Jump to content

Another Lithium battery thread


Dr Bob

Featured Posts

The weather has been bad, and we've been going slow. As a consequence, the batteries have not been fully charged for three weeks, until yesterday. I was disappointed to note that the amp-hour counter was 100Ah out: it still read -100 Ah when the tail current dropped below C/20. However I've just gone over the logs: in those three weeks we charged (and discharged) a total of about 2000Ah, so 100Ah error is 5%, which doesn't look too bad. It is 25% of the usable capacity of the batteries though, so there's an real incentive to get a more accurate SOC indication that works even if the batteries are only infrequently fullly charged.

 

MP.

 

 

 

 

Edited by MoominPapa
Link to comment
Share on other sites

9 hours ago, MoominPapa said:

The weather has been bad, and we've been going slow. As a consequence, the batteries have not been fully charged for three weeks, until yesterday. I was disappointed to note that the amp-hour counter was 100Ah out: it still read -100 Ah when the tail current dropped below C/20. However I've just gone over the logs: in those three weeks we charged (and discharged) a total of about 2000Ah, so 100Ah error is 5%, which doesn't look too bad. It is 25% of the usable capacity of the batteries though, so there's an real incentive to get a more accurate SOC indication that works even if the batteries are only infrequently fullly charged.

 

MP.

 

I think you have hit the fundamental limitation of an open loop Hall effect sensor. In fact 5% is quite good, considering! Hall effect sensors usually have quite a temperature coefficient both in terms of offset and gain, so I wonder if measuring the temperature of the sensor and applying some compensation in software might help improve it (unless of course it already has such compensation built it).

 

You also have the issue of the differing charge efficiency factors of the LA and the Li batteries, the former of which will vary according to state and rate of charge.

 

Anyway, if the 5% is fairly repeatable it would be easy to build compensation into the code. That would likely improve things. But ultimately I think you should consider using a more conventional resistive shunt.

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

8 hours ago, MoominPapa said:

The history of the last month: green is current and purple is depth of discharge in Ah.

 

MP.

 

Screenshot from 2019-03-19 22-53-32.png

But is it really?  It would be if the Ah in and out are accurate, but what is measuring the in/out energy flow to produce the graph, and what is it calibrated against?  You have said above that the Ah counter is out by 100 Ah so one measurement at least is not perfect.

 

Then there are the twin issues of what is a sufficient accuracy and, once you have calibrated the instruments, how do you know when they need recalibration?

N

Link to comment
Share on other sites

7 minutes ago, BEngo said:

But is it really?  It would be if the Ah in and out are accurate, but what is measuring the in/out energy flow to produce the graph, and what is it calibrated against?  You have said above that the Ah counter is out by 100 Ah so one measurement at least is not perfect.

 

Then there are the twin issues of what is a sufficient accuracy and, once you have calibrated the instruments, how do you know when they need recalibration?

N

Of course the problem is that this is an integrating process, and when you integrate - in this case current - over time, the error is cumulative and can increase without limit over time. This applies no matter how accurate the current measurement is, unless the error is zero which it can never be. So it is inevitable a compromise of how long the device is expected to run without being reset. That is, unless one can include some sort of automatic resetting of SoC based on some other factor. Battery fully charged is of course the normal way of doing this but as we know, one of the major points of lithiums is the absence of need to fully charge. Trying to do it via measuring the rested voltage might be possible, but difficult bearing in mind the flat voltage profile.

Link to comment
Share on other sites

37 minutes ago, nicknorman said:

Trying to do it via measuring the rested voltage might be possible, but difficult bearing in mind the flat voltage profile.

Plus the difficulty of ensuring that it is a truly resting voltage with absolutely zero current draw. Comparing the voltage with the output of something like an AD584LH would be relatively simple I would have thought, but knowing when to do it would be harder. Unless you can think of a simpler solution. 

