Discussions about LiFePO4 cells and Battery University

Are these pouch cells or cylindrical cells? If they're pouches, are they compressed in the Enginer setup?

Either way, my understanding is that plating comes well before vapor can form, though it can also come together with electolyte breakdown. Electrolyte breakdown doesn't start in our cells until about 3.9 Volts or so, and there it takes significant charge current. Very high temps (above 60 degrees C) can also lead to electrolyte breakdown. If you've got vapor buildup, it should be very obvious when looking at the cell (provided you can actually see the cell). Pouches swell up like balloons (or spring a leak) if not compressed, cylindrical cells will bulge at the ends until their safety vent cracks open. It's hard for me to imagine the issues you've had being the result of that unless these are uncompressed pouch cells.

Again, if you do a nice slow manual balance at 3.6V and take a few measurements with the cells at 100% SOC, you can rule many things in or out.

 

Agreed on all counts. Interesting storage scheme. If you look at the website for the Rimac Concept One (my favorite electric supercar at the moment), you can go to the page on the battery system and see how they're using springs inside a steel box to apply pressure to their pouch cells (also LiFePo). As you probably already know, A123 puts bands around their modules to accomplish this.

One other thing that occurred to me in this conversation--your max charge rate of 15A is well below 1C for your pack. I would think charging all the way to 3.8 with such low (relative to C-rate) current would be far worse for Li plating than charging at, say 2C, until one cell hits 3.8V, then pulling current back to 1 or 1.5 C. Then wait to hit 3.8 (or 3.7, or whatever) again, drop current some more, and so on in 3 or 4 stages. Li plates as a result of too much total energy being pushed into the cell. Because "terminal Voltage" (not actual "cell Voltage", which is what we see once the OCV has settled to rest) is pushed up more and more over "cell Voltage" with greater charge current, it's actually safer to hit 3.8V briefly at a very high charge current than it is to slowly creep up on 3.8 with a low charge current where it is very likely that a greater amount of total (excess) energy has been put into the cell. Hopefully that's clear.

 

It's not so much that there's a need to "hit it hard", but when you get into the higher Voltage range, the speed at which you got there is critical in finding what the safe max terminal Voltage actually is. I think we're just saying the same thing different ways...

The wall wart setup you describe is probably a very good cost-effective way to balance your pack periodically. You could charge at your 15Amps until a cell hits 3.6 or 3.65 then get this setup going to finish things off. Charging at the cell group level works well, but is difficult and expensive to implement in most real-world setups. At the least, it would afford you a relatively convenient way to top off the laggers like cell 11 in the Real Force pack.

 

Dan2l

You probably know that these 16 charger must then be insulated, that's to say with a transformer inside.

 

Taking W9BK's terminal volts opposed to cell volts at <1C charging current theory then having a HVC of 3.6 would be a good idea.
You could just simply just set your HVC to 3.60 volts. Start burning excess charge at about 3.45 volts. This is what the miniBMS does if you get the modules with a HVC of 3.6 volts. This would be better with a charger with a end voltage of about 55.3 volts (rather than 58.4 volts so balancing is mostly just done during CC mode (when balanced). The miniBMS burns a max of 0.75 amps on each cell. The burning current varies between the cell values of 3.45 volts and 3.6 volts and I could not find a graph "cell volts vs burning current" for it, so it could be near linear.

I also noticed that the LVC on the minBMS is temperature compensated. 2.5v at 0deg.C, 2.6v at 25 deg.c and 2.7v at 50deg.C. Why have temperature compensation. Would it avoid Li plating during discharge?

Dimitri who is the creator of the miniBMS informed me that I should use minimum AWG16 (1.5mm2) wires and make them as short as possible for the cell volts sensor wires. He also said that this is a drawback of the centralised system and this is not required in a distributed system. He did not explain why. I imagine that the wires can pick up induced EMF affecting the accuracy of the cell volts measurement or the measurement current was so great that the wires resistance could cause enough volts drop to make the measurement inaccurate in a centralised system. Where as the distributed system the miniBMS module is placed directly across the cell and the wire is very short. I know that the cell volts 9pin plug/loom cell volts wiring on the RFE packs are less than AWG16 and this may be a problem. In my PHEV kit I twisted the wires in pairs to try to avoid picking up noise. Noise from the Enginer DC converter could affect cell volts measurements at the time of LVC.

