Jump to content

Balancing LiFePO4s


MtB

Featured Posts

7 minutes ago, Mike the Boilerman said:

 

You might be right there, but equally you could do your case some good if you accepted there is a possibility as a non-user, you really don't see the full picture. Even if you are right about meter inaccuracies on my meter in particular (I have form for getting sold poorly calibrated meters), it seems likely to me the effects seen by Dr Bob are still real given his meter is known to be accurate.

 

I'll do some more comparison photos in a day or so when I'm back on the boat.

 

This is turning out a bit like live aboard vs non-live aboard boaters, with the former claiming that the latter’s opinions are worthless because they don’t live aboard!

When you say “his meter is known to be accurate” what does that mean? What accuracy and resolution of a meter was it tested against, just at one point or over a range of points?
Much of the “explanations” he gives, whilst valid in themselves, are not relevant to the specific point about the discrepancies between your two meters. And as we mentioned, it is surely questionable that a £30 device can be more than about 1% accurate. That presumes that these meters are genuine - the Chinese have a habit of cloning stuff!

 

“Observational” science is great up to a point, but can lead to misconceptions, eg “I’m standing on this grassy plain and I can see flat ground as far as the eye can see, therefore I deduce the earth is flat!”. It needs the reality check of some theoretical knowledge too. I’m afraid he lost a lot of technical credibility when he said authoritatively that having added (say) 10 AH to a 4S battery, in order to return a cell to the state it was before that charge was added, you have to remove 10/4=2.5 AH from that cell. An absolutely basic and fundamental error, and so far I’m not convinced he realises it.

Link to comment
Share on other sites

6 minutes ago, nicknorman said:

This is turning out a bit like live aboard vs non-live aboard boaters, with the former claiming that the latter’s opinions are worthless because they don’t live aboard!

When you say “his meter is known to be accurate” what does that mean? What accuracy and resolution of a meter was it tested against, just at one point or over a range of points?
Much of the “explanations” he gives, whilst valid in themselves, are not relevant to the specific point about the discrepancies between your two meters. And as we mentioned, it is surely questionable that a £30 device can be more than about 1% accurate. That presumes that these meters are genuine - the Chinese have a habit of cloning stuff!

 

“Observational” science is great up to a point, but can lead to misconceptions, eg “I’m standing on this grassy plain and I can see flat ground as far as the eye can see, therefore I deduce the earth is flat!”. It needs the reality check of some theoretical knowledge too. I’m afraid he lost a lot of technical credibility when he said authoritatively that having added (say) 10 AH to a 4S battery, in order to return a cell to the state it was before that charge was added, you have to remove 10/4=2.5 AH from that cell. An absolutely basic and fundamental error, and so far I’m not convinced he realises it.

 

I am surprised at just how accurate some of the chapo DVMs actually are. Deigning something to 1be % accurate is pretty easy. Designing something to be long term 1% accurate across a wide temperature range is a little harder. Deigning something to be 0.1% accurate over a temperature range is really difficult.

 

I would like lithiums but I don't want to be using bench power supplies to charge up individual cells, that's more hassle than putting a bit of water into my Trojans once every 200 hours.  It feels to me that Dr Bob is turning himself into a human battery management system, and one day he will be replaced by a microprocessor and a few power transistors.:)

 

(still its probably better than using a voltmeter and state of charge tables to turn yourself into a human Smartgage)

 

..........Dave

  • Haha 1
Link to comment
Share on other sites

36 minutes ago, Mike the Boilerman said:

 

You might be right there, but equally you could do your case some good if you accepted there is a possibility as a non-user, you really don't see the full picture. Even if you are right about meter inaccuracies on my meter in particular (I have form for getting sold poorly calibrated meters), it seems likely to me the effects seen by Dr Bob are still real given his meter is known to be accurate.

 

I'll do some more comparison photos in a day or so when I'm back on the boat.

 

I'd be very surprised if there are separate meters (DVMs) for each channel in battery monitors like this, the obvious approach (cheaper and more accurate) is to use an analogue multiplexer (no voltage error) to switch a single DVM across each cell in turn to take a reading. In this case there might be an overall error in the readings due to meter inaccuracy (e.g. 3.33V reading for 3.34V cell voltage) but all the cells will read the same if they have the same voltage. To reduce reading-to-reading error (noise) you'd then internally average multiple readings to get a display where the last digit is reasonably stable. It would be very easy (and cheap) to get cell voltage differences accurate to 1mV.