Link to comment
Share on other sites

10 hours ago, MoominPapa said:

The weather has been bad, and we've been going slow. As a consequence, the batteries have not been fully charged for three weeks, until yesterday. I was disappointed to note that the amp-hour counter was 100Ah out: it still read -100 Ah when the tail current dropped below C/20. However I've just gone over the logs: in those three weeks we charged (and discharged) a total of about 2000Ah, so 100Ah error is 5%, which doesn't look too bad. It is 25% of the usable capacity of the batteries though, so there's an real incentive to get a more accurate SOC indication that works even if the batteries are only infrequently fullly charged.

 

MP.

 

 

 

 

MP, great info.

I am in the process of scoping out my lower voltages and comparing them to yours. In a previous email, you said you had set your 0 limit at the bottom at 12.7V which was at -400Ahrs depleted ( i.e. SoC of 15% ish). That must have been 10 days ago at the bottom voltage on your graph. I guess you were therefore -70Ahr out at that time so you were only -330Ahr depleted so 25% SoC.  Would you agree with that? I am down to 40% SoC (measured with a shunt) and am trying to get it down to 12.7V ....it's 12.9V at the moment with a bit of load. Mine are performing the same as yours at the top .......and wondering if they will be the same at the bottom. I will isolate my bank today as we are moving and run them down further tonight. Mine can't go down fast after 12.7V as they have the 6 LAs behind them.

Link to comment
Share on other sites

56 minutes ago, WotEver said:

Plus the difficulty of ensuring that it is a truly resting voltage with absolutely zero current draw. Comparing the voltage with the output of something like an AD584LH would be relatively simple I would have thought, but knowing when to do it would be harder. Unless you can think of a simpler solution. 

I think there is a simpler solution. You can use the voltage vs current draw to help you.

The voltage variation at rest goes from 12.8V at lowish SoC (20% ish...not been there yet) up to 3.35V when 100%. There is a gradient on that line. It should therefore be possible to pick a point at say 60% SoC at which to reset the Amp hour counter. The voltage is varying 0.1V every 20% SoC ish.

Now, the variation in  voltage with load is a factor of 10 less, but I am seeing very consistent 0.01 - 0.03V variation with say 2-5A draw or charge. I have a few weeks of data and have not looked in detail but I bet there is a good correlation.

If MP therefore filled his lithiums to 100%, discharged to 60% via his amp hour counter. Then spent a few hours recording voltage with varying charge levels ie loads via putting more and more load on, and then varying degrees of solar, he would have enough data to build some graphs and therefore find the 60% SoC point via volatage and charge rate.

There is also another level of detail that MP could do and that is measuring cell voltage against load at the mid point. Some simple linear regression at that SoC point on cell voltages could give a more accurate answer. All the cells will behave differently so monitoring each cell (and MP has the data) could give a relationship that can identify 60%.

You cannot do this with LA's as the surface charge wrecks it. If you take out 10A, then it takes forever to get it back to rest. I am finding with my lithiums that they will be say 0.02V down on drawing 5A. On removing the load they have recovered to the 'at rest' within a minute. (Its easy on my system as I just disconnect the lithiums from the LA bank so no current in or out).

We would be using voltages down to an accuracy of 0.01V which would not be universally be ok, but for an individual system it would work as the voltmeters are consistent (I have 5 different meters with all the various alarms and disconnects and they are consistent in themselves but all with slightly different offsets).

The unknown for me here is the effect of temperature on voltage. I guess MPs temp will be like mine ie 12°C to 25°C (under the bed), which may need to be re-assessed when the weather gets hotter.

If you come up with a predicted volatage under a certain load to achieve 60%, you can then fill them up to 100%, discharge to that voltage/load and compare against the amp hour counter.

I guess you could easily be within 10% with little effort and maybe 5% with a bit of data graphing etc. 5% should be perfectly ok.

