The LiFePO4 Headway cell 38120P TEST REPORT inside

[Patriot]

As I went through the last 18 pages of thread, I didn't see anyone bring up the ZEVA website, so I'll post the link, and a graph below of what he got for discharge at 3C. He tested both P & L versions of the Headway 38120.

The author also claims it had marginal performance with a 0.5C charging rate.

I wonder if that would be enough for an Ebike.

http://zeva.com.au/tech/headway/

3C charge

Eeeww...

That didn't turn out so good.

Well, anyway, he shows a total 10.38ah, but it looks like it really started dropping off around 9ah.

 

[ZapPat]

As I went through the last 18 pages of thread, I didn't see anyone bring up the ZEVA website, so I'll post the link, and a graph below of what he got for discharge at 3C. He tested both P & L versions of the Headway 38120.

The author also claims it had marginal performance with a 0.5C charging rate.

I wonder if that would be enough for an Ebike.

http://zeva.com.au/tech/headway/

3C charge

Eeeww...

That didn't turn out so good.

Well, anyway, he shows a total 10.38ah, but it looks like it really started dropping off around 9ah.

From these zeva charts I get around 13-14milli-ohms DC resistance , so that's about halfway between the 17milli-ohms and 9milli-ohms that others have reported! I think as some have hinted at that headway may be having production problems, or for some other reason they have large variations in their cell characteristics.

Has anyone who bought from headway inquired about this high DC resistance, and why it seems to have changed for the worse since the first cells were tested?

 

[julesa]

why not join the headway group buy and save a ton of money on the shipping? joshua seems a straight up sorta guy who just ran into the lifebatt problem which seems to have followed us to this thread.

I would do exactly that, except that I refuse to do any business with Joshua on principle. He's rude and unprofessional, to say the least.

 

[Patriot]

Yes, they may be having some production issues, which is causing the variance in resistance between cells.

My guess is that over the next few months, as production production steadies with their new line, bugs worked out and QA is enhanced, they will be better than what we may be seeing now. Until then, I think they need to deal with the fact that they are going to have a few customers returning cells.

My guess, is that even though the cells aren't as up to snuff with what they may have originally designed in the lab, they will be good enough to satisfy the general use industry until the quality is brought up to the level intended.

You all have to remember, we are a finicky crowd, and we are the most critical. The average person will simply see the specs, buy the battery, use it, and be happy. If one cell cr@ps out, they replace it with expectation of doing so. Whereas with our bunch, that is entirely unacceptable. People like us drive the industry with very high demands, whereas most people just take it for what it is.

Most people see it as, "Oooo, hey this is better than those older batteries."

We see it as, "Hmmm.... this is better, but not quite good enough."

The ever unforgiving mind of the engineer.

 

[dnmun]

thanks for the link to the zeva site, i did some calculations of internal resistance on the cells they had analyzed. i also will fill in some other numbers i had scribbled in my notebook and posted over on the group buy yahoo site.

cell#25 at sub C gave meat .14Ah mark)
IR(2A,5A)=34mohm
IR(5A,8A)=34mohm

the average of #20 and #27:
IR(5A,10A)=30mohm
IR(5A,20A)=27.3mohm

#20 going from 5A to 29A.15Ah)
IR(5A,10A)=(3.19V-3.02V)/5A=.17V/5=34mohms
IR(5A,29A)=(3.19-2.52)/24=.67/24=27.9mohms

#27 going from 5 to 29A.14Ah)
IR(5A,10A)=(3.21-3.07)/5=.14/5=28mohms
IR(5A,29A)=(3.21-2.64)/24=23.8mohms
-------------------------------------------------------------------------------------------------------------

IR calculations from the headway charts i got off the group buy site:
(i used 20% of the chart width, since it is scaled to the 2 volt cutoff endpoint. and i used 20% after this on the other cells) i don't know how to make the correlation to either RLT's graphs or the graphs from zeva, headway used 1C,2C, and 3C ranges for their graphs)

1C: 3.18V,9.38A into .33ohm load
2C: 3.077V,17.945A into .167ohm load
3C: 2.948V, 27.742A into .1ohm load
IR(1C,2C)=(3.18-3.08)/(17.95-9.38)=11.7mohm
IR(1C,3C)=(3.18-2.94)/(27.74-9.38)=13.1mohm
----------------------------------------------------------------------------------------------------------

