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

[deckofficer]

I have a gift for all of us early adopters, and that is a peek into the marine diesel/electric industry as it is gearing up. New trade magazine that will allow download of their 1st edition for free.
UKIP Media & Events

Click on "Read now"

Some amazing advancements in batteries by the big players. Also a new a-hr rating system that finally addresses the difference in capacity ratings of lead acid vs LiFePO4, and guess what, we as a group of early adopters already came up with that relationship being 2 to 1 on our own (160 ahr LiFePO4 = 320 ahr lead acid). Many advertisements of LiFePO4 complete batteries with built in BMS for plug and play. A lot more players in this field than I knew of, so indicates the future of batteries.

This is the first magazine introduction that I plan to read cover to cover since "Multihulls" came out in the mid 70's. For me at least, it is that good.

I would be interested in comments from others that read this 1st issue, just post to this thread.

[caipirinha]

Cell-log 8 accuracy
I was getting quite ready to order one Cell-Log 8 when i came across the following user comment

Quote
Nice TOY, but can't be taken seriously. Readings are temperature dependent, because no precise internal voltage reference is used and all input voltages pass through diodes with -2.1mV/K behavior. So calibration with 1mV resolution is nice .. but valid on one temperature only. Tested with Fluke 187 DMM.

Unquote

That worries me since if trying to limit LiFePO4 to within 0.02 V (whether charge or discharge alarm and/or cutoff) a variation of ambient temperature of 10°c or more would defeat the effectiveness of the Cell-Log.

Any other user having similar experience ? Or to the contrary, a check at various ambient temperatures showing stable readings ?

[goboatingnow]

Originally Posted by caipirinha

Cell-log 8 accuracy
I was getting quite ready to order one Cell-Log 8 when i came across the following user comment

Quote
Nice TOY, but can't be taken seriously. Readings are temperature dependent, because no precise internal voltage reference is used and all input voltages pass through diodes with -2.1mV/K behavior. So calibration with 1mV resolution is nice .. but valid on one temperature only. Tested with Fluke 187 DMM.

Unquote

That worries me since if trying to limit LiFePO4 to within 0.02 V (whether charge or discharge alarm and/or cutoff) a variation of ambient temperature of 10°c or more would defeat the effectiveness of the Cell-Log.

Any other user having similar experience ? Or to the contrary, a check at various ambient temperatures showing stable readings ?

Why would u want such accuracy

Dave

[ebaugh]

Originally Posted by caipirinha

Cell-log 8 accuracy
I was getting quite ready to order one Cell-Log 8 when i came across the following user comment

Quote
Nice TOY, but can't be taken seriously. Readings are temperature dependent, because no precise internal voltage reference is used and all input voltages pass through diodes with -2.1mV/K behavior. So calibration with 1mV resolution is nice .. but valid on one temperature only. Tested with Fluke 187 DMM.

Unquote

That worries me since if trying to limit LiFePO4 to within 0.02 V (whether charge or discharge alarm and/or cutoff) a variation of ambient temperature of 10°c or more would defeat the effectiveness of the Cell-Log.

Any other user having similar experience ? Or to the contrary, a check at various ambient temperatures showing stable readings ?

I've been using a CellPro 6 for some battery testing. From what I've observed it does not have .001V resolution, even though it displays to that level. I think it's interpreted to some degree, there is a built in tendency to report all cells the same until they start to deviate. All my multimeters show some deviation between cells, even though the CellLog doesn't. But it does show up when the weakest cell starts to move ahead of the others.

However, I don't see an issue with accuracy sufficient to manage LFP on the ends of the discharge curve, say 0-5% SOC or 95-100% SOC since there are bigger voltage deviations there.

I attached some data collected from a CellPro, one shows the cell voltages for 4 cells at the beginning of a charge cycle. You can see they merge, from that point on, they stay in lock step down to .001V until diverging at the end of the charge above about 3.4V.

The CellPro box is consistent though, I attached another chart I just created today. One I'm still analyzing and validating. But look at the the 30AT5 discharge column and the 30A column. Those are from different days, and different cells (new out of the box). But the data collected is essentially the same. The T5 columns are both the same cells cycled from 3.65-2-3.65V, the other columns are a different group of cells, all cycled from 3.6-2.85-3.6V at different rates.

No experience with temperature. All this testing was done at 85-90F.

[roetter]

Originally Posted by ebaugh

I've been using a CellPro 6 for some battery testing. From what I've observed it does not have .001V resolution, even though it displays to that level. I think it's interpreted to some degree, there is a built in tendency to report all cells the same until they start to deviate. All my multimeters show some deviation between cells, even though the CellLog doesn't. But it does show up when the weakest cell starts to move ahead of the others.

