Lithium Battery Reports & Tests' 9/08

BMI said:
Ok now for the answers to the above questions-
PSI manufacture and sell the original/first generation 40138 cell (the old green ones)
BMI and Lifebatt sell the latest generation grey and red cells which are an improvement on the PSI cells. The BMI and Lifebatt cells use patented Phostech LiFePO4 powder which is manufactured in Germany and shipped to Taiwan for manufacture into 40138 cells.
So to put it simply once and for all BMI and Lifebatt cells are the same and come off the same production line while PSI cells are a different cell and use non Phostech LiFePO4 powder in their manufacture.

Nothing personal to anyone at PSI, BMI, or LiFeBatt. But my data collection and analysis suggests the existence of data inconsistencies...

1. BMI and LiFeBatt have made great use of a mis-representation that green cells are old. BMI continues to do so in their ebay ads, website text, and in written articles. Not only are green cells old, but they could very well be cheap Chinese copies made with inferior components.

Here's the problem: Green cells are current production cells from PSI - the only company that makes 40138 LiFePO4 cells with this terminal construction. Green is their color, as grey and red are used by BMI/LiFeBatt. PSI produces cells with and without Phostech material. Their standard green cells does NOT use Phostech. I know that PSI manufactured LiFeBatt and BMI cells under contract as recently as fall quarter 2008.

One demonstration of that fact is this: Visit Armin's product page, scroll down 1/2 way, and look at the photos of his grey cells. Note the serial number barcode and sequence: 4BF1M305-0593. Compare it to one of the cells I received from PSI in Dec 08:

View attachment IMG_1557.JPG

IF these cells were produced by BMI and/or LiFeBatt in another facility, why did they duplicate the PSI barcode, the PSI serial number sequence, and then choose to begin numbering their 'new' cells 'within' the PSI serial number sequence?

[edit] Could 'old' mean 'first generation'? It's obviously not 'old' with regard to time. Since the cells share the same cannister, manufacturing crew and (presumably) processes, this could leave 'old' to mean that a Phostech forumula, or other formula change, makes 'next generation' cells. Could be. But since it's the same factory, and they would presumably make a run of Urban-Assault black Endless-Sphere cells with a proprietary amount of Phostech if we wrote a check, could we claim all the other colors to be 'old copies' too? [/edit]

2. It's very easy for anyone to say device 'a' is an improvement over device 'b'. The first question this creates in my mind is 'compared to what?' They're an improvement how? They're better how? In regulated industries, confirmation must be performed by independent labs in accordance with established ISO, ANSI, or other specification. I don't know of a performance specification for LiFePO4, so it appears we're in a bit of a 'wild west' situation where anyone can say what they wish. It would be a significant marketing advantage for the folks at LiFeBatt and/or BMI to produce charts to prove the claimed superiority of their products. I cannot understand why they do not - unless they cannot. I've already placed a bet in Doc's testing thread that he will find no significant difference between the BMI cell and a PSI cell once his testing is complete.

3. Armin states that BMI and LiFeBatt cells now come from another manufacturer and are no longer connected with PSI. That could be. It appears that as of at least December 11th 2008 BMI and LiFeBatt purchase cells from the same manufacturer. As already stated, I have it from the source that it was PSI until fall '08.

I understand that PSI holds the patents on the design and manufacturing of these non-standard size cells with their terminals posts and seals. I also suspect that PSI is a larger corporation than BMI and/or LiFeBatt (this could be incorrect - I cannot find any information on BMI. I would expect a larger company to act a bit more 'corporate' though. This seems to favor PSI.). It doesn't make sense to me that a small company would put themselves in a position to be sued out of existence by violating PSI's patents - that's generally not a good long-term business strategy.

In addition, the photos posted on Armin's site, and that he may have supplied to the folks at Electric Echo, do show the manufacturing facility used to make these cells - and it's the PSI production floor with PSI's equipment and people. I don't know where they assemble packs with the completed cells. It would seem that a company with a new production facility might use photos from that facility instead of a facility they go to so much trouble to distance themselves from in print.