IR calculations from the zeva site: (using 20% of chart width, arbitrarily. 1.5" from the left edge, 7.4" wide chart, voltage and current normalized to the scale)

Headway:
1C(.33ohm load)=(3.188V@9.45A), 5C(.167ohm)=(3.067V@17.75A), 10C(.1ohm)=(2.933V@27.6A)
IR(1C,5C)=(3.19-3.07)/(17.75-9.45)=14.5mohm
IR(1C,10C)=(3.19-2.93)/(27.6-9.45)=14.3mohm

for the other lithium cells, i used 1C,3C, and 10C

PHET PE-1150: (3.176V@1.164A), (2.982@2.92A), (2.546V@9.99A)
IR(1C,3C)=110.5mohm
IR(1C,10C)=71.4mohm

Huanyu: (3.08V@1.14A), (2.96V@2.73A), (2.50V@9.79A)
IR(1C,3C)=75.5mohm
IR(1C,10C)=67.1mohm

Valence 18650: (3.21V@1.21A), (3.10V@3.03A), (2.812V@10.76A)
IR(1C,3C)=60.4mohm
IR(1C,10C)= 41.7mohm

Valence 26650: (3.18V@3.12A), (3.10V@6.11A), (2.73V@21.24A)
IR(1C,3C)=26.8mohm
IR(1C,10C)=24.8mohm

A123 systems M1: (3.18V@3.15A), (3.08V@6.16A), (2.67V@20.17A)
IR(1C,3C)=33.2mohm
IR(1C,10C)=30mohm

for the first of the K2 cells, there is no current track on the graph at 10C, current is off the scale so i use 5C for the first cell's final measurement, 10C for the others.

K218650E: (3.13V@1.21A), (2.98V@2.93A), (2.74V@5.45A)
IR(1C,3C)=87.2mohm
IR(1C,5C)=92mohm

K218650P: (3.18V@1.27A), (3.03V@3.01A), (2.73@10.67A)
IR(1C,3C)=86.2mohm
IR(1C,10C)=47.9mohm

K26650EV: (3.13V@3.13A), (2.97V@5.94A), (2.42V@18.81A)
IR(1C,3C)=56.9mohm
IR(1C,10C)=45.3mohm

K26650P: (3.10V@3.02A), (3.02@6.04A), (2.64V@19.98A)
IR(1C,3C)=26.5mohm
IR(1C,10C)=27.1mohm

i think rick has some lifebatts, it would be nice to run them through the CBA too. i was a little concerned earlier about the 11mohm readings of the headway graphs before i saw the 14mohm meausurements of the zeva headway cells, some of the current independent measurements are much higher, but they appear to be well within the lower end of the range of measurements of the other cells, including the Valence and a123 cells.

hope this helps, i learned a lot. dennis

 

[ZapPat]

i think rick has some lifebatts, it would be nice to run them through the CBA too. i was a little concerned earlier about the 11mohm readings of the headway graphs before i saw the 14mohm meausurements of the zeva headway cells, some of the current independent measurements are much higher, but they appear to be well within the lower end of the range of measurements of the other cells, including the Valence and a123 cells.

Hi Denis,
Great initiative here in comparing all the different cells for internal resistance... BUT make sure you scale the resistance results to each cell's capacity!!! very important here since your conclusion that the headway cells are pretty much as good as the others tested is much scewed because of this. Headway cells are 10Amp-hours each (10 000mAh), and here are the capacities of the much smaller cells tested by zeva (http://zeva.com.au/tech/LiFePO4.php):

PHET PE-1150 - 18650 - 1100mAh [9X]
Huanyu - 18650 - 1000mAh [10X]
Valence - 18650 - 1200mAh [8.3X]
Valence - 26650 - 2400mAh [4.2X]
A123Systems M1 - 26650 - 2300mAh [4.35X]

This means you'll have to scale all your results by these factors in bold for each cell.

Maybe we should have a standard measurement for DC cell resistance? Calculated on a standard 1 Ah base maybe? We have to be comparing apples with apples, so a 10 000mAh headway cell with 15 milli-ohms DC resistance would be 150mohms/Ah, whereas the Huanyu cell results are already in the right scale, so about 70mohms/Ah (twice as good as headway cells). And the A123 cells look like about 7mohms/Ah... 20 times better internal resistance than headway cells for the same capacity!!

