Influence of DOD and temperature on the internal resistance

[bellfador]

The internal resistance is the most important factor defining the efficiency of the battery. But the data for the internal resistance of a specific battery is incomplete. In pdf's of some batteries there is some information such as: "Internal Resistance - less than 38mOhm/ AC impedance 1 kHz". This is an inadequate way for optimal choice of battery in high current DC devices.

Let me shed some light on such a problem:
1) I have made a battery tester with an arduino, fast current sources and a custom electrical thermometer;
2) I have made some tests and I've compared the results with other devices like SkyRC B6 and several multimeters with an accuracy of 0.5%;
3) I have tested different batteries and I've colected data;
4) I have changed the test method and return to the point 3).

I am giving a link of the electrical scheme: http://i.imgur.com/mmsbqGt.png. The current sources are not perfect, but on the other hand they are fast and stable. The current is 0.98-0.82A and 8.4-7.5A.
The current shape is 8A and the delay is 1000us + 350us from arduino... : http://i.imgur.com/YyQBkrc.jpg. Now delay is 700us + 350us. The voltage sag for 25R: http://i.imgur.com/RH8PfdU.jpg and GA http://i.imgur.com/g9HPCfW.jpg. Look how high the influence of the induction is.

The final accuracy of the battery tester is 0.5% for the voltage, 1% for the current, 3% for the temp and 0.0001% for the time. All batteries are charged with 1.2A to 4.200V with 0.1A cutt off, the resting time is about 30 minutes. I have done 3 full cycles before the first test of every single battery.

Testing method:
1) if voltage > 2.630V: discharge with 0.9A, sending data every second;
2) for every 200 seconds:
a) measure ACIR: ten cycles 1000/1000us, 8/0A, ACIR = average ACIR;
b) if voltage > 3.1V (3.2V for MJ1 and GA): one cycle 10/10s, 8/0A, sending data every 0.1s, DCIR = (U1-U2)/(I1-I2);

Some quantities are given:
T - temp, *C;
U - voltage, mV;
I - current, mA;
time in ms or s;
DOD - dept of discharge, %;
DOD = 100 - SOC;
IR- internal resistance, mOhm.
ACIR - internal resistance in AC current, fast change;
DCIR - internalresistance in DC current, slow change;

First, I decided to test the Influence of the DOD on the internal resistance with a minimal influece of the temperature.
Living room, temp 21-24 *C.

1. MJ1 (new):
3300 mAh from SkyRC B6 @1A;
3228 mAh from the battery tester 0.9A / 8A;

2. GA (new):
3320 mAh from SkyRC B6 @1A;
3233 mAh from the battery tester 0.9A / 8A;

3. R25 (6 month / 15 cycles @ +1.5A/-8A):
2418 mAh from SkyRC B6 @1A (2520 new);
2387 mAh from the battery tester 0.9 / 8A;

4. VTC4 (new but maybe old brand or fake):
2040 mAh from SkyRC B6 @1A;
1978 mAh from the battery tester 0.9A / 8A;

Tests for HG2 and 30Q are coming soon. After that I will make tests for the influence of temperature in 40-60% DOD.

 

[Hillhater]

I'm sure this is good data, but it's hard to fully understand without parameter descriptors and units on the axis.

 

[bellfador]

5.HG2 (new):
2930 mAh from SkyRC B6 @1A;
2891 mAh from the battery tester 0.9A / 8A;

6.30Q (new):
2950 mAh from SkyRC B6 @1A;
2938 mAh from the battery tester 0.9A / 8A;

 

[Hillhater]

Excellent data bellfador.
I have seen simple single point data comparisons of AC and DC IR before, but never over the entire DOD range.
Thanks for adding the axis labels etc.
Are you contemplating monitoring the effects of extreme cell temperature also ?..Hot and cold .

 

[dogman dan]

My math learning disability makes me have a very hard time with graphs.

But looking at the first graph, am I correct in seeing that as the resistance increased, so did the battery temp? That makes sense to me, if you pull hard on batteries closer to the end of their capacity, they seem to heat fast at that point. when they are close to empty.