However, I don't see an issue with accuracy sufficient to manage LFP on the ends of the discharge curve, say 0-5% SOC or 95-100% SOC since there are bigger voltage deviations there.

I attached some data collected from a CellPro, one shows the cell voltages for 4 cells at the beginning of a charge cycle. You can see they merge, from that point on, they stay in lock step down to .001V until diverging at the end of the charge above about 3.4V.

The CellPro box is consistent though, I attached another chart I just created today. One I'm still analyzing and validating. But look at the the 30AT5 discharge column and the 30A column. Those are from different days, and different cells (new out of the box). But the data collected is essentially the same. The T5 columns are both the same cells cycled from 3.65-2-3.65V, the other columns are a different group of cells, all cycled from 3.6-2.85-3.6V at different rates.

No experience with temperature. All this testing was done at 85-90F.

I have several CellLog 8s. I adjusted all of them to agree with my Mastek multimeter.

I do not have the problem that it tries to 'merge' the cell voltages. I have one cell that stays a little low even after top balancing. The CellLog keeps agreeing with the multi meter showing this cell low.

Since my boat is usually at a similar temperature, I do not see any drifting of the voltage in real life. Anyway, 2mV is no something we are worried about. We need to terminate charging maybe within 50mV (0.05V). I am using a voltmeter with programmable high and low relay to control the cut-offs.

I use the CellLog for checking on cell balance, especially past the upper knee between 3.45V and 3.55V. I have an audible alarm on the CellLog that ail let me know when cells drift more than 50mV apart. Then I just use a automotive head lamp with cables and alligator clips to drain a little bit out of the high cells.

I think the CellLogs are great devices for our purpose.

[ebaugh]

Whoops! I confused the CellLog with the PowerLab....too many Labor Day (USA) beers today.

I have only seen a CellLog, discovered my eyes required a magnifying glass to read it and stopped there. The Powerlab is a different class of product, designed to charge and regeneratively discharge batteries while sending stats to a PC...in our world, mostly for testing, not a permanent installation.

It has no relay output like the CellLog, so can't control external devices. But has been very useful for collecting data and measuring actual cell capacities. It's also useful for correcting balance, since it will charge one cell at a time at up to 30A from any 12V source.

[T1 Terry]

If you want to monitor cells down to 0.02v yet they will vary 0.1v between charging and discharging, then you definitely have a problem. It won't matter what meter you use, the cells themselves won't play that game for you. Why would anyone be interested in measuring cell voltage down to 0.001v, what purpose would it serve?. Stop discharging at 2.8v, that can be 2.75v or 2.85v, it's not that critical, stop charging at 3.6v, that can be 3.55v or 3.65v, again, it's not that critical. These cells are tough buggers, a few hundred Ah or more and you are concerned at anything more accurate than the second decimal point full scale means you have far to much time on your hands. If you were talking about cells with milliamp capacity.... nah, still can't see the value. Now if they were hand grenade cells like Lipo, then maybe a bit more attention may be required if you are trying to squeeze that last little iddy bitty in there, but you are off the planet doing that anyway so invest in a high accuracy fluke meter, but make sure it's a master unit and fully calibrated each time you use it.

T1 Terry

[ebaugh]

That's right for full charge and discharge, but of you want to cycle between 15-85%, it gets more difficult to measure with precision.

Does it matter? Probably not, just stop before full charge and stop before full discharge. Which is easily achievable, but you won't know precisely what's left. A good reason to oversize the bank a little.

[SVI]

[QUOTE=hellosailor;1318765]"Where would even find a 12v, 320 amp charger"
QUOTE]

Not a problem - Ebay. Get 4x80A or 2x200A as reasonable prices - 230v -> 12V server power blocks, I believe.
Some people even use the inverter welder but it has to be heavy-duty, 100% cycle.

[roetter]

I use a JLD5740 programmable voltmeter from lightobjects.com for my low and high cutoff.
It lets you set individually a high trigger say 13.8V. Then you also program the voltage for it to trigger off, say 13.3V. This way your charge source do not cycle on and off all the time.

You do the same for the low triggers. Cut the low-solenoid for example at 11.8V and have come back on at 12.2V. Again, this makes sure you had some current go in before allowing draw from the battery.

The voltmeter has two internal relays with COM, NC, NO, that can probably drive 10As. I drive a relay board with for relays to drive my solenoids and other devices.

