What happened to my battery tester?

[katou]

I'm happily moving along, testing cells at 2c for a friend's pack, and whammo, suddenly, I hit a string of like 20 cells that all test out abysmally.

I check for resting voltage before I start testing, which removes all of the dead cells. Additionally, any cell with a resting voltage under 3.0 v, I write it on the can. That way when I test it I am aware that it will probably not be up to snuff. I test those cells anyway, just to see what happens.

There shouldn't be so many cells testing so poorly (like 0.01 ah) or even less. I suspect something else is going on.

Step 1: test cell at 2c, then recharge and test again at 1c
Result: See graph below:

View attachment Bizarre.jpg

The 2c test is on the far left. I ran it a few times (with the same cell), and each time it tracked nearly the same, 0.010ah. The long green line is the 1c test, 2.280 ah. Looks pretty good, but let's compare it to some very normal results:

View attachment Normal.jpg

Here I have overlaid test results from a week or so ago. As you can see, the green line from the first graph tracks the "normal" 1C curve, indicating that this cell is in great shape, no undue sag, good capacity, etc.

Step 2: try to eliminate other possible sources of error
I checked the charger, it is putting out 3.65v
I checked the andersons into the CBA, they're tight
I checked the cells coming off the Voltphreaks charger, they are coming off at 3.60-3.69

Step 3: ask for help

What would suddenly cause these results?

Thanks for the help, until I fix this, I'm pretty much dead in the water.

Katou

 

[webfootguy]

A couple of ideas. It looks like the "charge" step is failing. You get a normal discharge the first time, but the second time the cell is empty so it goes right to the low voltage cutoff. Look carefully at your charge setup. If possible put a watt meter inline with the charger that shows you the watt hours put in on recharge so you can see it is in fact charged properly. If the watts went in but you can not get them out, the cell might be shorted internally and thus the charge is lost to heat. If so, why only on high discharge?

Another possibility is the cells have some kind of discharge protection, like a PTC polyswitch. The 2 C discharge trips it to limit the current to the level you see. Are these raw cells or protected cells?

I'm out of any other ideas.

 

[katou]

These are raw M1 A123 26650 cells from (used) 36v Dewalt packs

Both times, the cell was fully charged, verified by voltage of 3.65.

I don't have a wattmeter, but I did note that:

1. charge cycle to top up from pack - approx 20 minutes, (recharge time with Voltphreaks of 20-45 min is typical for used toolpack cells)
2. did 2c test, lasted 30 seconds, cap shows as 0.010 ah
3. recharge cell to test again - 1 hr 40 min, typical
4. did 2c test, lasted standard 1 hr, cap shows 2.28

Wait a minute. If the cell tester is malfunctioning, ending the test early, thus giving an incorrect cell cap, why is the recharge in step 3 taking so long?

If the test is ending early, and the cell is in fact still full, the recharge should be very quick, even less than the initial charge time of 20 minutes.

If the chargetime for cycle 2 is indicative of an empty cell, then that suggests that somehow the cell emptied completely during the 2c test?

I will throw that cell back on the charger and report back with the chargetime. The wattmeter is on my list of things to get, I know, I know.

Katou

 

[katou]

Recharge time 40 min from empty.

Katou

 

[katou]

Discharged the cell again, 2c, then without recharging, did a discharge at 1c. Results were similar, but strangely not exactly the same. Capacity reads 2.148

View attachment Odd again.jpg

Katou

 

[katou]

I have decided that I am going to continue testing cells and see what happens. I will throw a few 1c discharges in there to see if cells are registering much higher capacities.

Here's the part that bothers me. One of my assumptions is that the behavior of the cell at lower C rate tests indicates the performance of the cell at much higher C rates.

These cells are saying that my assumption is incorrect.

Katou

 

[katou]

I am very troubled by the data, and what it suggests:

that just because a cell tests out great at 1c, it is not applicable at all to predicted performance at other C rates

If anyone can give some perspective on this, that would be great.