I will take mine up to 100% on a 6-8 week basis to synchronise as I havent got an Arduino that can do all that clever stuff. I can estimate my 60% point as above but cant change my BMV setting unless I am at 100%.

 

Link to comment
Share on other sites

2 hours ago, nicknorman said:

I think you have hit the fundamental limitation of an open loop Hall effect sensor. In fact 5% is quite good, considering! Hall effect sensors usually have quite a temperature coefficient both in terms of offset and gain, so I wonder if measuring the temperature of the sensor and applying some compensation in software might help improve it (unless of course it already has such compensation built it).

 

You also have the issue of the differing charge efficiency factors of the LA and the Li batteries, the former of which will vary according to state and rate of charge.

 

Anyway, if the 5% is fairly repeatable it would be easy to build compensation into the code. That would likely improve things. But ultimately I think you should consider using a more conventional resistive shunt.

 

I'm not sure a resistive shunt would help. I  think the error is coming from the integration, rather than the raw readings. I reviewed the code, and found one possible source of systematic error in the current readings. (The scaling code rounds by truncating towards minus inifiinity, not towards zero, so the rounding error always tends to overestimate the current during discharge and underestimate it during charge. I've fixed that now.) The error does always seem to determine SOC as being less than reality, so I could blindly add a fudge-factor. I'd rather not do that. I can check the calibration of the current sensor at no current, by opening the contactor, so I'm happy there's no no systematic bias, at least at zero amps.

 

The graph below adds the compensated battery voltage (ie calculated by determining the internal resistance and removing the resulting voltage drop at the instantaneous current.) Eyeballing, there seems to be enough SOC info in there to get a signal that's good to better than 25% accuracy, so it should be worth trying to build a filter to do that and adjusting SOC based on its output.

 

MP.

 

Screenshot from 2019-03-20 10-43-44.png

1 hour ago, Dr Bob said:

MP, great info.

I am in the process of scoping out my lower voltages and comparing them to yours. In a previous email, you said you had set your 0 limit at the bottom at 12.7V which was at -400Ahrs depleted ( i.e. SoC of 15% ish). That must have been 10 days ago at the bottom voltage on your graph. I guess you were therefore -70Ahr out at that time so you were only -330Ahr depleted so 25% SoC.  Would you agree with that? I am down to 40% SoC (measured with a shunt) and am trying to get it down to 12.7V ....it's 12.9V at the moment with a bit of load. Mine are performing the same as yours at the top .......and wondering if they will be the same at the bottom. I will isolate my bank today as we are moving and run them down further tonight. Mine can't go down fast after 12.7V as they have the 6 LAs behind them.

I'm certainly treating the 400Ah floor with some scepticism for the time being, based on the information I have now.

 

MP.

 

Link to comment
Share on other sites

8 minutes ago, Dr Bob said:

I think there is a simpler solution. You can use the voltage vs current draw to help you.

The voltage variation at rest goes from 12.8V at lowish SoC (20% ish...not been there yet) up to 3.35V when 100%. There is a gradient on that line. It should therefore be possible to pick a point at say 60% SoC at which to reset the Amp hour counter. The voltage is varying 0.1V every 20% SoC ish.

Now, the variation in  voltage with load is a factor of 10 less, but I am seeing very consistent 0.01 - 0.03V variation with say 2-5A draw or charge. I have a few weeks of data and have not looked in detail but I bet there is a good correlation.

If MP therefore filled his lithiums to 100%, discharged to 60% via his amp hour counter. Then spent a few hours recording voltage with varying charge levels ie loads via putting more and more load on, and then varying degrees of solar, he would have enough data to build some graphs and therefore find the 60% SoC point via volatage and charge rate.

There is also another level of detail that MP could do and that is measuring cell voltage against load at the mid point. Some simple linear regression at that SoC point on cell voltages could give a more accurate answer. All the cells will behave differently so monitoring each cell (and MP has the data) could give a relationship that can identify 60%.