The miniBMS instructions recommends that the battery pack be manually balanced before using the miniBMS which some users have overlooked. The RFE battery pack can be so out of balance (caused by the BMS16D) that it could take the miniBMS about four months to balance. Once the pack has been manually balanced the miniBMS keeps it balanced long term. In the unlikely event of a HVC the BMS needs to isolate the charger output (not input) so that the charger will latch and not do another charge and keep cycling all night on the HVC.

The cost would be 16 cells x $10 = about $160 for the Walwart setup.

I would rather pay the extra money for a miniBMS and not have to periodically balance my cells manually.

Whether the miniBMS approach works I do not know as I have not collected any data since I started using this method so I cannot say for sure that this approach would prevent Li plating during charging. I use this approach with my two Emginer kits which both use RFE battery packs.

 

I have a question for the battery expert:

I have read somewhere that pouch cells go soft when completely discharged. They also go soft if they lose their vacuum. Is it just the vacuum that makes the cell stiff?

 

If by "go soft" you mean they become limp and flexible, then neither of these is true for A123 cells. I don't think there's much of a vaccuum pulled in these cells to begin with. Pouches that have been allowed to swell up (by whatever mechanism) will appear to be less stiff than a good cell, but this has nothing to do with vaccuum, but a pushing apart of the many layers of cathode and anode inside the cell. It is the the tight stacking of the electrodes and separator layers that give the cell the bulk of its stiffness. Absent swelling, SOC should have no effect on this whatsoever.

 

I see two possibilities here--balance load resistance is constant, and duty cycle is 100%, which would mean current moves with Voltage in a more or less linear fashion. Some of the more intelligent systems out there control the balance load duty cycle with a PWM-type scheme. In the latter case, duty cycle (and thus balancer current) awould most likely be determined by the delta between the min and max cell Voltages.

This does not make sense to me. If anything, I would expect the BMS to lessen the depth of allowable discharge as temperature goes down, not up. I know of no mechanism that causes plating during discharge--only charging. Those would be charging at too high of a cell Voltage, or charging at too high of a C-rate when the cells are very cold, AND at too low an SOC. The C-rates we're talking about would not be a problem above -10 degrees C or so.

Noise could be a problem, but I would think stray inductance would be a bigger problem. Perhaps the balancers are located within the BMS? This would mean balance current would be pulled thorugh the sense lines, leading to significant enough current through the sense lines that the load presented by the balancers could cause a Voltage drop across the sense lines and wreak havoc on good monitoring of cell Voltage. Noise can generally be dealt with through shielding and careful routing of wiring. I don't know the layout of the mini-BMS system, so I could well be missing something here.

ALL Li battery packs should be well balanced BEFORE being put into service. Not doing so is to beg for problems. The very long balance time by the BMS is just one issue. "Cycling on the HVC" is not necessarily a bad thing though--done right, it speeds the balancing process considerably.

I think you said above that the mini-BMS is ~$300. If so, and--more importantly--if it works well, then I would agree the mini-BMS would be the better buy.

 

Thanks.

I just happened to "accidently" overcharge a A123 20AH pouch cell lately. It did not blow up like a balloon possibly because the cell volts only went as high as 4.00 volts for a long period of time and went soft and it was also compressed by two aluminium sheet plates.

1. It did not swell up like a balloon but like you said it swelled up just enough to seperate many layers of cathode and anode.

2. I am not sure but maybe the two compression plates may have helped in some way. Maybe it just avoided the cell's bag from splitting.

3. I assume it did not split as I could not smell any electrolyte. I read that the electrolyte smells "sweet".

4. The cell developed a very high internal resistance which regardless of SoC the LVC or HVC would happen at about -2C and 2C respectively.

5. I noticed that liquid (electrolyte) had seperated from the many layers of anode and cathode.

Are these the characteristics you would expect to find for an overcharged A123 pouch cell?

Is the high internal resistance the result of Li plating?

I am contemplating doing a autopsy in a well ventilated workshop. Wear the required PPE i.e. rubber gloves, dust mask and safety glasses. Take photos and post them on this thread.

 

Ah, ritual sacrifice followed by a gory post-mortem....you'll get better results if you paint pentagrams around the lab and drip the blood of a freshly killed rooster all over the floor. Bear in mind that specific results of overcharge/overdischarge can vary significantly with charge/discharge current, absolute Voltages reached, thermal considerations (which are profoundly affected by "compression plates" that also act as heat sinks), and normal cell variability.

Most of the electrolyte in a cell is soaked into the active material on the plates. Gas buildup in a cell comes from electrolyte being chemically broken down, so it stands to reason that this will quickly affect spacing of the electrodes throughout the cell. What C rate did you overcharge at? What did the cell finally settle to? Sounds like you didn't abuse it TOO bad.