 

I should also point out that for single-range voltages like this (3V-4V) it's dirt cheap to get an internal DVM which is accurate (absolute reading) to better than 1%, 0.1% is pretty cheap, and even 0.03% (which would give a reliable 3.333V reading with 1mV absolute accuracy) is not that expensive.

 

Not saying for sure that this is how these monitors *are* built, but a single multiplexed DVM is cheaper than separate ones (and averaging costs nothing) so why do it any other way? The absolute accuracy might not be good to the last digit, but the differences should be.

 

None of which stops a cheapskate designer screwing up any product if they try really hard, but there's no reason to assume this here unless somebody has measurements to prove it... ?

Edited by IanD
Link to comment
Share on other sites

5 minutes ago, IanD said:

I'd be very surprised if there are separate meters (DVMs) for each channel in battery monitors like this, the obvious approach (cheaper and more accurate) is to use an analogue multiplexer (no voltage error) to switch a single DVM across each cell in turn to take a reading. In this case there might be an overall error in the readings due to meter inaccuracy (e.g. 3.33V reading for 3.34V cell voltage) but all the cells will read the same if they have the same voltage. To reduce reading-to-reading error (noise) you'd then internally average multiple readings to get a display where the last digit is reasonably stable. It would be very easy (and cheap) to get cell voltage differences accurate to 1mV.

 

I should also point out that for single-range voltages like this (3V-4V) it's dirt cheap to get an internal DVM which is accurate (absolute reading) to better than 1%, 0.1% is pretty cheap, and even 0.03% (which would give a reliable 3.333V reading with 1mV absolute accuracy) is not that expensive.

 

Not saying for sure that this is how these monitors *are* built, but a single multiplexed DVM is cheaper than separate ones (and averaging costs nothing) so why do it any other way? The absolute accuracy might not be good to the last digit, but the differences should be.

 

None of which stops a cheapskate designer screwing up any product if they try really hard, but there's no reason to assume this here unless somebody has measurements to prove it... ?

Well yes but it’s not quite that simple since you are not just measuring 8 voltages all referenced to 0v. You have to change the reference point for each cell since the cells are connected in series. 

Link to comment
Share on other sites

Just now, nicknorman said:

Well yes but it’s not quite that simple since you are not just measuring 8 voltages all referenced to 0v. You have to change the reference point for each cell since the cells are connected in series. 

No you don't because all DVM ICs have a differential input, you would switch the two input pins to the top and bottom of each cell in turn and measure the voltage across it. You need a dual-channel 8:1 analogue multiplexer that can deal with the overall voltage range (easily available) and possibly (depending on the DVM chip) an instrumentation amplifier to reject the common-mode voltage (ditto). All basic electronics, cheap and easily available.

Link to comment
Share on other sites

30 minutes ago, IanD said:

No you don't because all DVM ICs have a differential input, you would switch the two input pins to the top and bottom of each cell in turn and measure the voltage across it. You need a dual-channel 8:1 analogue multiplexer that can deal with the overall voltage range (easily available) and possibly (depending on the DVM chip) an instrumentation amplifier to reject the common-mode voltage (ditto). All basic electronics, cheap and easily available.

Can’t be bothered to look it up but can an analogue multiplexer and DVM ic that are both powered from 3.3v, easily measure the differential voltage across 2 points one being 23v and one 26.5v, and give the result to the same accuracy as measuring across 2 points one being 0v and the other 3.3v?

Link to comment
Share on other sites

2 hours ago, nicknorman said:

Can’t be bothered to look it up but can an analogue multiplexer and DVM ic that are both powered from 3.3v, easily measure the differential voltage across 2 points one being 23v and one 26.5v, and give the result to the same accuracy as measuring across 2 points one being 0v and the other 3.3v?