Please edit your post with new scaled results so we really can compare cell quality on an even basis.
Ciao!

 

[dnmun]

i now think that the headway charts were from the zeva report. the numbers are just too similar, the loads identical, and the reason that headway labeled it 1C,2C,3C was because that was the current regime they saw in the graphs. the original zeva report used C designations almost arbitrarily(27.6A for the headway cell is not 10C). i was stunned when i was looking for the current track on that K2 cell, that they published without having the current track on the graph was something i would not expect from an experienced experimenter. all he had to do was change the scale of the data recorder, simple error which shoulda been apparent immediately.

because of the huge drop that flip found with the 38A load, down to 2.47V, i looked at the graph of the zeva/headway at 27.7A(3C) current and the voltage has only sagged to about 2.94V(similar to the Huanya), and would be expected to drop another .14V with the addition of 10 amps of current to get it to 38 amps flip was pulling. so the cells in that graph would have only shown sag down to 2.8V, but flip measured 2.47V, and RLT was showing 2.6V at the 29A current. add another 9 amps to RLT's graph of #20&#27 and the current would drop the voltage down to about 2.4V, very close to the voltage that flip measured. repeatability.

what stands out is that the cells in the zeva report appear to not have the same amount of voltage sag at 3C that is now being observed in these cells. i think jeremy had a similar observation, and non of the other cells have such a severe sag even though the apparent internal resistance is in the same range.

looking at the 1C voltages from an average of #25,20,and 27 we would expect the cell voltage on the current headway cells to be about 3.05V at 1C. the Huanyu cell is the only one with a 1C voltage in this range(all the others held a voltage around 3.18V), but the voltage on the Huanya cell did not sag nearly as much at high currents, and held 2.96V at almost 3C when the headways had dropped .4V below that. all the other cells held about .6V more than the headway at the same load around 3C.

if the apparent internal resistance remains in this range around 20-40mohm, drawing another 10 amps, to 4C+ could draw the cell voltage down to 2.1V, essentially at LVC.

a 16 cell headway battery, drawing 3C(30A) would be at 40V, delivering 1200watts. a similar 10Ah battery at 30A of valence or a123 cells would be at 48.3V, producing about 1450 watts.

 

[RLT]

Hey, I finally got over 10AH out of one of my 38120L cells!

Unfortunately I had to test it at a measly 1A to get it.

Still have about 10 cells that haven't been tested yet, so it is possible there will be a better one;
but I decided to take one of the best cells (9.63AH @ my 'standard' 8A discharge, down to a cutoff of 2.44V (Instead of the normal 2.5V, so as to offset the resistance of the CBA)) and test it at one Amp.

Ended up with 10.78 AH.

So, that's pretty good news for anyone running 20P packs, right?

The 45 or so cells that I have tested since changing the cutoff down to 2.44V have been looking a little better AH wise, a larger percentage making it over 9.0AH, but they are still very inconsistent, some as low as 8.6AH (@ 8A)

One kind of odd thing is that one of the cells that had been a little lower than the average voltage when I first received the shipment (3.2V instead of 3.35 on average) had somehow self discharged down to 0.79V while waiting for me to get around to testing it. There was no EXTERNAL reason for it to have self discharged, but it did, nonetheless.

The voltphreaks chargers wouldn't accept it when I hooked it up, but after half an hour on the lab power supply to bring it up to about 3V, it charged up fine the rest of the way on the voltphreaks. And it managed to preform amongst the best through two test cycles at 8 amps. 9.63 and 9.65 AH.

Had another cell that took about three times as long to 'top off' from the 'factory fresh' voltage as the rest of them have done, but it tested out pretty decently 8.90AH.

I'm not going to do the math right now, but it looks like since I changed the cutoff voltage down to 2.44, that the average of the cells is now a little bit over 9.0v... I'd guess somewhere between 9.05 and 9.12.

When I finish testing the last 10 or so, I'll crunch the numbers more accurately, but I just thought I'd give you all a little update.

 

[Ypedal]

Have you ever taken one down to the 2.1/2.0v Max minimum level ?

