LiFePO4 - The future for off-grid battery banks?

[CrazyJerry]

Ironically, I only use resting voltages (12 hours!) as a relative state of SOC, but even then, your metering needs to be quite accurate as the spread between full and dead is only 1 volt! Shirt pocket meters are out. The Optimate TS-126 tester lines up with my Fluke 87, so I felt better. This is what I use:

12.7v / 3.175v dead
12.8v / 3.20v 20% <-- I try to never see a resting voltage below this
12.95v / 3.237v 25%
13.1v / 3.275v 50%
13.275v / 3.318 75%
13.6v / 3.40v 100% <---- gen1's typically rest at 3.38v

Anything *resting* over 3.4v after many hours is some bad overcharge! Discharge immediately.



Kinda worries me about float - ideally if you can't disable it, then the float voltage should be so low that no current actually flows.

For me, that is a bit too conservative at 13.6v while under charge. That might be too slow in a solar application - it all depends on you. Even though I run at 14v, with a .05C absorb, if I were to go totally conservative, then perhaps 13.8v or maybe nudge it to 13.7v. You'll have to see what level you are comfortable sitting at taking into account solar time issues.

It's fun fine-tuning LFP to fit one's own needs.

PNjunction,

You are correct that "it's fine-tuning LFP to fit one's needs". You've been pretty up front about your charging values and why. My own questions fall into the "what if scenario", in this case, what if I went just a tad lower, etc? Somewhere within all the data supplied by all the folks willing to share should give us a reasonable set of guidelines that yield performance and longevity that is repeatable. Looking back to the 2011 literature I have on the 20ah cells now reveals that the parameters were not optimal - hence changes we see today. Combine that with Sunking's statement that we are in many ways dealing with an immature technology from manufacturers that have changed their chemistry formulations several times in a relatively short time-frame and this leaves me to try to err on the side of conservative/caution this time (even though those little 20 cells are still surviving just fine right now.)

To your point about disabling float - yes, I can set it lower so essentially there is none but at this point in the season I may be wishing to take advantage of any available solar input. I may be incorrect to do this float setting for now but I do want to track any impact this has in relation to heat with these cells - if there is any measurable amount. I'm still not having luck sifting through studies looking for answers to a couple of questions I feel are important to this new setup. One being heat dissipation in the GBS 200ah cell with opposing terminal ends. I understand about activity dispersal uniformly through the cell but I'd like to find any type of data related to heat with this design. I was reading about distances traveled in LFP cells and as I look at these 200ah cells there appears to be some real estate from one end to the other. Whether that translates to more real estate between the anode and cathode is not something I can easily find either.

One thing that concerned me (that I believe you mentioned some posts ago) with the 14v and .05C absorb was the scenario of not enough solar input during the day to reach that point - or even if that was the case for days - never reaching that set-point, but instead just hovering near it for hours and hours. This would (in my opinion) be less desirable than the float setting I'm using (mentioned above) but I could be wrong, or this could be splitting hairs.. To potentially address this issue - maybe the even more conservative approach I'm taking (albeit with the float setting you're not 100% in love with) may be a good compromise. Time will tell and your posts are great at sparking some thinking - a bit of a challenge. Let's continue to exchange ideas and outcomes

~CrazyJerry

[CrazyJerry]

Jerry, I was looking at your EMS system. I was wondering how the two statements below work together when your EMS system uses cell balancing ?? Does the system do both top and bottom balancing ?? Is the " New Pack " your older cells ??

The functions of the Elite system.

Hi Willy T,
The two statements below should have been clarified on my end a bit more. So, the initial install of this particular pack followed the bottom balance approach first. This is not a function of the EMS - it is something I initiated by all the very convincing posts I read on this forum. From a common sense point of view in a "sub-C" environment (as PNjunction has labeled it), it makes good sense as a starting point.

The EMS is setup to do a top balance at ~3.55 per cell. The pdf for this EMS can be viewed online here:

http://elitepowersolutions.com/docs/User%20Manual%20for%20EMS%20System.pdf

It does a decent job explaining what it does and more or less how it does it. (See pages 17 & 18)

~CrazyJerry

[karrak]

The 12 volt pack was charged to 13.6 and then held there until the end amps setting (.05c, 10 amps in this case) was reached. At that point, the pack entered the "float" stage at 13.5volts. Once at float, a small current of 1.9amps(charging) is reflected on the screen. Upon initial install, the pack/brain/ needs to be charged to 14.2v to start at the full 100%.

