.80V as shown in graph 2. If you look at the first graph, the 13.8v point still has 30A going in and this is the same in graph 2. As the voltage on the lithiums hit 13.8v, the auto disconnect worked and isolated the bank, hence no load, hence a gradual decay to 13.4V. By looking at the area under the current curve on graph 1, you can estimate the Ahrs put in between the 13.8V point and the 14.00v point which is circa 15-20Ahrs, so that means likely 13.8V is 95% SoC.

As Tom says, it is likely that this graph will change if 70-80A are going in.

One point to note that the orginal capacity of the batteries was 480A but these are 2nd hand so the capacity may be less. For now I will try and set the disconnect at 13.8V and monitor the Ahrs in/out.

289 charge cycle 1 6th March.jpg

289 charge cycle 2 7th march.png

  • Greenie 2
Link to comment
Share on other sites

5 hours ago, nicknorman said:

I think a neural net is what you want. In summary, during the learning phase you show it the answers you want, along with multiple parameters of data that were extant at the time the answers were valid. It learns the correlations between the data and the answers, even though at first these may not be obvious. After the learning phase, it can then work out the current “answer” from live data sets.

 

 

Genuine question,  are you being flippant or is the code available for a basic neural net that is small enough for home use?

Link to comment
Share on other sites

29 minutes ago, tree monkey said:

Genuine question,  are you being flippant or is the code available for a basic neural net that is small enough for home use?

No, not being flippant. I think the size depends on the number of parameters and volume of data. But within the size of a PC, definitely. Maybe not for an arduino, depends. At least there is no need for fast processing as batteries don’t change second by second.

 

But I don’t think a Kalman filter is relevant. That is for processing several dimensions of data in the presence of noise, and predicting the outcome. It should be possible to measure the necessary parameters with very little noise.

Edited by nicknorman
  • Greenie 1
Link to comment
Share on other sites

38 minutes ago, nicknorman said:

No, not being flippant. I think the size depends on the number of parameters and volume of data. But within the size of a PC, definitely. Maybe not for an arduino, depends. At least there is no need for fast processing as batteries don’t change second by second.

 

That's Interesting thank you

Link to comment
Share on other sites

On 08/03/2019 at 09:28, nicknorman said:

I think a neural net is what you want. In summary, during the learning phase you show it the answers you want, along with multiple parameters of data that were extant at the time the answers were valid. It learns the correlations between the data and the answers, even though at first these may not be obvious. After the learning phase, it can then work out the current “answer” from live data sets.

 

 

It would be really interesting to see what modelling you could do with a neural net on the data from a Lithium bank. The modelling of SoC from voltage on a Lead acid bank (Gibbo's) used only one voltage reading (and its rate of change and then its derivative) but for Lithiums you also have each individual cell voltage...giving 5 voltages to consider. As MP says, each cell performs differently. I havent got a cell voltage logger on mine as MP has so can only auto log the overall bank voltage, but from the limited manual cell logging I am doing, I can see all sorts of trends at the cell level. The rate of change of cell voltage varies significantly from cell to cell under high load or high charge compared to low load - viz one cell will go down faster than another on deep discharge but on high charge they will be very similar. There is therefore 4 times (maybe 5 times) more data to consider in building a model and multiple effects going on all the time. The 'old' PLS or MLR techniques wouldn't be capable of following what is going on but the new techniques of neural nets, topology, sparse data etc are ideal.

The question then though is what could you do with advanced modelling?

From my limited experience it seems easy to see when you are at 100% SoC, ie the tail current drops to 2-5% at a rising charge voltage - and this rise happens quickly - (as Tom and MP have advised).

The charge voltage at this point appears from what Tom has said will vary dependent on Charge rate, so at a charge of 30A, both MP and I have seen a voltage of 14.00V giving 100% and a voltage of 13.80V giving circa 95%. A futher set of Tom's data shows (IIRC) a voltage of nearly 13.9V giving 95% at 40A charge. Tom is reporting 14.0V needed to get over 80% at higher charge rates (80A? - Tom's bank is also 33% smaller than MP and mine hence that 80A is more like 100A for us).  It would be useful to collect data to see if you can plot charge rate versus voltage to give 95% charge. I will try and do it myself via my different charge sources. For each charge source you should be able to identify the voltage limit for 95% charge....and then just avoid that unless you want to re-set your BMV or amp hour counter. Gibbo put a lot of effort into getting a model to determine SoC but in his case, he needed to be accurate as it was important to get to 100%. 90% wasnt good enough. Lithium's are far less demanding. Dont let the voltage get over a certain point for each charge source. ( I guess the next question is how much does temperature vary this decision point). I assume the same is true at the low voltage end....but not a clue as I have not been there yet.

There will be an issue with the amp hour counter varying and I am sure that is where the neural net would work, but just taking the bank up to 100% (or 95%) every 6-8 weeks resolves this ...and you will then understand how much drift you get in that 6-8 week period....but one of the references used earlier in this thread did push the point that you need to go to 95-100% now and again to keep the batteries at their best, claiming a memory effect. So bottom line is that you will go to 100% now and again and that will re-set the BMV/amp hour counter.

It really does seem easy to look after these batteries with little or no input, providing you scope out first your charge source characteristics and learn the limitations of your system.

If anyone has data on voltages at 95% or 100% SoC with different charge rates, then let me have it and I will see if they are comparable.

 

 

  • Greenie 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.