4. BMI continues to cling to the '20C' marking on their packs, and uses it in their advertizing. LiFeBatt did as well, but Don Harmon posted on this forum that it was an erroneous discharge rate and would no longer be used by LiFeBatt. Here are two facts that suggest this is marketing-speak only and has no function in the 'real world'. First is that 20C on a 10Ah cell means a 200A discharge rate. Even if the cell could flow that much energy, it would be very dramatic - for about three minutes. Then the cell is empty. Assuming they follow pack building guidelines published by the VMS manufacturer, All New Energy, the packs are built with a 100A fast-blow or a 75A slow blow fuse between two adjacent cells. The pack won't produce more than 10C for very long before the fusible link blows and the pack is off-line.

[edit] This portion about pack construction, though correct, doesn't speak to the capabilities of the cell. There might be differences in performance, but it doesn't appear to be openly available. [/edit]

If any of this is incorrect, I ask anyone to correct my position. If they can prove their point, I will correct my posts. I will provide sources and contact info if an independent party wishes to verify my comments. I don't care who the company is or what they sell - I am interested only in accurate information. I'll take my store down before I'll use bad info or untruth to sell. That dog don't hunt.

Sorry mods...maybe this stuff can be stripped out and moved to fix the severe thread creep.
 
AndyH said:
Sorry mods...maybe this stuff can be stripped out and moved to fix the severe thread creep.
Thanks for chiming in.

I suggest we continue here, it is about time to start a different thread on battery testing anyway... it's almost June '09 and lots of new cells are popping up: Headway, PingV3, blue-blocks, red-cells, etc.

Interesting that despite claims that there is a new BMI/LB plant, there has been no horn-tooting about it, let alone a verifiable street address. :roll:
 
Andy -- I don't consider your post to be 'Thread Creep' but very relevant info. Let's see if any mfgs can accept the 'Accuracy Challenge'
 
Hi everyone,

I am new to this forum, and am glad to have found it.
I am posting here because I have an EV - a modified scooter which currently uses Lead Acid batteries, and I'm considering upgrading to LiFePO4.
As such I have been in contact with Armin at Lithium batteries Australia, and as previously noted, he also was not forthcoming with technical data, he did however suggest that I purchase one cell, and perform tests on it to my heart's content. Which I did. I was particularly interested in the internal resistance of this type of battery, since this affects the rate of charge and discharge. I am however having some difficulty doing the math, the formula I'm using yield very different values for internal resistance depending on which data point I used. - Which shouldn't be so. I read that if I use the difference in voltage drop divided by the difference in the discharge rate (in Amps) this would yield the internal resistance. In other words: R = delta V / delta I.

Anyway, here is the data I collected. Take a look and see what you think.
These are rated at 10 AmpHr with 3.3 Volts nominal voltage - of course that varies with the load.
I did 3 tests, one at 1 Amp, one at 10 Amps, and one at 25 Amps which works out to C/10, 1C, and 2.5C

Any advice on calculating internal resistance from the data would be greatly appreciated.
Armin has been suspiciously quiet since I sent him the data.

Cheers everyone, and good luck with your projects
3wheeler2View attachment BMI Test.jpg
 
Nice data. To measure cell resistance, you normally alternate the load between two values, but your graph looks close enough. If you take the 10A and 25A graphs, the difference is about 0.1v.
Then take 0.1/15 (the difference) and you get about 6.7 milliohms, which sounds pretty good.
 
Thanks for the reply, Actually at 1 Ahr. it's .088V / 15 = 5.86 mOhms. But if I do the same calculation between the 10 Amp and 1 Amp test I get .14V / 9A = 15.5 mOhms.
that's almost 3 times more. I may have to do the 1 Amp test over again. The cell seams to have gained some strength after the first few cycles.
 
I think you need to cycle the cell 4 - 5 times before the tests will yield accurate data. But carry on, this is a great test!
 
i have been making some comparison tests of a grey BMI cell i just received from Armin and one of the green Lifebatt cells i received from the factory in Taiwan. There is a HUGE difference. The impedance at 1 kHz of the original green cells is typically 5.5 milliohms. The grey cell is 2.5 milliohms fully charged and the impedance repeatably drops to 2 milliohms at 50% discharge. The grey cell provides a significantly higher voltage over the whole discharge at 10A; more voltage than can be explained just by the impedance difference. i have been waiting to have complete data before releasing my tests; i am finishing up my 100A test set and will have that data in the next few days. I expect that the higher current tests will show even more clearly how much improvement there is in the grey cells.

my results for the green cell are in line with the results i posted long ago.

i should point out that the CBA2 has some fairly significant errors even at only 10A. My unit reads 40 mv high across the full range, and the current was off about 5% even at 1 amp. This makes comparisons between two units a bit suspect unless they have been calibrated. I have calibrated out the current error, but i am not going to pay West Mountain Radio $100 for software that would let me adjust out the voltage error. IMO this is not the way to do business; charging people for software that fixes problems with their hardware.

to calibrate out the current error in the standard software just put a 1% resistor of 10 milliohms or so in series with a test cell or supply, set the CBA2 to a current and measure the voltage across the resistor then enter the error into the space provided. the voltage calibration function is not available unless the pro software is purchased. As long as two cells are compared on the same CBA2 the error is not a big problem.
 