I use the CellLog to keep and eye on the individual cell balance. Also, I will hook up an alarm on the CellLog that will trigger on individual cells going way beyond my safe limits and if there is a difference between highest and lowest cell of more that 0.1V.

EVDrive 03/2012 Digital Voltmeter JLD5740 - YouTube

[goboatingnow]

Originally Posted by T1 Terry

If you want to monitor cells down to 0.02v yet they will vary 0.1v between charging and discharging, then you definitely have a problem. It won't matter what meter you use, the cells themselves won't play that game for you. Why would anyone be interested in measuring cell voltage down to 0.001v, what purpose would it serve?. Stop discharging at 2.8v, that can be 2.75v or 2.85v, it's not that critical, stop charging at 3.6v, that can be 3.55v or 3.65v, again, it's not that critical. These cells are tough buggers, a few hundred Ah or more and you are concerned at anything more accurate than the second decimal point full scale means you have far to much time on your hands. If you were talking about cells with milliamp capacity.... nah, still can't see the value. Now if they were hand grenade cells like Lipo, then maybe a bit more attention may be required if you are trying to squeeze that last little iddy bitty in there, but you are off the planet doing that anyway so invest in a high accuracy fluke meter, but make sure it's a master unit and fully calibrated each time you use it.

T1 Terry

spot on, thats why I asked , " why would u want such accuracy"

dave

[caipirinha]

Hello & thanks for the feedback

Originally Posted by T1 Terry

If you want to monitor cells down to 0.02v yet they will vary 0.1v between charging and discharging, then you definitely have a problem. It won't matter what meter you use, the cells themselves won't play that game for you. Why would anyone be interested in measuring cell voltage down to 0.001v, what purpose would it serve?. Stop discharging at 2.8v, that can be 2.75v or 2.85v, it's not that critical, stop charging at 3.6v, that can be 3.55v or 3.65v, again, it's not that critical. These cells are tough buggers, a few hundred Ah or more and you are concerned at anything more accurate than the second decimal point full scale means you have far to much time on your hands. If you were talking about cells with milliamp capacity.... nah, still can't see the value. Now if they were hand grenade cells like Lipo, then maybe a bit more attention may be required if you are trying to squeeze that last little iddy bitty in there, but you are off the planet doing that anyway so invest in a high accuracy fluke meter, but make sure it's a master unit and fully calibrated each time you use it.

T1 Terry

No i don't want to monitor cells to any particular value, but i want to make sure that any tool used to monitor battery condition or trigger alarms will have as few false actions or alarms as possible. The 0.02 V i mentioned is only for a 10°C variation - typically what you will have between early morning and late morning.
In my usual harbour the yearly variation could be about 50°C, after all that results in a variation of 0.1 V which you say is still good enough for end-of-charge cutoff so my fears appear excessive.
I would still prefer that something which is marketed as a measuring device and displays with a resolution of 0.001V would have either a temperature compensation (which seems built into a variety of inexpensive electronic sensors) or would cutoff its display to accurate digits (which for the cell-log 8 would appear to be 0.01 V)
Thanks again for the discussion

[vturbo]

I have thought of another issue regarding charging LiFePo4 batteries noone seems to have covered.

There is potential for the ripple voltage and maybe ripple current from an alternator to heat up and damage the battery banks.

Has anyone had an issue with this or is it still too early to tell? It is a fact that voltage ripple can and eventually will cause damage to a battery if it is above the batteries spec on ripple.

Is there any cheap way to smooth or remove the AC ripple voltage from a charging line? I have seen VW use big capacitors on their alternator lines when the battery is remote e.g. in the boot of a car in order to help protect the ECU and radios etc.

[roetter]

Originally Posted by vturbo

I have thought of another issue regarding charging LiFePo4 batteries noone seems to have covered.

There is potential for the ripple voltage and maybe ripple current from an alternator to heat up and damage the battery banks.

Has anyone had an issue with this or is it still too early to tell? It is a fact that voltage ripple can and eventually will cause damage to a battery if it is above the batteries spec on ripple.

Is there any cheap way to smooth or remove the AC ripple voltage from a charging line? I have seen VW use big capacitors on their alternator lines when the battery is remote e.g. in the boot of a car in order to help protect the ECU and radios etc.

I would think that the LFP batteries act as the super-large capacitor. You would probably measure ripple directly at the alternator, but not at the batteries. With my bench power supply, raising the voltage during charge by just 0.1V usually make the amps go from somewhere in the 20 range all the way to the max of 50A.

[goboatingnow]

Yes in practice the alternator is driving into a very low resistance. No ripple

Dave