In the meantime, I suspect that my cell holder may have some issues, perhaps some connections are not good?

I am going to make another cell holder which will use much better construction, but it might take me a bit to get that done.

I will report back when I have the cell holder done and tested.

Katou

 

[katou]

I'm thinking further, and I have realized something. The only 2c tests I have done so far were with little 1.4 ah cells, so the max amp draw through my holder has been 2.8a and it worked fine at that level.

Now I"m testing at 2c for 2.3 ah cells, or 4.6a

Since I started at that amperage level, EVERY CELL HAS FAILED. This makes me confident that my holder is the problem.

Now, to find the time to build the damn thing. I just hope that's the problem.

Katou

 

[amberwolf]

Interesting thought. It certainly could be, if it's contact resistance is high.

 

[texaspyro]

I built a small AA NiMH cell powered device that swigged a healthy slug of juice (but not at all an unreasonable amount). The batteries were mounted in your typical black plastic holders with spring/rivet contacts. When hit with the current the springs, contacts, and plastic heated up and stopped making contact. When they cooled down some they made contact again. Lather, rinse, repeat... until magic smoke escaped.

 

[katou]

AW, could you elaborate on that? If the contact resistance was high, what would happen? How would that create the results I'm seeing?

Katou

 

[cell_man]

If it's the holder it can easily be confirmed by checking the voltage across the terminals on the cell when under load. If there is a big difference between what the DMM says and what the CBA displays, there's your answer. If the DMM voltage also drops very low, it's a cell problem. This is precisely why a battery tester should be a 4 wire system.

 

[amberwolf]

AW, could you elaborate on that? If the contact resistance was high, what would happen? How would that create the results I'm seeing?

If it was high resistance, it would cause a voltage drop across it. If the drop is sufficient for the CBA to see what amounts to LVC at the cell, it would cut the test off early and show significantly lower capacity than was actually there.

No, wait--if the contact resistance was high and the CBA is measuring the whole system (contacts and battery) then the voltage drop wouldn't matter, because it would still be voltage across the whole thing totalling up. I think. I may have confused myself. :? Stupid numbers.

But...no, wait again: If the contact resistance is high then the voltage drop across it plus the cell voltage would actually read *higher* than just the cell without load, at higher currents, and maybe that causes some error condition that stops the test early.

That still doesn't jive with the earlier report that recharging the cells did take the amount of Wh that they should have to recharge from empty, though.

I don't know. Do the test Cell_man describes and it'll tell you more.

 

[cell_man]

The voltage at the load can never be higher than the source. Any voltage drop across high resistance connections will have a potential voltage in the opposite direction, so therefore subtract from the supply voltage, not add to it.

Kerchoff's 2nd law states:
The directed sum of the electrical potential differences (voltage) around any closed circuit must be zero.

Unless you have a 4 wire system, any electrical resistance between probes and cell will look like higher internal resistance and make the cells look worse than they are. If you can measure the actual voltage on the cell, contact resistance is not important as the constant current load would compensate for this. Put a DMM on the cell while under load, that you will tell you what is going on. The contact resistance may not be linear and could possibly increase significantly at higher current levels.

 

[amberwolf]

Ok, so I had it right in my head the first time, before I started trying to run the numbers in there, too, which melted stuff and got me confused. :lol: I'm ok when I trust my instincts on stuff but when I try doing the math to "prove" it I usually screw up somewhere and end up lost.

 

[katou]

Cell man, thank you so much. I will test that straightaway. AW, thanks for the context. It's nice to have the techy info, and some less-techy info for context.

I will post back here after I check.

Katou

 

[katou]

Sonuvabitch. That was it. I put the mutimeter on the cell during a test and the cell never went below 2.55v, yet the CBA clearly registered voltage drop to 2v and stopped the test.

I did the test 3 times, and each time the lowest voltage recorded was slightly different.

Test 1: 2.72v
Test 2: 2.64v
Test 3: 2.55v

Yeeehaaa! When I was young, I wanted my problems to be very complex, so I couldn't be blamed for having them. Now I'm older, I HOPE that problems are because I screwed up something simple.