You cannot do this with LA's as the surface charge wrecks it. If you take out 10A, then it takes forever to get it back to rest. I am finding with my lithiums that they will be say 0.02V down on drawing 5A. On removing the load they have recovered to the 'at rest' within a minute. (Its easy on my system as I just disconnect the lithiums from the LA bank so no current in or out).

We would be using voltages down to an accuracy of 0.01V which would not be universally be ok, but for an individual system it would work as the voltmeters are consistent (I have 5 different meters with all the various alarms and disconnects and they are consistent in themselves but all with slightly different offsets).

The unknown for me here is the effect of temperature on voltage. I guess MPs temp will be like mine ie 12°C to 25°C (under the bed), which may need to be re-assessed when the weather gets hotter.

If you come up with a predicted volatage under a certain load to achieve 60%, you can then fill them up to 100%, discharge to that voltage/load and compare against the amp hour counter.

I guess you could easily be within 10% with little effort and maybe 5% with a bit of data graphing etc. 5% should be perfectly ok.

I will take mine up to 100% on a 6-8 week basis to synchronise as I havent got an Arduino that can do all that clever stuff. I can estimate my 60% point as above but cant change my BMV setting unless I am at 100%.

 

See the graph I just posted. Doing the lenear regression yields a signal for the volatage which looks promising as an SOC information source but clearly needs more filtering.

 

MP.

 

 

 

Link to comment
Share on other sites

3 hours ago, BEngo said:

But is it really?  It would be if the Ah in and out are accurate, but what is measuring the in/out energy flow to produce the graph, and what is it calibrated against?  You have said above that the Ah counter is out by 100 Ah so one measurement at least is not perfect.

 

Then there are the twin issues of what is a sufficient accuracy and, once you have calibrated the instruments, how do you know when they need recalibration?

N

 

The only data in the graph comes from the battery current measurements. The SOC is generated by integrating the current with time. The independent check is that "battery full" is defined by battery voltage above 13.9 and charge current less that C/20 (24A in this case) The time axis is too long to see the detail, but the SOC was set to zero under this condition 22 days ago, and then again on Monday (and again on Tuesday). When the fully charged condition was reached on Monday, the SOC was still at -100Ah, and then forcibly reset to 0 because the "battery full" condition was satisfied.

 

So there's no guarantee that either of these two  are accurate, but we can say that the two methods of determining full charge (current < C/20 and amp-hours zero)are mutually contradictory. It might be that t he Ah measurement is good, and the state-of-charge from charge current in wrong, but given how the two are generated, I thing that it's more likely that cumulative errors in the integration are the main source of error.

 

MP.

 

Link to comment
Share on other sites

I know you new to lifePo4s are enjoying yourselves but why? I have solar, every day it charges, at the end of the day my state of charge says 13.3 volts next morning they charge again until they are full or the sun stops shining? It really is that simples!

These batteries work and work well, far better than LAs and the only things that will kill them is over or under voltage's so stop worrying and stop scaring the general boater

  • Greenie 1
Link to comment
Share on other sites

11 minutes ago, peterboat said:

I know you new to lifePo4s are enjoying yourselves but why? I have solar, every day it charges, at the end of the day my state of charge says 13.3 volts next morning they charge again until they are full or the sun stops shining? It really is that simples!

These batteries work and work well, far better than LAs and the only things that will kill them is over or under voltage's so stop worrying and stop scaring the general boater

That's great, but I'm using mine differently, and getting accurate answers to questions like "how many days/hours 'till the batteries are flat" and "can I go away for a week and leave the fridge on" would be useful.

 

MP.

 

 

 

 

  • Greenie 1
Link to comment
Share on other sites

4 hours ago, Dr Bob said:

We would be using voltages down to an accuracy of 0.01V...

Hence my suggestion to compare the voltage to an AD584LH, which is cheap, and accurate to 5mV (0.005V) at 10V. Maybe that’s not accurate enough and maybe it’s just too much faffing. 