Indeed. If testing for comparison's sake, then just having a consistent error is often good enough.
 
3wheeler2,
Armin's offer to buy just one cell for testing should be welcomed,
that is what we asked cell manufacturers all the time on this forum.
How many cell distibutors will sell you one cell. ?? LifeBatt, Headway??
Very good practice, Armin.
MC
 
armin sent me a cell for testing, because he knows i will give it the same objective review i gave the lifebatt cells. no matter what i think of don h. i will provide objective results. i also purchased a cell out of the recent bulk buy of headway cells, and will post the test results in 5-6 weeks when i get that one. this is one of the ways i feel i can pay back the favors others have done for me.

i am sorry that when i called tests made with light bulbs and a cycle analyst "reasonable" it was taken as an insult. it was not meant that way. i have 4 of the 6-1/2 digit hp meters, calibrated to a lab standard, an hp 4328 impedance meter, also with NIST traceable calibration, and an assortment of 1% resistors in .005 ohm increments. i can graph temperature to .5C accuracy over the discharge cycle, and i will be providing test data that can be repeated by anyone else with traceable calibration. I still would not call my data anything more than "reasonable".

i apologize for not having the data today, but i am going fishing for a couple of days. if anyone has a grey Lifebatt cell they want to loan me for comparison with the latest BMI cell, i will be happy to test it and pay return shipping.
 
bobmcree said:
... IMO a teflon washer secured by a nut and no pressure relief valve is not a good way to keep electrolyte in a cell...

I can update this info a bit. The cells do have pressure valves - one is installed in each end. They correspond with the 'X' cut in the black and red stickers on the cell end.

View attachment outside_vent.jpg
View attachment Inside_vent.jpg

The seal is more sigificant than just a washer. Yes - there is a stepped washer on the outside. While part of the seal, it seems the outer washer's primary function might be electrically isolating the terminal from the cell case. This outer shoulder washer is much thicker and more solid - this is the one that appears to take the physical loads placed on the terminal.

There's a more complex seal on the inside which includes a recess on the inner surface of the end cap, a square plate machined into the inner portion of the terminal, and a thinner, more pliable, seal. The inner seal material has much more surface area and a couple of 90 degree transitions - including a shoulder around the terminal stud. This configuration allows the seal to tighten as internal pressure builds.

Worst case scenario for any seal is loss of clamping pressure. I expect this seal would be quite robust as long as the outer brass nut is properly tightened before connecting the cell. I've received a couple of cells with loose nuts - presumably from vibration during shipping. I won't sell them, but they've given good service so far.

View attachment seal.jpg

I'd like to say that no cells were harmed to bring this info to the group, but can't. This one was quite upset.

View attachment steamed.jpg
 
Nimbuzz said:
I think you need to cycle the cell 4 - 5 times before the tests will yield accurate data. But carry on, this is a great test!

It will take 10-12 cycles before one sees the cell begin to stabilize. I've seen the capacity increase thru cycle 14 with PSI's cells.
 
Andy -- Very interesting -- Thanks for the inside dope!

Why is that cell smokin? And isn't it smoking from the end cap rather than the 'vent?'

ATTN TESTERS; Repeat, cells may not stabilize for testing until after 12 cycles! Good to know = thanks again.
 
Nimbuzz said:
Why is that cell smokin? And isn't it smoking from the end cap rather than the 'vent?'

The cell wasn't completely discharged when I attacked the case with the cut-off wheel. My second cut went into the cell structure, shorted it, and the electrolyte started to cook off. It didn't need to use the vent - it had a two inch cut in the case at that point. The cell got hot enough to melt the rubber vise protectors for the first couple of inches but not hot enough to burn. The smoke was cool and had a very light solvent/alkaline smell.
 