 

[RLT]

Have you ever taken one down to the 2.1/2.0v Max minimum level ?

Not on the CBA.... I did one (accidentally) down to 1.92V when I was trying the lightbulb/WattsUp test... According to WU, that one did 9.77AH at around 0.84->0.76 V draw.
(I'm not really certain that I trust the WU readings though... When running on external power for the meter itself, I think there is some sort of bleed through effect... I know there is when there is no load on the battery under test... whether that goes away when you put it under load, I can't say for certain.)

After that I tested it twice at the nomral 8A and got 8.74 then 8.69AH out of it... That was at a 2.5V cutoff, before I started setting it at 2.44V to compensate.

I was under the impression that 2.5v was about the minimum before you start to degrade performance / life span significantly.?

I don't have enough extra batteries to want to do anything destructive on purpose... I'm afraid by the time I get finished doing bad things accidentally, and the ones that aren't any good in the first place, I'm barely going to have enough to build the packs I need.

 

[dnmun]

i can sympathize with the fear of ruining the battery for our amusement. i discovered rcgroups and they are a good place to read to feel comfortable about how much you want to abuse your baby, er battery. bob used 2.1V for the LVC setpoint on the BMS, so i would feel comfortable myself in that case if they were my batteries. you are paying the price of being the lucky dog, and you have the CBA too!

the CBA looks like a neat hack, not sure how much parts would be, but it will be as good a download as winzip. imo.

 

[RLT]

the CBA looks like a neat hack, not sure how much parts would be, but it will be as good a download as winzip. imo.

Huh?
I must have missed something. I am totally lost as to what you are talking about there.

 

[dnmun]

i was thinking that it would be neat to be able to have an open source program that would run a battery analyzer similar to your west mountain radio unit, using parts that could be assembled by users. then the software to run it could be developed open source and shared. this would allow people to test their own battery packs, and i would like to see it be part of an integrated controller so that there would be some data storage of the batteries performance on the bike. justin is thinking of a new type controller, where the battery is depleted in a minimal fashion to optimize the amount of power delivered to the motor. a real coulomb counter. i can imagine a monster block of a controller housing, that would house a big AC transformer too, with rectifiers and some filtering to charge the batteries, so the onboard charger would be built into the controller housing, and the processor to run the controller fets could also be used in that charge mode to manage the charging, or a second IC to control the charging and data logging. maybe with another bank of rectifier IC's capable of delivering 4-5A@48V or something similar, to allow charging, fairly rapid, at the charging hot spots. justin was able to carry enuff batteries to go 60km, imagine smaller battery packs, but with a charger built into the controller, it could plug in anywhere, put back 250Wh of power in an hour while drinking coffee or shopping, then another 15 miles, etc, lower the overall weight by reducing the battery size in spite of carrying the charging transfromer in the controller case. that's why i thought the hack of the CBA would be popular.

 

[RLT]

Oh, I stand under you now.

Well, the regular CBA isn't all that complex, but it isn't really simple either.

I did post some photos of the 'guts' of one a couple of months ago, here:
http://endless-sphere.com/forums/viewtopic.php?f=2&t=278&p=76594#p83327

Maybe someone will get inspired and start hacking . It is well beyond my level of technical competence though.

 

[dnmun]

i saw the pictures which is why i thought it might be an easy hack. the program to record and display the data on the screen is beyond me, but i am thinking more and more that a fast charger could be built into the same housing with the controller like mr keywin is making for us at shenzen electric. they also manufacture huge power supplies too, i just bot two big 10A power supplies on ebay they manufactured, so they know how to build a rapid charger for a battery pack, and headway is also in shenzen too i think, shenzen is second largest export seaport in china.

this is entirely separate from what justin is doing with his controller, but i see no reason they will not be building such a system before long and using something like justin's drainbrain type display which means they will have current and voltage available all the time.

this controller could display and record the power consumption from the battery pack to protect it, for the lithium battery packs mainly to protect the cells from overdischarge and reversal, and if they know how much charge is taken out, then they can keep track of how much charge they are storing back into the batteries during a recharge, and that could be used to reimburse the supplier of electricity.

scenario like this: client stops at charging station and plugs in, his account is accessed, his battery pack requests a 250Wh replenishment charge, the charging hot spot lamp post does the credit transaction, approves the charge and begins a metered delivery, 250Wh for 5 cents say, go into store, come out later, unlock bike and unplug, then ride off. why not every street lamp post?

the problem is the transformer, for 48V 5A it would be big, and a lot of weight to carry, and the rectifiers and capacitors and parts would add 3-4 lbs, but with it you would have mobile fast charge, which along with universal charging hot spots, around the metro area, people could charge up when they needed it. then people could carry a smaller battery pack for the equivalent mobility. lowering the weight and cost quotient of an expensive component in the ebike system, the battery.

but this is a long way off the headway thread. except for the shenzen connection.