 

 

Edited by WotEver
Add a bit
Link to comment
Share on other sites

4 hours ago, peterboat said:

I know you new to lifePo4s are enjoying yourselves but why? I have solar, every day it charges, at the end of the day my state of charge says 13.3 volts next morning they charge again until they are full or the sun stops shining? It really is that simples!

These batteries work and work well, far better than LAs and the only things that will kill them is over or under voltage's so stop worrying and stop scaring the general boater

An update from here on progress after being out a few days and why setting up a system is not that straightforward. Once commissioned then yes, simple.

These batteries are outstanding. Sucking up every bit of power thrown at them. None of this watching tail current decaying slowly from 80% full.

In normal use, these batteries are going to be simpler to look after than LA's. We use 100Ahrs overnight (typical) and the last few days the solar has put an additional 50Ahrs back in per day. With 300 useable Ahrs in the 480Ahr bank (maybe more), that means you can go 5 days without running the engine. In the summer with a bit more sun then we will never need to run the engine. No need to fully charge. Once you get into a cruising routine, then maybe once every 6-8 weeks think about taking them up to 100% to synchronise the amp hour counter......and not think about them otherwise. This fits in well with Peter's....it is simple. It really is.....

 

BUT.

 

 

Initially, there is the 'not so simple' construction of a system to prevent overcharging and over -voltage. The 3 peeps on here have all got slightly different DIY systems and mine is different again. Anyone thinking about going the LiFePo4 route needs to pay attention to the differences. The snag I am coming up against is that I had hoped to terminate charging from the various charging devices by voltage alone. Peter does his by voltage but he is ONLY using Solar so easy to control that source. T&B charge normally via the alternator and that backs off when it reaches a voltage and tail current (via an Arduino device) and MP has an Arduino device that disconnects all his charge devices based on voltage AND tail current.

In practice, the voltage target to get to 85% SoC increases with increasing charge current, so I would need to terminate charge at say 14.0V if charging with our alternator on a 45A setting, at say 13.8V for charging via an IP22 at 30A and maybe 13.6-13.7V if the solar is only putting out 10-20A. MP has been trying to tell me this for weeks. Its only now the message has been understood having played with it for 2 weeks.

I have found that tail current drops rapidly over 30 mins or so once SoC gets over 95%. Up to 95% the charge current is rock solid. No reduction at all. If you have a controller that can measure and use the current measurement, then termination is easy. Problem is that you need some sort of 'brain' (ie an Arduino) to monitor voltage and current. Unfortunately the BMV 712 measures tail current but current is not one of the options you can set off alarms or the relay. The Arduino seems a very good solution.

I will try and see if I can find a compromise based on voltage with maybe choosing 13.8V as the terminating voltage and accepting that if charging over 50A I will only get to 70% SoC (but that is fine anyway). The only issue I see then is if I have a day of 4-5hours motoring and getting up 90% (so it doesnt terminate) and then getting lots of solar late in the afternoon. I am not sure if the backstop of the cell monitoring BMS box that disconnects at 3.8V would ever be triggered.

With the Sterling AtoB attached to the alternator I have 4 battery type settings that vary not only the absorption voltage but the current. The two gel settings give me 45A and 55A so I have a number of various permutations. I need a few days of good sunshine to test the high voltage disconnect.

One other problem I found was my two solar controllers fighting with each other. Sorted that by just using the 30A mppt for all the panels (total 500W).

Peter's 'solar only' is a very simple system with no dramas at all. Once I understand the top end disconnection, mine will be very simple as well.

  • Greenie 2
Link to comment
Share on other sites

23 minutes ago, Dr Bob said:

An update from here on progress after being out a few days and why setting up a system is not that straightforward.<snip>

Dunno what any of that means........but it sounds fantastic. Well done!