Thank you bobmcree for your excellent information, I look forward to your test results.
I was unaware of the inaccuracies of the CBA II, of course calibration is necessary for any in depth tests.
I'm kind of new to this so...

Anyway I neglected some information that might be useful;
1 The BMI cell is a grey cell number 4BF10I13-1033. there is probably some information incoded there but I don't know how to decode it.
2. There was no temperature data recorded in the original post because I don't have the "pro" version of CBA II, but I do have the temperature probe and collected data manually as the test proceeded. Note at 1C or 10 Amp discharge there was little increase in cell temperature, about 2 degrees Celsius over the entire one hour test. However there was a significant increase in cell temperature at the 2.5 or 25 amp discharge rate. I observed a linear increase of approximately 0.9 degrees Celsius per Amp Hour of discharge. A total of 8.3 degree increase over the entire 20 minute test. Test started at 23.0 and ended at 31.3

Whatever the internal resistance/impedance of the cell, this is really the telling factor for me. I plan to use these cells at about 1.5 C continuously with peaks of up to 4 C during hard acceleration. I figure the cells will not head significantly under these conditions and therefore not cause premature damage.

I have placed an order with Armin for an HPS-4830 battery and 10 A charger. I'll let you guys know when I get it.

looking forward to more test results.
3wheeler2
 
3wheeler2 said:
Thank you bobmcree for your excellent information, I look forward to your test results.
I was unaware of the inaccuracies of the CBA II, of course calibration is necessary for any in depth tests.
I'm kind of new to this so...

Just another data point....
My three CBA II's are all witin 1% of the set current value and voltage reading up to about 19A-20A. Above that the sense resistor drift rises up to about 1.7% off at 30A. The voltage is off by whatever the drop is in the leads from the CBA II.

That's one of the biggest reasons I developed my CC-400 load, to allow the CBA to operate at very low (and very accurate) current levels that won't exceed the, IMHO, 65W or so max rating for the CBA. :mrgreen:
 
Do you guys have the CBA Amplifier? if so do you use it -- or is it for packs rather than cells?
 
I have loads that I designed set up for cell or pack discharging up to 750A at 2000W. I've seen load users (mine and the CBA Amp) using them for both packs and cells. They all handle discharges down to low voltages so can handle single cells.
 
Nimbuzz said:
Do you guys have the CBA Amplifier? if so do you use it -- or is it for packs rather than cells?

I have a CBA amplifier and use it for tests above 10A. Since it's a 10x amplifier, a 100A discharge only requires 10A from the CBA. I've found it to be very accurate.

Two amplifiers can be paralleled for up to 1000W. It has a 48V max voltage, so it can't handle a fully-charged 48V LiFePo4 pack.
 
the CBA2 is a good tool for what it is, but at higher currents there are many sources of error. the amplifiers they sell are great, but way too expensive. i have calibrated my CBA2 to work at 25A but beyond there it is not much good. an extra fan will permit longer duration tests, as if you tell the software it is >1Ah the current is limited to 10A and if you run it at 25A / 75W for more than a few minutes it will overheat and kick off.

one solution is to use the CBA2 as a data collection meter and use an external load for higher currents. i have a load that works up to 100A and have promised test data this week comparing the green lifebatt cells with the later grey bmi cells. some people seem to still believe there is no difference, but tests show there is a big difference, mainly that the impedance of the grey cells is less than half that of the green cells, resulting in higher discharge voltage over the whole cycle and reduced cell heating in the grey cells.

i wanted to wait until i had more complete data, but i have been convinced to release meaningful data as i collect it. i have a headway cell coming from the bulk order, but if anybody wants to loan me one sooner i will provide test data sooner. i would also love to get ahold of one of the ping cells if anybody has one to loan me. i will be publishing the data on my 100A load this week for anybody else who wants to duplicate it, and i will be willing to loan the load out to people i trust.

an average load of 25A is fairly realistic for ebikes, and is within the capability of the CBA2 with an extra fan. i have calibrated my unit so that the voltage and current on the graph are accurate to 1%. I will have more photos of the load construction and more graphs in the next day or two.View attachment green lifebatt vs grey bmi cells at 25A.JPG

the graph shows that the voltage output is lower across the full range for the green cell, by almost exactly the expected difference between 5.5 milliohms and 2.3 milliohms for the grey cell, and that all that lost power is heating up the cell instead of going to the load. graphs at higher current should this effect even more clearly, and i expect to fry an egg on the lifebatt cell after discharge for 6 minutes at 100A :)


one member has already offered to send me a couple of ping cells, so if anyone has an extra headway cell we will not have to wait until mine comes from the bulk order.

btw i tried to insert the graph into the post but could not figure out how to get it large enough to see. any tips?
 