Your charge regime fits pretty closely with a number of others who post on the Energy Matters forum. Only difference from my charge regime is that I charge at 3.4 volts/cell (=13.6/4) with an end amp setting of .02C and float at a lower voltage of 3.3 volts/cell (=13.2/4) which basically terminates the charge when the current reaches .02C@3.4 volts/cell. I used to be of the opinion that keeping LFP batteries at a float voltage of around 3.35 volts/cell (=13.4/4) was bad for them, but now think this is probably not the case. Others on the Energy Matters forum, have found that if you do float them at this level after charging at 3.4-3.45 volts/cell, assuming enough sun, you end up with a fully charged battery at the end of the day which is the perfect scenario.

Sun permitting I charge my battery up to ~3.47 volts/cell (=13.9/4) once a week to reset the SOC counter. I would be interested to know whether your system does something like this.

As at two days ago charging my battery to 3.47 volts/cell gave a voltage deviation between the cells of 0.081 volts (3.528-3.447). In March of this year the battery was balanced and that deviation would have been less than 0.010 volts. So my battery is going out of balance. Maybe newer LFP batteries or more expensive ones from three years ago would stay more in balance. Ideally, one would hope that at some point in the future, that all LFP cells from the same manufacturer with the same capacity would remain perfectly matched in capacity and coulomb/current efficiency regardless of the operating environment during their entire life, so we don't have to take precautions against them going out of balance.

Simon

[PNjunction]

PNjunction, the display is handy. One of my questions with it, and the likes of, is how accurate is it? Or is it accurate enough for basic needs thus making it useful.

Ironically, I only use resting voltages (12 hours!) as a relative state of SOC, but even then, your metering needs to be quite accurate as the spread between full and dead is only 1 volt! Shirt pocket meters are out. The Optimate TS-126 tester lines up with my Fluke 87, so I felt better. This is what I use:

12.7v / 3.175v dead
12.8v / 3.20v 20% <-- I try to never see a resting voltage below this
12.95v / 3.237v 25%
13.1v / 3.275v 50%
13.275v / 3.318 75%
13.6v / 3.40v 100% <---- gen1's typically rest at 3.38v

Anything *resting* over 3.4v after many hours is some bad overcharge! Discharge immediately.

The 12 volt pack was charged to 13.6 and then held there until the end amps setting (.05c, 10 amps in this case) was reached. At that point, the pack entered the "float" stage at 13.5volts. Once at float, a small current of 1.9amps(charging) is reflected on the screen. Upon initial install, the pack/brain/ needs to be charged to 14.2v to start at the full 100%.

Kinda worries me about float - ideally if you can't disable it, then the float voltage should be so low that no current actually flows.

For me, that is a bit too conservative at 13.6v while under charge. That might be too slow in a solar application - it all depends on you. Even though I run at 14v, with a .05C absorb, if I were to go totally conservative, then perhaps 13.8v or maybe nudge it to 13.7v. You'll have to see what level you are comfortable sitting at taking into account solar time issues.

It's fun fine-tuning LFP to fit one's own needs.

[Willy T]

Jerry, I was looking at your EMS system. I was wondering how the two statements below work together when your EMS system uses cell balancing ?? Does the system do both top and bottom balancing ?? Is the " New Pack " your older cells ??

The new pack that runs the other side of the house is bottom balance (again thanks to this forum).

Upon initial install, the pack/brain/ needs to be charged to 14.2v to start at the full 100%.

The functions of the Elite system.

The functions of an Energy Management Systems include voltage and temperature monitoring of individual cells, pack current measurement, state of charge (SOC) calculation, alarm interlocks to prevent over-charging and over-discharging, cell balancing, ground fault detection, video output, CAN communication, etc.

[CrazyJerry]

Pretty cool - although I think what I'm seeing are *rested* voltages, not while under charge. Most GBS cells, especially Gen1's rest (4 hours or more) at 3.38 / 3.39 when considered 100% fully charged. So that meter is relatively accurate.