Is this from higher resistance at the end? or just that the battery has been running and warmed up?

I'd be curious to see how a nearly empty battery resistance is compared, testing one battery that is near empty and warm, and the same battery at the same state of discharge when it's cold.

Also a full battery too, starting from warm, and starting from cold.

 

[Hillhater]

.
But looking at the first graph, am I correct in seeing that as the resistance increased, so did the battery temp? That makes sense to me, if you pull hard on batteries closer to the end of their capacity, they seem to heat fast at that point. when they are close to empty.

Is this from higher resistance at the end? or just that the battery has been running and warmed up?
.

Doubtful that temp is the only cause.
That IR increase at the end , corresponds to a 1 deg temp increase !
Also note the higher IR at the start, but that is when the temp is 3 deg lower ?
If you look at the data in the Samsung PDFs for the 25R, it shows the effect of temp on the discharge voltage graph over the full DOD range. And we know that voltage sag is directly related to IR.

 

[bellfador]

Sorry for late answer.
I have limitation and don't want to damage my cells. But maybe in future I will make tests in extreme conditions.
In last chart (for 30Q) near to 35% DOD, DCIR rise up with 1 mOhm because opening of a window in the room and the temperature drop with 2 *C.

 

[Ohbse]

This is fantastic work! Interesting to see how this changes over time as a cell ages and of course with varying temperatures. Thank you for taking this on, it's very helpful data!

 

[bellfador]

The Influence of the temp on the internal resistance with a minimal influece of the DOD.
All cells will be charged to 3.8V, 0.2A cut off.

1. 30Q (new):

 

[Hillhater]

Excellent info.....Which cell is that ?
How about a spell in the fridge/ freezer :wink:

 

[bellfador]

30Q vs HG2:

GA vs MJ1:

25R vs VTC4:

 

[Hillhater]

Are you able to do repeated IR tests and get consistent close results ?
.. thanks again. :wink:

 

[bellfador]

Yes, I can. I have repeated some tests for the Influence of the DOD and the differences is little mostly because of the different temperature in the room.

Edit: I will repeat the test for influece of temp for MJ1 without tabs, because of each tab bring 0.7 mOm and I don't sure in my mathematical correction.
After that, I can repeat test for another cell, any suggestions?

Soon I will make test in pulse mode @8A, something like 10s on /30s off.

 

[bellfador]

From second test on MJ1 I measure internal resistance with 0.5 mOhm higher and updated the upper post.

I have started the pulse mode tests @8A, 10s on /30s off.
All cells will be charged to 4.1V, 0.2A cut off. The end voltage is 2.9V. Room temp 23*C +/- 0.5*C.

MJ1:
C= 2409 mAh; E= 29627 J; pulse counter= 107;

Weight: 47.18 g

GA:
C= 2497 mAh; E= 30897 J; pulse counter= 111;

Weight: 47.86 g

HG2:
C= 2306 mAh; E= 29350 J; pulse counter= 102;

Weight: 45.20 g

30Q:
C= 2327 mAh; E= 29675 J; pulse counter= 103;

Weight: 46.01 g

VTC4:
C= 1743 mAh; E= 22207 J; pulse counter= 77;

Weight: 45.45 g

25R (6 month old / 15 cycles @ +1.5A/-8A):
C= 2029 mAh; E= 25924 J; pulse counter= 90;

Weight: 43.63 g

VTC5A :
C= 2029 mAh; E= 26111 J; pulse counter= 89;

Weight: 47.34 g

 

[bellfador]

VTC5A is added.
2460 mAh from SkyRC B6 @1A;
2400 mAh from the battery tester 0.9A / 8A;

Pulse test:
C= 2029 mAh; E= 26111 J; pulse counter= 89;

Weight: 47.34 g

 

[bellfador]

A123 26650m1b is added.
2530 mAh from SkyRC B6 @1A;
2470 mAh from the battery tester 0.9A / 8A;

Pulse test:
C= 2245 mAh; E= 25819 J; pulse counter=101;

Weight: 75.55 g

 

[Nuts&Volts]

This is great data!! Thank you for sharing

 

[bellfador]

LiPo Graphene 1.5Ah 65C is added.