Nope that's not how it works, you have a high-voltage MUX powered between the battery positive and 0V, then shift the resulting differential voltage down using a high-voltage instrumentation amp. There are also mux/amps available that can do this directly, they can take in a floating (on a high voltage) differential signal from multiple pairs of inputs and shift it down to a ground-referenced output.

 

Go and look it up, this is a standard measurement problem with plenty of off-the-shelf ways to solve it ?

Link to comment
Share on other sites

24 minutes ago, IanD said:

Nope that's not how it works, you have a high-voltage MUX powered between the battery positive and 0V, then shift the resulting differential voltage down using a high-voltage instrumentation amp. There are also mux/amps available that can do this directly, they can take in a floating (on a high voltage) differential signal from multiple pairs of inputs and shift it down to a ground-referenced output.

 

Go and look it up, this is a standard measurement problem with plenty of off-the-shelf ways to solve it ?

That is how it works! The device has to work when connected to anything between 1 and 8 cells (ie 3.3v to 26v) and as has been mentioned, it takes its power only from the first cell, ie a 3.3v supply. Can a mux/amp such as you describe, directly deal with a 22.7v and 26v differential input when powered from 3.3v?

 

As you can probably tell, my expertise is mostly in digital electronics / software. I’m not very good at analogue electronics.

Link to comment
Share on other sites

4 hours ago, nicknorman said:

This is turning out a bit like live aboard vs non-live aboard boaters, with the former claiming that the latter’s opinions are worthless because they don’t live aboard!

When you say “his meter is known to be accurate” what does that mean?

I'm not going to argue with you any more Nick over accuracy of meters after this post. It is a red herring and only serves to confuse others and that could be dangerous if people ignore the messge. What is needed for a cell monitor is not absolute accuracy but consistency of measuring each cell ….so when one cell goes up 10mV, the meter shows it has gone up that much. My BG8S does that as checked by my multimeter over all 4 cells and at voltages between 3.2 and 3.4V Mike should check his monitor against a good multimeter to convince himself his meter is ok.

 

Now the remainder of this post is data for Nick. It is long and boring. Please go to line starting '* *' if you have trouble staying awake.

The real problem (I think you are raising) is you can't use these meters to monitor Li's especially in the plateau range because they are not accurate enough. I agree you cant use them in the plateau range but not for that reason.

Let me present some data to show you that the voltages are real and how they do vary. This will be new to you as you have never seen a system in action.

First up my recording of voltages as I charged my bank to 98% in October after 3 months off shore power. (the page without the pen at the top).

Let me explain the sheet. The column headings are self evident. Amps is what the system was doing ie discharging at the top and charging with a 30A victron IP 22 charger at the bottom of the page.

Cell voltages are in the 4 columns but a lot of the readings have the first '3.' missing ie if it says 486 – that is 3.486V.

The Amp hr deficit is the reading from the BMV on Ahrs used but after 3 months without syncing. This is way out and a reason why I wanted to get up to 100% to re-synchronise. If you look at the readings near the top of the page you can see -222Ahr. At the bottom it is +50Ahr. That is because the -222 should have been -272Ahrs. It was 50 Ahrs out at the point where I terminated charge. I need to change the CE factor on the BMV (now done). The Ahr deficit figures after 16.10 on the 21/10/19 are estimates as the BMV does not count + values..but I know the charge rate and the charge time (unless my watch is not accurate enough....must remember to check it against the talking clock....not)

Total V is the voltage showing on the BG8S and my BMV. They are pretty much the same. The final column outside of the table is the delta in mV from highest to lowest cell.

Now to the data itself. Look first at cell 1. At the bottom of the table, in the last row of data, my Amps are starting to drop from the 28.5A which is the start of the Li's controlling the current it will take. I am therefore nearing full. Maybe another 10Ahrs to go. I don't care. Its nearly full and I can re-synch the BMV and terminate voltage. Cell 1 is the highest V. It got to the knee first . Cell 2 is the laggard. The delta is 104 mV which is fine for me. I could try harder to balance but it is not worth it for now. I will comment in a minute on how to get it better.

Now look at Cell 1 in the middle of the table ie at 11.04Z on the 21/10 just before I started to charge. Cell 1 is the lowest of the 4. That is typical of what I see at the moment. Today at -150Ahs, the at rest voltage in cell 1 is the laggard. Typically I have a delta of 15mV in the plateau.