I'm using something similar, but just a bit smaller:

http://tecmate.com/products/ts-126/

Tested against a Fluke, and a bit more convenient than a chart taped on the wall. It isn't lab-quality, but close enough for daily checks and so forth. Note that it too draws a bit of current, so I don't leave it attached.

Although rested voltages are not always the most convenient way to deal with things. But I do use it from time to time to judge the effectiveness of various voltage and current tests with lifepo4.

Example tests of charging (>.05C current, .1C or more typical) and later evaluation of rested voltages:

1) Charge to 3.5v and stop.
2) Charge to 3.6v and stop.
3) Charge to 3.5v and allow absorb current to drop to .05C.
4) Charge to 3.6v and allow absorb current to drop to .05C.

It was easy to find that options 3 and 4 both result in a "fully" charged cell, but option 3 merely takes longer.

PNjunction, the display is handy. One of my questions with it, and the likes of, is how accurate is it? Or is it accurate enough for basic needs thus making it useful.
-
You did list a few charging scenarios above and this pic lines up with #4. The 12 volt pack was charged to 13.6 and then held there until the end amps setting (.05c, 10 amps in this case) was reached. At that point, the pack entered the "float" stage at 13.5volts. Once at float, a small current of 1.9amps(charging) is reflected on the screen. Upon initial install, the pack/brain/ needs to be charged to 14.2v to start at the full 100%.
-
The only thing I noticed was the lcd screen's viewing angle didn't jive with my (aging) eyesight. I did notice when viewing it from just below the horizon it was better. No menu settings helped, but turning the lcd upside down did - and there's a menu adjustment to make the visual right-side-up.
-
This system has a secondary screen that shows the individual cell voltage and temperature. Haven't gotten that far to check the accuracy but it's on the list.
-
Thanks for the charge scenarios. You and others have mentioned these over the many posts. I'll pan on going through them just to observe but am slightly biased to #3. That may be some of the residual from 11 years of lead... Have a good day!
-
~CrazyJerry

[PNjunction]

Pretty cool - although I think what I'm seeing are *rested* voltages, not while under charge. Most GBS cells, especially Gen1's rest (4 hours or more) at 3.38 / 3.39 when considered 100% fully charged. So that meter is relatively accurate.

I'm using something similar, but just a bit smaller:

http://tecmate.com/products/ts-126/

Tested against a Fluke, and a bit more convenient than a chart taped on the wall. It isn't lab-quality, but close enough for daily checks and so forth. Note that it too draws a bit of current, so I don't leave it attached.

Although rested voltages are not always the most convenient way to deal with things. But I do use it from time to time to judge the effectiveness of various voltage and current tests with lifepo4.

Example tests of charging (>.05C current, .1C or more typical) and later evaluation of rested voltages:

1) Charge to 3.5v and stop.
2) Charge to 3.6v and stop.
3) Charge to 3.5v and allow absorb current to drop to .05C.
4) Charge to 3.6v and allow absorb current to drop to .05C.

It was easy to find that options 3 and 4 both result in a "fully" charged cell, but option 3 merely takes longer.

[CrazyJerry]

When running conservatively, such as only reaching 3.5v at the high end, and just dipping under 3.2, say 3.19v at the low, this seems to dovetail nicely into what I plan for when doing my power budget based on the 20% nameplate de-rating.

PNjunction,
So today in looking at the new pack monitoring (this place is now starting to look like (Frankenstein's lab!) I thought you might get a kick out of this. Very interesting that at just 13.46 volts (3.36-3.37 per cell) matches up with 95% capacity. They may even be a little more conservative than you - lol! (Take with a grain of salt)
-

-
~CrazyJerry

[CrazyJerry]

The CellLog8 is a great little device but it does have its issues. The main one being that unless it is modified it draws its power from only the bottom six cells so will unbalance your battery. I have fixed this by powering it via an external switch mode power supply. There are other ways to fix this problem and other information on the CellLog8 can be found here Cellog 8 hacking. Let me know if you want more information about my fix.

It can be calibrated and I have found that it seems to hold its calibration.

Simon

Simon, I remember that post (actually right where you referenced) and I left it on the to-do list someday mainly because the celllog8's are only powered on when I check the cells. When I was using the velomobile there wasn't room for anything else so pack monitoring at a pack level was it. Now, the home unit normally just sits on a shelf all rolled up.