After 2 cycles with +1A / -1A and 3 cycles with +2.3A / -4A:

1492 mAh from SkyRC B6 @1A;
1467 mAh from the battery tester 0.9A / 8A;

Pulse test:
4.1V - 3.4V
C= 1265 mAh; E= 17379 J; pulse counter=58;

Weight of one naked cell: 42.71 g

 

[okashira]

Wow, pretty impressive power.
Great measurments

 

[okashira]

So your reported DCIR is essentially an "instantaneous" DCIR.

I would also like to see a "C/200" DCIR.
IE, for a 3.1Ah cell..
Run 0.9A, hold for 30 seconds
Run 9A, hold for 6.2 seconds. (9A for 6.2 seconds would be capacity/200)
You would adjust the 6.2 for the cell capacity. for a 1.5Ah cell, you'd use 3.0 seconds.
Record difference in voltage between the voltage just prior to initiating 9A, and at the end of the 9A run.
Use that for the pulse DCIR measurement. The number will be higher.

 

[bellfador]

What kind of information gives C/200 DCIR ?
No load voltage drop when cell lose some of its capacity. This is not happening bеcаuse of the internal resistance.

 

[okashira]

What kind of information gives C/200 DCIR ?
No load voltage drop when cell lose some of its capacity. This is not happening bеcаuse of the internal resistance.

Using c/200 gives you better real world indicator of DCIR. With your test, which is essentially C/1000000 or some really small amount , you are getting the "surface charge" from the ions thatbare right there on the surface of the anode so the dcir result will be artificially low.

My theory is that your method will actually help high capacity cells (with a thicker anode) show a low dcir relative to high power cells

C/200 is small enough so that any voltage drop due to SoC change will be very small relative to the IR voltage drop
But you can also correct for that easily with a linear approximation of the SoC voltage curve.

 

[fechter]

Yes, great measurements there. This is very useful information.

A great demonstration of why you need to avoid running batteries to near zero SOC.

 

[bellfador]

Using c/200 gives you better real world indicator of DCIR. With your test, which is essentially C/1000000 or some really small amount , you are getting the "surface charge" from the ions thatbare right there on the surface of the anode so the dcir result will be artificially low.

My theory is that your method will actually help high capacity cells (with a thicker anode) show a low dcir relative to high power cells

C/200 is small enough so that any voltage drop due to SoC change will be very small relative to the IR voltage drop
But you can also correct for that easily with a linear approximation of the SoC voltage curve.

My DCIR measure pulse is 10 seconds with 8A. This is near to C/100 for 2.2Ah cell.
Maybe you mean my ACIR test pulse - 1.05 ms with 8A or C/ 7 500 000 for 2.2Ah cell.

 

[okashira]

Ok. I must have misunderstood your methodology: "The current is 0.98-0.82A and 8.4-7.5A.
The current shape is 8A and the delay is 1000us + 350us from arduino... : http://i.imgur.com/YyQBkrc.jpg. Now delay is 700us + 350us. "

I thought you were getting DCIR from that.

In that case, that is much better. So my next critique is related, since you're at about C/100 - try to be consistent.
Id recommend targeting C/100 for every cell you test, and adjust the current to fit, keeping the same time.

So, 10 seconds for 8A on a 2.2 Ah cell

For a 3.1 Ah cell, it would be 10 seconds at 8*3.1/2.2=11.27A

For a 31 Ah cell it would be 112.7A for 10 seconds.
Of course this would be problematic for your setup for larger capacity cells, so you may want to choose a lower current to begin with

I think 10 seconds is a good # to use, as that is what the DoE uses in their test methodology:
http://avt.inl.gov/battery/pdf/PLUG_IN_HYBRID_Manual%20Rev%202.pdf (section 4.3)
They also give guidance on how to correct for voltage drop due to SoC change. I'd just use a linear curve fit though

They don't adjust test current based on battery size, which I don't agree with