 

[olaf-lampe]

@dnmun

if you want to discuss charger-issues, please open a new thread.

I'm not moderating this thread, this is just my 2 cents
Olaf

 

[Jozzer]

Got some new cells today from Victoria, they have a new covering, i'm hoping that when I test them I will find they have lower resistance too!
Keep you posted...

Steve

 

[Ypedal]

Woooo.. nice !!!

Is the weight similar to the tabbed cells ?

 

[biggs]

Got some new cells today from Victoria, they have a new covering, i'm hoping that when I test them I will find they have lower resistance too!
Keep you posted...

Steve

Hello Steve,

that was what i want from victoria, my cells came in the paper pack. but i think that they are the same cells like mine. i have gott them 3 weeks ago. and i have build my pack 24S4P.

Today i measured the resistance for 4p and it was 3.7 mOhm. kets say that one cell is about 16 mOhm. but next week i will measure all cells.

bg Alex

 

[bquick]

I have a pack of 80 38120L cells for my motorcycle. They are arranged in 4P, 20S. I charge them with two FMA 10S chargers which balance the cells very closely with each other. I don't have my low voltage BMS working yet, so I limit the amount of discharge to around 50%.

After a couple of weeks, I had a cell fail. Since 4 are in parallel, it took three other cells with it. I was suspicious of another cell, so I told Victoria that I had 5 bad cells. It took a while, but she sent me 5 replacements, free of charge.

I replaced the 4 bad cells and got everything balanced again. The other suspicious cell seemed to be doing fine until a couple of days ago. With the pack at about 60% discharge, the bad cell and the other 3 cells with it all went to zero. I have asked Victoria for three more cells, but I haven't heard back from her in a couple of days.

I'm starting to doubt the QC on these cells. Unfortunately, it only takes one bad cell to make an entire pack unusable.

- Brad

 

[Ypedal]

Were the failed cells on the bottom of the stack by any chance ? Top, middle, ?

 

[Jozzer]

I have some of the cardboard wrapped cells too, I'm hoping that these blue ones are better than those.

If reliability turns out to be an issue I'll not be using any more. Why do you ask where the cell was in the pack Ypedal?

I'm glad my motorbike has lifeBatts, I'd hate to break down for the sake of a weak cell, and stripping one cell out of each pack would be a nightmare!
I originally ordered the Headway's to try on a motorcycle, but found them underpowered for the job

I'll try to remember to weigh a cell tomorrow too, but I think they are the same size/weight as before.


Steve

 

[Ypedal]

If reliability turns out to be an issue I'll not be using any more. Why do you ask where the cell was in the pack Ypedal?
Steve

Looking at the picture, and something i have been putting alot of thought into.. pack assembly. Looking at the makita packs for example, they use a white plastic >< shaped support between 2P stacks of 18650's, light weight cells, 2 stacked, not really a problem.

However, a stack of 350 gram cells, has to weigh down on the bottom row, and over time deforming the cans and squishing the cells, does this make a difference or not. i have no clue.. but it's one of those things, like building a better mouse trap ! lol

 

[bquick]

The first cells that failed were right in the middle. I don't think there was enough weight on them to cause damage. The last cells to fail were on the bottom. I'd say that there's a better chance that the cells were defective than that the weight deformed them. I have two packs. The cells that failed were in different packs.

- Brad

 

[RLT]

Very depressing news, guys.

I guess I'm going to have to re-think my future pack building designs. I was planning to build mine into easily reconfigurable 4S2P subassemblies, but with so many cells failing, (I had the one DOA cell, and four very questionable ones in my batch of 100) I guess I'd better stock up on schottky diodes and just build 1P groups, then parallel them. Easier troubleshooting that way.