Major thanks for the direction, all.

Out to make a new cell holder. I believe I now feel confident stating that hotglue is not the material of choice in making a cell holder for higher amperage operation.

Katou

 

[amberwolf]

Well, that is good news; much easier to fix a cell holder than a CBA.

 

[katou]

Alright, I have made the new cell holder, and things are much better!!

That said, not perfect. I am registering voltage on the cell tabs of 3.28v but the CBA graph says 3.21v. What's up with that?

I wouldn't bother with the minor difference, but with these lithium chemistries, a little voltage change is a lot.

I think that perhaps there is still some resistance in my setup. I am using battery clips which might not be as solidly attached as they could be.

Do you think that is it? Or should I be looking into the 3 wire sensing that I've seen mentioned?

Katou

ps. or am I really lucky and I can just ignore the discrepancy?

 

[texaspyro]

One question is just how good are your meters (and the CBA)? A lot of cheap (and not so cheap) meters are rather awful... worse than that 2% you are seeing. See http://endless-sphere.com/forums/viewtopic.php?f=14&t=19805 for some cheap little toys that can answer that question.

 

[katou]

Umm, pretty good? Mine has very few scratches. My dog chewed on it a bit, so the belt-clip doesn't work anymore, but other than that...

The voltage reference device is a good idea. It is difficult to know what's what, especially when, as you say, my the discrepancy I'm seeing is well within the standard limits of tolerance of a cheapo DMM like I have.

I never thought of that before though, I'll probably get one, but hopefully I can test a few cells and sell them to pay for it!

I just finished the first cell at 2c, capacity of 2.29 ah, which is excellent!!

Katou

 

[amberwolf]

That said, not perfect. I am registering voltage on the cell tabs of 3.28v but the CBA graph says 3.21v. What's up with that?

What does your meter read on the other side of the clips that hold the CBA wires to the tabs?

If it reads the same then the reading is probably calibration difference between your meter and the CBA. Otherwise it's probably connection resistance, especially if it goes up with higher current draw.

 

[texaspyro]

Yeah, a lot of people mistake resolution for accuracy or assume that they go hand in hand. I have a 7.5 digit meter here (resolution to 0.1 microvolts) that is less accurate than my 3.5 digit (resolution to millivolts) hand held Chinese special... and that one is not very accurate.

Guess I gotta get off my duff and calibrate the things (along with a large pile of Tektronix DM5010's that have had the calibration memory nicads replaced with CR123A primary lithium cells). I'm a bit of a volt nut and have the equipment to calibrate 8.5 digit meters. The 7.5 digit meter is easy (I have the service manual). That hand held is another matter... lots of inscrutable Chinese diddle doohickies to work out... it does capacitance, inductance, freq, and temperature. Maybe best to leave it alone. Or not.

 

[katou]

AW, testing another right now, check this out, test done at 2c.

tips on cell: 3.11v

tips on clips: 3.08v

tips on bare wire running straight to CBA with no joints: 3.02v

CBA reads 3.00v

I see a pattern here, I guess that the pattern indicates increasing resistance with each step/connection?

I know my theory, but I've like zero experience on real live electronics stuff.

TP, it sounds like you've thrown away better stuff than I'll ever use!

Katou

 

[katou]

Experimenting with different C rates and measuring the voltage discrepancy.

2c = 0.11v
5c = 0.26v
10c = 0.46v
15c = 0.81v

These calculations are:

Vcell - Vgraphed by CBA = voltage discrepancy.

It looks linear, or pretty close, linear within the probable error of my equipment anyway.

I've been messing with this thing, I and I think that I have discovered the troublesome connection that might be causing the resistance.

I'm using 14awg wire, that's enough right? I could get lower, but I figured that would be good enough. Total length of wire is approx 8" between holder and CBA.

I will address that one connection, and then test again. If that doesn't work, I'll post a picture of the holder so you can suggest more possibilities.

Katou