How much pressure was applied by the plates? Did they get warm during overcharge? The heat sinking they provide may have done a lot to keep swelling in check.

I guess you could say it has a sweet element to it, but it's quite acrid as well--not a smell I particularly like. Once you smell it, you'll recognize it forever. You'll get a real good whiff once you cut in.

I think you mean you can have the cell just about dead, apply a 2C charge and you will immediately hit HVC--is that right? If so, yeah, you've made a nice crispy piece of toast there.

Do you mean you have liquid sloshing around inside the cell? Not unusual for an abused cell.

They are among the possibilities, no real surprises.

Probably a main contributor. Impedance rises because, among other things, cyclable Li in the cell has to take a much more tortuous path to get to where it can intercalate into the remaining usable structure of the active material on the plates.

Sounds good. Safest to discharge the cell to a fairly low SOC before you start cutting, but you can tear them down with a lot of charge on them without anything bad happening. A healthy cell will even continue to perform better than you might imagine for quite some time after being cut open. Make sure the workspace is well lit and very clean. You don't want any debris or crud getting in there and reacting in unpredictable ways. Other than PPE and a camera, a hobby knife and a pair of tweezers are about all the tools you'll need. If you have some kind of fume extraction system available to you, use it. A ventless system with activated carbon filters can also quite effectively capture electrolyte fumes. Even a benchtop fume extractor for soldering with a carbon pad filter is better than nothing.

 

Yeap, I have a fume extraction fan in the welding bay. I will probably use it.

 

Hi wb9k,

The Enginer charger has a CC current of 15 amps. So each of the four buddy cells are charged at a rate of 15\4= 3.75 amps i.e. 3.75/20=0.8175C. So at a charge of 0.8175C what should the BMS HVC be set to?

Using a charger CC mode of 0.8175C, is a BMS HVC of 3.8v too high if the HVC only occured once during each charge cycle?

If the charger kept cycling all night on the HVC of 3.80v with a CC of 0.8175C, would you expect the RFE cell to be overcharged?

Is there some sort of formula to determine the approprioate HVC for a BMS or is there a thumb of rule or a look-up table or you must go to the RFE specifications sheet.



I have two RFE 20AH specification sheets in attached PDF files below and I do not know which one or if both are being used by Enginer.

Maybe you could help explain the graph. The blue line represents the charging voltage or cell voltage.
The black line represents AH capacity of the cell. The red line represents the charging current of the cell.

According to the graph the cell reaches full capacity at about 3.5 volts so why bother going any further unless you are going to top balance.

The only function of CV mode that I can see is that it allows top balancing. Is there another reason? I cannot see any point in further charging just to top-up the cell if you can run a risk of overcharging a cell. The small gain in AH during CV mode is small and is not really necessary unless you are balancing at the top end.

There is no mention of an appropriate HVC setting for a BMS to provide protection for the RFE cell. So the graph above shows the values of a balanced cell and the charger cutoff voltage should be 3.65(+or- 0.05v). Does this mean the BMS cutoff voltage should not exceed 3.65(+or-0.05v).

Can you explain what is "0.5C5mA".

Do the RFE spec sheets support your quoted statement above.

Thanks.

 

I think you transposed some numbers here. The C rate is .1875, not .8175. Big difference. You could simplify your math to get to the same answer: 15 Amps/80Ah=.1875C. At this C rate--and based on recent experience monitoring 12V packs with a Cell Log 8s through a couple cycles--I would put the HVC pretty low, 3.62 to 3.65.

Even at .8175C, 3.8 may be just a bit too high. I would not go that high unless charging at at least 1C. I'd go around 3.7, but I have no hard data to support that. At under C/5 (.2C), 3.8V is too high, IMO.

Yes, definitely--and at virtually any C-rate.

For this type of detail, I would always refer to the manufacturer specs, no matter what type of battery we are talking about. They're all a little different, reliance of "rules of thumb" is unlikely to get you to peak performance for any given cell.

Almost--that black line isn't quite flat at any point, but you are correct that there is very little capacity to gain by charging any higher. At 3.5V (rested), you are at 95% SOC or more.The reason you want to do the CV thing at the end is to keep the pack balanced. It is better to keep a pack balanced by top-balancing at every charge than it is to wait for things to get way out of whack and then fix the problem. In DIY systems like we're discussing here, that's perhaps a luxury, but for a normal setup for everyday users, it's a must. Balancing systems work fairly slowly. If they are allowed to work on a very regular basis, this is OK. If we have no balancing except when things get out of balance far enough to notice, fixing the problem becomes time consuming and/or labor intensive (if manually balancing.)