So we have Cell 1 the lowest at the plateau voltages and Cell 1 the highest when near full. Of course it must be the meter! No. Take a look at the second table. Data from 1st April before I started using them full time off line. Cell 1 was the laggard at mid plateau (always) and also the laggard at full charge. In early July I charged cell 1 up by 10Ahrs (at the same time reducing cells 3 and 4 a lot with the resistor circuit. The meter is not making up these numbers. Yes, you could make up a scenario where the variation in these cell voltages could be explained by some weird variation in measurement of voltage between cells but that is highly unlikely. The voltages respond in a totally expected manner when you individually charge or discharge each cell or charge or discharge the full bank.

By the way, temperature is not a variable as the range was 15-18°C in all cases....batteries under the bed.

* * The key reason for banging on about this is not to argue with you Nick but to explain to peeps how it is not a good idea to react to voltage differences between the cells on the plateau. If a cell is low or high on the plateau in relation to the other cells – do not use that to try and balance. The important example is cell 2. Look at the first table. At 11.42 on 19/10 I was getting low in voltage ie 12.62V so power was being taken from the LA bank. Cell 2 was starting to accelerate downwards which was concerning, ie the delta was 76mV vs my typical 15mV. It is clear that cell 2 is down on SoC as it is the laggard at low AND high voltage. It is however in the pack in the middle of the charge curve – see 11.00 on 21/10. I do need to adjust cell 2 upwards and can do that by putting maybe 10Ahrs into the cell to help balance the top end but I cant make that decision from the plateau voltages. If I tried to auto balance or manual balance from the middle values, I would be putting power into Cell 1 as the priority which is totally the wrong one. If you are using a cell monitor to look at cell voltage it is important that you understand why you only balance at the top or bottom but you can use the plateau voltages to aid your understanding and assist in decisions on how much manual balancing to do.

 

IF there was an issue (not with accuracy – but with consistency across cells) with the meter then this would show itself up with voltage changes that cannot be explained with no accompanying actions. I know this meter now. I know how it varies when you put 1A, 5A or 10A load – cell to cell. I know how it varies when you charge with 5A, 10A or 40A. Do you? If the meter was telling me porkies, I would know.

If you plot the delta mV against charge/discharge (amount of charge or discharge, position on SoC, total voltage) you get smooth lines that are agree with data from other users. That supports the data that says any accuracy issues cell to cell are negligible as Ian above alludes to. I have 16 pages of this data and it is consistent. I know what I am talking about.

Screen Shot 2019-11-13 at 11.36.42.png

Screen Shot 2019-11-13 at 11.37.14.png

  • Greenie 1
Link to comment
Share on other sites

On 11/11/2019 at 17:07, rusty69 said:

An electric heater.

 

My old dad has Lithiums in his glider, apparently he has a special plug in gadget he bought to get the voltage down when storing them.It may have been a light bulb.

 

Carry on everyone.............

I use a light bulb on my Honda Insight batteries to discharge them to 75 volts 

Link to comment
Share on other sites

Bob makes me wonder how I survived these last16 months or so?

In all honesty as both you and I know living with our batteries is far easier than living with LAs, and I had full tractions, not semi tractions, which can take more abuse than other LAs. I know I have a BMS in every battery but at my level of voltage it's passively balancing because the active balance isn't lit up. For me and I have said this all along 20 80% keeps away any problems, I can't be bothered arguing what I know to be true because john, james, Johnv and myself experience daily, that's well balanced cells 

  • Greenie 1
Link to comment
Share on other sites

1 hour ago, peterboat said:

Bob makes me wonder how I survived these last16 months or so?

In all honesty as both you and I know living with our batteries is far easier than living with LAs, and I had full tractions, not semi tractions, which can take more abuse than other LAs. I know I have a BMS in every battery but at my level of voltage it's passively balancing because the active balance isn't lit up. For me and I have said this all along 20 80% keeps away any problems, I can't be bothered arguing what I know to be true because john, james, Johnv and myself experience daily, that's well balanced cells 

I agree with you Peter. In between my 3 monthly skirmishes with 100%, I stick to 30-80% with zero problem, zero intervention and very little thinking.