When it is hooked up to the repurposed cells that are now part of the house bank it no longer has to monitor 8 cells per celllog meter (there are two separate meters in the box). The setup now is that each meter only "sees" 4 cells. This is a result of how they are all wired up (4p4S)... Basically the first group (or 16 cells in series @ 48V) that I purchased is still together as a happy family but now in it's own 12volt pack. The second group is also wired identically. So each group has it's own celllog8 that's now only seeing 4 cells each. I used the flat trailer whips plus a separate ground wire for quick connect/disconnect capability. I don't watch much tv.
~CrazyJerry

[CrazyJerry]

Well Jery you have discovered Lithium cells do not drift and there is no need to for them. In fact the Balance Boards are the root of the problems. Most notable destruction. If you want to eliminate the possibility of over discharging the cells, you have acquired the skills and knowledge to Bottom Balance the cells making it impossible to over discharge them and drive them into reverse polarity.

Thanks SunKing. Some time ago when I discovered this forum I was initially quite entertained by your label of "vampire boards" however, it is a great way to remember what they do well. Currently, they are in a small box not far from my antiquated Frontier bag phone.
The new pack that runs the other side of the house is bottom balance (again thanks to this forum).
-
~CrazyJerry

[karrak]

With the string of 16 cells drifting (actually 32 cells since both packs were drifting), I resorted to charging individual 12 volt modules (like you were mentioning.) I reconfigured the solar panel wiring and could see approx 42 amps or so. Charging by 12 volt module was pretty time consuming until I found the the hobby market / Celllog8. After building the monitor, I went back to charging the entire 48volt pack. I could then give an individual cell a jolt or a discharge through the corresponding speaker jacks in an attempt to get the cells back to a closer relationship.

The CellLog8 is a great little device but it does have its issues. The main one being that unless it is modified it draws its power from only the bottom six cells so will unbalance your battery. I have fixed this by powering it via an external switch mode power supply. There are other ways to fix this problem and other information on the CellLog8 can be found here Cellog 8 hacking. Let me know if you want more information about my fix.

It can be calibrated and I have found that it seems to hold its calibration.

Simon

[PNjunction]

The only thing I've seen about the 200ah cells is that the positive and negative are on opposite ends to try and help current distribution and avoid hot-spot crowding at the terminals.

This may not be anything noticeable in a "sub-c" solar housebank environment.

Re the Cellog - do power it from a *separate* power source, and avoid the temptation to tap off the battery itself. Again, another point of failure, de-balancer. I would verify the accuracy of each channel with a separate reading from a Fluke or otherwise comparing terminal voltages to the cellog 8 display.

I know T1 Terry and others have used them, but keep an eye on it and do a pm comparison once in awhile.

[CrazyJerry]

Thanks Jerry for all the information, there is allot to digest. All very interesting and informative.

I might have missed some information and am unclear about some details of your charging procedure and temperature measurements. I have a few questions.

Did you charge the whole 48 volts battery or split it up into its twelve volt battery modules and charge each one individually?

Were your temperature measurements done using your solar charger or the 6 amp 110volt charger?

Would I be right in calculating the charge current from the solar charger to be around 11 amps(0.6C)