  • Greenie 1
Link to comment
Share on other sites

1 hour ago, WotEver said:

Hence my suggestion to compare the voltage to an AD584LH, which is cheap, and accurate to 5mV (0.005V) at 10V. Maybe that’s not accurate enough and maybe it’s just too much faffing. 

 

 

I'm using an internal band-gap reference in the Arduino MPU as the reference for the voltage readings, so fairly happy about accuracy.

 

MP.

 

  • Happy 1
Link to comment
Share on other sites

5 hours ago, MoominPapa said:

That's great, but I'm using mine differently, and getting accurate answers to questions like "how many days/hours 'till the batteries are flat" and "can I go away for a week and leave the fridge on" would be useful.

 

MP.

 

 

 

 

I have the whisper gen which in the event of low battery starts and charges, I find leaving the fridge on is a safety device, against solar just putting in an amp at the end, and so maybe eventually overcharging

  • Greenie 1
Link to comment
Share on other sites

I think the time and research involved in the design stage is worth it in the end. I certainly spent a long time reading up and researching the different options before deciding the route I went down. Pleased I did as now they're installed they just sit there and work! Occasionally glance at the BMV to see if I need to charge or not. Definitely less monitoring and much easier to look after than lead acid. 

 

Getting very good results from solar at the moment, much better than expected based on last years performance. Might need to keep eye on charging and voltages as solar input increases to check I've got the optimum settings for the solar controller. Once set up they really are fit and forget though. 

 

As I mentioned previously, we charge to 100% to synchronise the BMV every 4-6 weeks or so, generally to coincide with a cruise to pump out. We find generally the BMV remains fairly accurate now we've tweaked the settings though. It's interesting to see the data from other installs, particularly how much the charge rate affects SOC as various voltage points are reached during charging. One of our alternators has developed a fault and consequently the batteries are now at much higher SOC by the time 14.0v is reached, although it is obviously taking a lot longer. Max charge at present is 65-70a compared to 100a with both alternators working well. 

 

We tend to leave everything switched on when leaving the boat now, it doesn't matter if they sit for a week or more at partial SOC. I figure the low voltage cut out on the fridge will stop the batteries getting excessively drained, and solar (even in winter) will keep up with minor power usage (WiFi router etc).

 

I notice peterboat and me just enjoy using the batteries now they've been installed for a while, enjoying their benefits and superiority over lead acid, and basically ignore them! Hopefully MP and Dr Bob will have the same benefits when their installs settle in and any teething problems are sorted out. It's been interesting reading the different routes that have people have taken to hopefully get the same end result. I think the 4 installs on this thread have taken 4 quite different routes, so will be interesting to follow how these different methods compare and hold up over time. 

  • Greenie 3
Link to comment
Share on other sites

Spent some time hacking this evening, working on the idea of using the battery voltage as another  data source to determine SOC. (See the graph I posted earlier.) Steps were.

 

1) Build simple function to estimate SOC from battery voltage. This is a simple linear scaling between 12.6 and 13.3v over the capacity of the battery.

2) Calculate an error function from the difference between the estimate, above, and the SOC as determined by Ah counting.

 

When these are plotted, I saw a generally increasing error term as time from last Ah reset increased, as expected, with the maximum error being 100Ah, again as expected.

 

3) Calculate a modified SOC, which, for each sample point, is the previous SOC, plus the coulombs in or out, as before, but with an additional term added. which is a small fraction of  the error function above between the SOC and the estimate of SOC from the voltage. If the voltage-based estimate agrees with the coulomb count, this has no effect, but if the voltage-based estimate is higher, the SOC gradually increases until the two agree, and similarly for an error on the other direction. The fraction that works best seems to be about one thousandth.