  • Greenie 1
Link to comment
Share on other sites

4 minutes ago, Dr Bob said:

I agree with you Peter. In between my 3 monthly skirmishes with 100%, I stick to 30-80% with zero problem, zero intervention and very little thinking.

So why do people keep on telling us disaster is just around the corner? ????

Link to comment
Share on other sites

5 minutes ago, peterboat said:

So why do people keep on telling us disaster is just around the corner? ????

In a few years time when we are still enjoying all the benefits of our Li's, others will follow. These 2nd hand batteries will become more and more plentiful and cheaper as EVs crash and their power sources put on the market. It will become the norm for anyone with the ability to manage batteries.

  • Greenie 1
Link to comment
Share on other sites

2 hours ago, nicknorman said:

That is how it works! The device has to work when connected to anything between 1 and 8 cells (ie 3.3v to 26v) and as has been mentioned, it takes its power only from the first cell, ie a 3.3v supply. Can a mux/amp such as you describe, directly deal with a 22.7v and 26v differential input when powered from 3.3v?

 

As you can probably tell, my expertise is mostly in digital electronics / software. I’m not very good at analogue electronics.

The main circuits are powered from (for example) 3.3V, and the high-voltage circuits (e.g. analogue mux) use the highest voltage node (e.g. 26V) as their power supply (they take negligible current from this). Then there are instrumentation amps which can take a differential voltage centred anywhere up to several hundred volts offset and output it relative to 0V -- look at LT6375 as an example.

Link to comment
Share on other sites

3 hours ago, Dr Bob said:

I'm not going to argue with you any more Nick over accuracy of meters after this post.

* * The key reason for banging on about this...

TLDR. If you are going to write another article, can I suggest you crisp it up a bit.

 

You are committing that unforgivable scientific sin of vigorously ignoring evidence that doesn't suit you - I am referring to the pictures of MtB’s 2 monitors. But I realise I am wasting my breath, despite the fact that we agree on much of it, it’s clear you resent anyone with an opinion on any aspect of it that is slightly different from yours, and instead of being able to debate around it, you see it as a threat. A modern ill, of course. I’ll leave you to it.

 

At some point in the future when I have the time and when our Trojans conk out, I will get some lithium batteries. I won’t fudge it by having a massive bank of LAs left in place (no room, apart from anything else), I will have a (probably home made) alternator controller that goes to float when the batteries are at the desired SoC and allows us to continue cruising. Perhaps then you will regard my opinion with less contempt, but then again probably not.

 

1 hour ago, IanD said:

The main circuits are powered from (for example) 3.3V, and the high-voltage circuits (e.g. analogue mux) use the highest voltage node (e.g. 26V) as their power supply (they take negligible current from this). Then there are instrumentation amps which can take a differential voltage centred anywhere up to several hundred volts offset and output it relative to 0V -- look at LT6375 as an example.

I’m tempted to buy one and open it up - they’re only £30!

Edited by nicknorman
Link to comment
Share on other sites

1 hour ago, nicknorman said:

 

 I will have a (probably home made) alternator controller that goes to float when the batteries are at the desired SoC and allows us to continue cruising. P

 

 

I think that would be a very good thing to do as that is a weakness of the currently available kit to easily integrate alternators into these batteries. It will make life a lot more simple for many.

Link to comment
Share on other sites

2 hours ago, nicknorman said:

TLDR. If you are going to write another article, can I suggest you crisp it up a bit.

 

You are committing that unforgivable scientific sin of vigorously ignoring evidence that doesn't suit you - I am referring to the pictures of MtB’s 2 monitors. But I realise I am wasting my breath, despite the fact that we agree on much of it, it’s clear you resent anyone with an opinion on any aspect of it that is slightly different from yours, and instead of being able to debate around it, you see it as a threat. A modern ill, of course. I’ll leave you to it.

 

At some point in the future when I have the time and when our Trojans conk out, I will get some lithium batteries. I won’t fudge it by having a massive bank of LAs left in place (no room, apart from anything else), I will have a (probably home made) alternator controller that goes to float when the batteries are at the desired SoC and allows us to continue cruising. Perhaps then you will regard my opinion with less contempt, but then again probably not.