Thanks
Simon

You're welcome Simon and I hope it may be of use to someone, or at least to help paint a bigger picture.
-
Bouncing back-n-forth a bit I can see where I could've cleaned up the post a bit.
Initially I charged the 48 volt pack as a whole with the supplied GBS listed 48v 6amp charger. It would bring the pack up to an advertised 58.2 volts. When using that charger on the second 48 volt pack (purchased about a month later) I noticed the balance boards came on way sooner (3.55 volt versions). After digging around on their website I noticed the spec sheets on the batteries and balance boards had changed. So, that's when I began to pay attention to the heat build up from charging with the supplied charger. At the end of the charging cycle with the supplied charger, ALL 16 of the 3.7volt balance boards would be lit (red) and would stay on for some time after. If those board turn on voltages were accurate (3.7 volts) then the charger was putting out more than the advertised 58.2 volts me thought. The only other way I could charge the 48 volt pack was through the solar/mppt so I hooked up the charge station. You are right in the ballpark with your reference to 11amps (0.6C) from the solar. Depending on sunlight/conditions/time of day it would be lower at times. I had 101pv in and the output to battery set at 58volts(approx same as GBS charger). After continuing to observe heat from both saddlebags, I lowered the figure to 56volts and then downward to 54volts.
-
Going beyond 11 amps into the pack was pretty easy. I could use both the solar and the supplied charger simultaneously to charge the pack, and just shutoff the supplied charger when things got close and let the solar finish it off. Right or wrong, this is what I did and the cells continued to survive.
-
With the string of 16 cells drifting (actually 32 cells since both packs were drifting), I resorted to charging individual 12 volt modules (like you were mentioning.) I reconfigured the solar panel wiring and could see approx 42 amps or so. Charging by 12 volt module was pretty time consuming until I found the the hobby market / Celllog8. After building the monitor, I went back to charging the entire 48volt pack. I could then give an individual cell a jolt or a discharge through the corresponding speaker jacks in an attempt to get the cells back to a closer relationship.
-
In lowering the upper charge voltage limit I did notice the lowering of the heat at the terminals so that's what I ended up staying with. The added benefit was "less" cell drift. Reducing upper charge voltage further like PNjunction and others have suggested has settled these cells of drifting although I do check them about once a month.
-
So, that's a lengthy post to your question, and in summary, I charged as a 48volt pack as well as by 12volt module. I used the GBS charger and I also charged directly from the solar panels through the MX60.
-
~CrazyJerry

[Sunking]

Every week I would check the paralleled cells for cell voltage drift and there isn’t any. This revelation is what made me question the use of ALL the balance boards (remember - half of them are of a differing value) so they were all removed. I check the cells about once a month and they’re fine with no adjustments needed.

Well Jery you have discovered Lithium cells do not drift and there is no need to for them. In fact the Balance Boards are the root of the problems. Most notable destruction. If you want to eliminate the possibility of over discharging the cells, you have acquired the skills and knowledge to Bottom Balance the cells making it impossible to over discharge them and drive them into reverse polarity.

[CrazyJerry]

Jerry - sounds like you are on the ball, although I read your posts with some trepidation when I saw the E-bike application. Many of those cells in that arena are used, abused, and murdered before their time.

Now that you are running with conservative values, you may want to do as I do when planning your next setup - purposely de-rate the ah specification by 20%. That is, my 20ah battery is really only 18ah, and the 40ah batt only 32ah.

When running conservatively, such as only reaching 3.5v at the high end, and just dipping under 3.2, say 3.19v at the low, this seems to dovetail nicely into what I plan for when doing my power budget based on the 20% nameplate de-rating.

Like you, since I'm running in the so-called "sub-c" environment, there is plenty of time to manually check cells when dropping just below 3.2v for any rogue cells out there - but I certainly wouldn't want to run with as many cells as you are, but understand the repurposing of your original e-bike setup cells. And of course hoping that you have an LVD in use just in case.

If you do go with new cells of larger capacity, but still need to parallel some, remember that with LFP, you build up your capacity first with parallel sets, and then series connect these "sets" or groups of paralleled cells to obtain your target voltage last.

So that's three of us now running GBS - you, me, and Sunking. I only picked GBS for the safety reasons of having the individual cell-terminal covers come with them, as opposed to making your own top-covers with say Winston or CALB.

PNjunction,
In the beginning I was unaware of this forum and its only been recently I discovered it. Before registering I've read a ton on here, and a couple other forums too. It's amazing how the comments in many cases, reflect different things I've also experienced along the way with these cells. What I never sat down and charted, was the upper and lower "knees" although I witnessed the sharp rise and drop-off on a meter. Going by the battery spec sheet, I was none the wiser but was suspicious due to heat build up and cells drifting... The knee graph speaks volumes when you know what you're looking at.
-
Moving forward, yes, I already have the second bank installed and it follows both the AH derating as well as the very conservative charge/discharge points. At the risk of inflaming some folks on here I do believe you, SunKing and a few others bring quite a lot to the table to include differing views. And, as evidenced by the 2011 20ah spec sheet, this has also been reduced to more conservative figures... Coincidence?
-
The Surrette replacements are the 200ah GBS cells. I do like the terminal cover design, but also wanted to try their GenIII recipe. I like the idea of the terminal on both ends of the cells but am having little luck downloading and reading through some of the mega studies where this has definitive benefits.
-
This is an interesting experiment and will share anything - good or bad - along the way.
-
~CrazyJerry