 

Below is a graph that shows the result. Ignore the labels: the green line is the original (coulomb-counting only SOC) the purple line is the new SOC, with the filter from cell voltage added. and the blue line is the error fumction between the new SOC calculation and the voltage-based estimate. You can see from the blue line that the voltage-based estimate is really noisy, but the noise does not affect the enhanced SOC value. The error stays around zero, and no longer diverges. The previous and new SOC estimates gradually diverge until the old SOC, which is known to be 100Ah too low, is 100Ah below the improved SOC estimate. The two converge again after the old SOC is reset on a full charge condition.

 

This makes me happy: it looks like the new algorithm gives me a good SOC estimate that's resistant to the drift over time that pure coulomb-counting gives. These experiments were done on the PC, using downloaded datasets. News stage is to implement to the algorithm in the BMS and see how well it works in practise.

 

 

MP

Screenshot_2019-03-20_22-57-17.png

  • Greenie 2
Link to comment
Share on other sites

11 hours ago, MoominPapa said:

Spent some time hacking this evening, working on the idea of using the battery voltage as another  data source to determine SOC. (See the graph I posted earlier.) Steps were.

 

1) Build simple function to estimate SOC from battery voltage. This is a simple linear scaling between 12.6 and 13.3v over the capacity of the battery.

2) Calculate an error function from the difference between the estimate, above, and the SOC as determined by Ah counting.

 

When these are plotted, I saw a generally increasing error term as time from last Ah reset increased, as expected, with the maximum error being 100Ah, again as expected.

 

3) Calculate a modified SOC, which, for each sample point, is the previous SOC, plus the coulombs in or out, as before, but with an additional term added. which is a small fraction of  the error function above between the SOC and the estimate of SOC from the voltage. If the voltage-based estimate agrees with the coulomb count, this has no effect, but if the voltage-based estimate is higher, the SOC gradually increases until the two agree, and similarly for an error on the other direction. The fraction that works best seems to be about one thousandth.

 

Below is a graph that shows the result. Ignore the labels: the green line is the original (coulomb-counting only SOC) the purple line is the new SOC, with the filter from cell voltage added. and the blue line is the error fumction between the new SOC calculation and the voltage-based estimate. You can see from the blue line that the voltage-based estimate is really noisy, but the noise does not affect the enhanced SOC value. The error stays around zero, and no longer diverges. The previous and new SOC estimates gradually diverge until the old SOC, which is known to be 100Ah too low, is 100Ah below the improved SOC estimate. The two converge again after the old SOC is reset on a full charge condition.

 

This makes me happy: it looks like the new algorithm gives me a good SOC estimate that's resistant to the drift over time that pure coulomb-counting gives. These experiments were done on the PC, using downloaded datasets. News stage is to implement to the algorithm in the BMS and see how well it works in practise.

 

 

MP

Screenshot_2019-03-20_22-57-17.png

Yes, good work.

 Can you expand on section 3? I guess this is to account for the increased or decreased voltage at load or charge for each voltage measurement. Is it possible to derive a simple relationship between voltage and Amps in/out for a given SoC from you voltage function in 3?

 

Also I guess this was all done at around 14-15 deg C. Any ideas on if this will change at the height of summer when it gets to 25 under the bed? I'm struggling to find voltage vs temp data.

Edited by Dr Bob
Link to comment
Share on other sites

11 hours ago, MoominPapa said:

Build simple function to estimate SOC from battery voltage. This is a simple linear scaling between 12.6 and 13.3v over the capacity of the battery.

At the end of the day isn’t this all you actually require?  If you’re using the voltage measurement to ‘correct’ the calculated SoC why can’t you simply use the voltage measurement as the final arbiter?

  • Greenie 1
Link to comment
Share on other sites

26 minutes ago, WotEver said:

At the end of the day isn’t this all you actually require?  If you’re using the voltage measurement to ‘correct’ the calculated SoC why can’t you simply use the voltage measurement as the final arbiter?

A bit like I have 13.8 it enters float which is 13.6,actually its double as mine is a 24 volt system. I have a disconnect for over and under voltage job done

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.