 

I’m tempted to buy one and open it up - they’re only £30!

If you get round to making this home made alternator controller, and assuming it does proper feedback control of voltage and/or current (rather than just a "switch" to turn off the alternator rotor current) then I would be interested in collaborating on your project.

 

................Dave

Link to comment
Share on other sites

2 hours ago, dmr said:

If you get round to making this home made alternator controller, and assuming it does proper feedback control of voltage and/or current (rather than just a "switch" to turn off the alternator rotor current) then I would be interested in collaborating on your project.

 

................Dave

I don’t think it will be too hard. An off the shelf alternator controller chip with LIN interface. A PIC micro to talk over LIN to the controller to set target voltage / max field current or whatever. Inputs from cell voltages, obviously (is there a BMS thing like the BG-8S that outputs serial data?); from battery current, maybe from alternator output current (neither needs to be particularly accurate); from alternator temperature; from battery temperature; Human interface to select max SoC. (I’m thinking a switch to select max 80% or (near) 100%). I’d probably go for a separate AH counting SoC gauge (BMV etc). Maybe a separate PIC to do emergency low and high voltage disconnect. Anything else?

 

Edit: The AD7280A looks good for battery cell monitoring, has an SPI interface.

Edited by nicknorman
Link to comment
Share on other sites

55 minutes ago, nicknorman said:

I don’t think it will be too hard. An off the shelf alternator controller chip with LIN interface. A PIC micro to talk over LIN to the controller to set target voltage / max field current or whatever. Inputs from cell voltages, obviously (is there a BMS thing like the BG-8S that outputs serial data?); from battery current, maybe from alternator output current (neither needs to be particularly accurate); from alternator temperature; from battery temperature; Human interface to select max SoC. (I’m thinking a switch to select max 80% or (near) 100%). I’d probably go for a separate AH counting SoC gauge (BMV etc). Maybe a separate PIC to do emergency low and high voltage disconnect. Anything else?

 

Edit: The AD7280A looks good for battery cell monitoring, has an SPI interface.

If you use a BMV for Ahr counting or SoC, remember if you are charging only to 80% then the Ahrs and SoC will be out of sync if the parameters are not set up right after a few months. I think Tom's is now reasonably accurate but drifts a bit. Mine was way out. You could of course get it to re sync at 80% but then you have to put your brain in gear if going up to 100%. 

A number of us are just terminating on voltage whereas MP does it properly and terminates on voltage plus tail current. 

Link to comment
Share on other sites

18 minutes ago, Dr Bob said:

If you use a BMV for Ahr counting or SoC, remember if you are charging only to 80% then the Ahrs and SoC will be out of sync if the parameters are not set up right after a few months. I think Tom's is now reasonably accurate but drifts a bit. Mine was way out. You could of course get it to re sync at 80% but then you have to put your brain in gear if going up to 100%. 

A number of us are just terminating on voltage whereas MP does it properly and terminates on voltage plus tail current. 

My system would have the option to terminate at roughly 80% or roughly 100% SoC. The general idea would be that for a normal trip, cruising each day, it would be on 80%. Or it was planned to tie up for a few days then the switch would be set to 100% some time before stopping. When set to 100% it would of course use a combination of overall voltage, individual cell voltages, and charge current to decide when to reduce alternator voltage to a non-charging value.
 

I wasn’t planning to use the BMV for any sort of control or health monitoring, just to give a fuel gauge type of display. As you say it will drift out of sync over time due to integration errors, but with the option to fully charge the batteries and with the right settings on the BMV, it should only slowly go out of sync, and auto-reset when the switch is on 100%.

 

The alternative is to design a built in SoC counter, using data from a proper shunt for short term accuracy, hybridised with voltage vs SoC data for long term accuracy (to remove integration errors). Which I think is what MP did with his kalman filter thingy. But projects that try to do all things at the outset, are doomed to never get finished!

 

Oh the AD7280A chip also has level shifted MOSFET drivers for each cell for balancing. Heat dissipation is always going to be an issue but it might be possible to devise an algorithm to balance the cells (at least partially) following a 100% charge as opposed to during it, thus meaning high currents/power weren’t needed.

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.