That would be my take-away.

I have never seen this nomenclature before. My guess would be that it's the same as "0.5C +/- 5mA".

I believe they do, but only indirectly. Look at the chart at the upper right of the second page of the RFE-20-10150225 sheet which shows Voltage curves for various charge rates. Note that the lower the C rate, the longer the battery has to CV at 3.65V. At 1C, 3.65 is hit just a minute or two before 60 minutes has gone by. At .5C, the time 3.65V is his about 10 minutes before two hours has gone by. At .2C, 3.65V is reached over 1/2 hour before 300 minutes has gone by--the amount of time required for a full charge (.2C*5=1C or 60minutes*5=300 minutes). This implies what you can prove to yourself in other ways--that max terminal Voltage is a function of BOTH C rate and SOC. Having said that, the data sheet as presented does not really make this clear, soooo...

Here's another experiment you can try to illustrate the point. Take a cell and charge it until OCV hits (for example) 3.7 Volts at 1C. Let the cell rest for 30 minutes and record the rested Voltage. Discharge the cell to 3.4 or so (rested) and charge again to 3.7 Volts, this time at .5C. Let cell rest and record Voltage again. Discharge again and charge to 3.7V at .2C. Let cell rest and record Voltage again. You should see that at lower charge rates, settled Voltage will be closer and closer to HVC. In other words, the lower the C-rate, the closer you are to true 100% SOC at HVC.

So why does RFE recommend a cutoff of 3.65 no matter what the charge rate? Probably for simplicity's sake and as a CYA against too-lax interpretation of the spec, but that's just a suspicion on my part. I could well be wrong. Hope that is of some help.

 

This may sound rude from the get go. However I want to step into this conversion a little.
First, I just purchase so some BVM cell logs (one tool/measure) of a in this case RFE pack.
I will also use BMS and multi meter to check some measure aka of voltage/amps

Next
Balancing
Top/ Bottom
Top balance of a pack is a method of balance(not the only one) which has a very recent poor history.
Just ask Boeing dreamliner engineers. Also the Arizona company that built/design Boeing battery pack reduce a company mfg plant to desert/ash while in operation. Both Fisker and A123 have had at least one (thermal runaway)
if not more that have gone public. Thermal runaway aka fire on all of the above. Maybe this just my thinking but top balance is a science with a jaded learning curve which is still with some questions????

 

The battery in the Boeing jetliner is not LFP, but a chemistry with Cobalt. Thermal runaway is primarily associated with Li chemistries that include Manganese or Cobalt. LFP is MUCH more stable because the Oxygen in the mix is much more tightly bound chemically, thus is cannot be easily stripped away to become fuel for a fire. My understanding at present is that the battery was overcharged, THE cardinal sin with all Li chemistries. Shame on any engineer who builds a system where this is possible. You can read the latest details on the investigation at the NTSB website, which is surprisingly clean and easy to navigate for a government website.
http://www.ntsb.gov/investigations/2013/boeing_787/boeing_787.html
Included are pictures of the burnt battery next to samples of good ones. The pack looks like something built by an amateur in a garage--many of us here are astounded that such a product could ever get into any airplane, let alone a commercial jetliner. I would further question the wisdom of using any Li chemistry with Co or Mn in a plane. Energy storage is always dangerous, no matter what the medium and fire is ALWAYS a possibility. LA batteries can explode during a jump start, yet they are everywhere. People have learned how to deal with the risks and manage them properly. Apparently, there are some who need to learn some of this the hard way with Li-ion. FWIW, I personally feel A123's cells are far and away the safest mode of energy storage (per energy density) that I have ever worked with. I work with these things every day without fear--albeit cautiously.

I think you do not understand the causes of the Fisker fires, which had to do with an overheating cooling fan--the Li battery had nothing at all to do with it. I don't know of a single incident with our cells where "thermal runaway" was induced by overcharging or top balancing. We have seen incidents of failure, but not like the one in the 787.

It's really a shame that all Li-ion chemistries have been tainted with this reputation for spontaneously bursting into flame. LFP has to be abused quite extremely to be a risk here. You can make an A123 battery flame out or even explode, but I've never seen it done without being above 4 Volts while charging at at least 4C. This is rougly the same risk as with LA. So build a system where that's not possible. This isn't rocket science. Top balancing is generally preferred because it allows by far the most simple, direct, and reliable way to keep a pack properly managed. If you're worried about cell damage or fire, bring down the HVC to around 3.6 and place hard limits on charge current above ~80% SOC and just move on.