Excellent. So, we can rule out that BMI is also Lifebatt (And that lifebatt is actually worse.). Actually, I kind of wonder if the lifebatt cell isn't a 4 to 5C cell like headways? That output voltage seems a bit low at only 2.5 C. Also, interestingly enough, the headways and Lifebatts have the same form factor.

Nevermind, I think the Headways are around 9 mOhm, Lifebatt's at 5.5 and BMI is at 2.3? What about PSI?

Anyways, that'd be VERY interesting if BMI's was 2.3 mOhm.

Standardized internal resistance = R*Ah/V

A123 = .01*2.2/3.2 = 6.8 milliHour
BMI = .0023*10/3.2 = 7.2 milliHour

Holy crap! BMIs are practically functionally identical to a123s! Where do you order these wonder cells?

Interesting, this thread might be old, but it seems like BMI costs 50 dollars per cell? At about 5 times the capacity of a123s, that sounds like it's about as expensive as a123s harvested from a123 packs in terms of wh/$. But, I can imagine, it'd certainly be easier to put together a pack!
 
swbluto said:
Excellent. So, we can rule out that BMI is also Lifebatt

actually that is not the case. as far as i know psi makes both, and has a new state of the art factory. i have not had the chance to test a newer lifebatt cell so i cannot say for sure that they are not as good as this grey bmi cell. Armin tells me that 2.3 milliohms is the average measured value of the recently manufactured BMI cells and he has given me no reason to doubt him.

even if lifebatt cells are exactly the same as the bmi, i would rather deal with armin. there is ample evidence that don h. has misrepresented the facts about the lifebatt cells, like claiming they were made with Phostec materials and later admitting they were not, where armin has always been completely up front.

the original spec on the lifebatt cells i have is 20C so at 200A there would be about 240W of heat INSIDE the cell during discharge. small wonder they got some 10C labels and quickly covered up the original 20C.

i believe the headway cells are the same size 40 x 138 mm but the headway has flat ends for solder tabs and not the threaded studs like the psi cells.

i bought one headway cell out of the recent bulk purchase but will not have it for about a month. i just cannot see paying bs.com $38 when i paid $17 for one, so unless somebody wants to loan me one that data will have to wait. A kind member is sending me a couple of ping cells for comparison.

clearly the lower impedance is better both in terms of power delivered to the load and internal cell heating. heat has got to shorten the cell life, and when we are talking about cells with a potential to last several thousand cycles paying twice as much for better cells may be a worthwhile investment. I totally understand that some of us must make initial cost a major factor, and that if the BMI cells last 6 years compared to the headway cells lasting 3 years that may be acceptable to some of us.


swbluto said:
Holy crap! BMIs are practically functionally identical to a123s! Where do you order these wonder cells?

you can email armin directly and ask him. armin@ihug.com.au

i will be placing an order soon and i intend to build packs from these cells for sale. He is giving me a killer deal on my initial order but we have agreed that it is not to be a bulk order for a bunch of people like the recent headway order. I am putting together a business plan and will try to get financing to bring in enough cells to make it worthwhile. Interested parties can contact me privately at the email address at the bottom of all my posts and we might work something out.

i am finishing up the higher current load now and should have more data in a day or two.

swbluto said:
Interesting, this thread might be old, but it seems like BMI costs 50 dollars per cell? At about 5 times the capacity of a123s, that sounds like it's about as expensive as a123s harvested from a123 packs in terms of wh/$. But, I can imagine, it'd certainly be easier to put together a pack!

one of the curves i will generate is 4 a123 cells in parallel so that would be a valid comparison at least in terms of impedance and output voltage but as for lifetime that will take time.

i believe i can sell them for less than that and still make a fair profit. it will depend to some degree on shipping cost and the level of interest. batteryspace is selling the headway cells for $38 and in pack level volume i may be able to come close to that. no promises today, but those who know me know i am not in it just for the money. i want us all to have the best batteries for our bikes, and if i can make enough profit to keep my 11 year old truck running until i can build something electric that is adequate for me .
 
Back
Top