# Designing a battery tester (Ah + internal resistance)

Discussion in 'The Projects Forum' started by Julia C, Oct 15, 2014.

1. ### Julia C Thread Starter New Member

Oct 15, 2014
9
0
I'm looking to create a battery tester (Ah counter + internal resistor checker).

Specs: Two different cells will be used - 100Ah, 3V cells & 200Ah, 3V cells
Ah counter discharge rate: 0.75C, so 75A & 150A respectively
Internal resistance: check to be completed at 400A (held for 10 seconds). Voltage measured before and after.

I've never done any high power design/wiring before. My experience is all in microprocessors and such. So I'd like some advice/someone to double check my design. Please and thank you.

For the Ah counter, I'm using resistor banks to draw power from the battery. I intend to use one circuit with a resistance of 0.04ohms to draw 75A (from a 100Ah battery), another circuit in parallel that would alter the overall resistance to 0.02ohms to draw 150A (from the 200Ah battery), and a third circuit that would load the battery at 400A for the IR test. I would open/close these circuits with contactors. Circuit diagram attached.

Are there any problems with this?

Questions:

1. The smallest resistors that I can find (that are rated for this power) are 0.25ohm resistors off Digikey. This means that I would need to wire 33 of them in parallel to achieve my desired voltage. Are there any better solutions?

2. I am looking to use two 200A, 12V contactors that we have in our shop, and purchase a 500A, 12V contactor as the primary one for the entire circuit. Something like this or this (the difference is coil current: 1.1A vs 130mA - I haven't figured out how this affects me yet). Is this reasonable?

3. How do I size and find a heatsink?

4. Anything else that I should be considering?

I appreciate you reading this far, and any help that you can provide!

Julia

File size:
1.5 MB
Views:
79
2. ### ronv AAC Fanatic!

Nov 12, 2008
3,404
1,478
Holy smoke.
Are these single cells that make up a larger battery? I'm guessing they are LIPO's???
Usually if you are testing to a manufactures spec they want constant current, which only makes it more difficult - but maybe you don't need that?
The problem with the chassis mount resistors is that they need a lot of heat sink to run at their rated power. I think the ones you are looking at need 290 square inches each. So by the time you count both the size and the cost of the heat sink you may be better off with just the old fashion tubular wire wound resistors and maybe some fans.
Would it be possible to water cool it?
At such high, and continuous currents, wire size becomes an issue. So I think I would mount the tabs of the resistors to a big copper buss bar to make it easier to attach the wires.
Don't know if this is much help.

3. ### wayneh Expert

Sep 9, 2010
12,394
3,246
There was a thread here not long ago on battery testing and the take-home message was to use a carbon pile. You may find it helpful.

4. ### Lestraveled Well-Known Member

May 19, 2014
1,957
1,218
Julia, Hello and Welcome to the forum.
I have a strong background in lead acid battery systems. I have a few questions:

- Most all of the major battery manufacturers base their amp hour ratings on a discharge rate of C/10 or C/20. Why are you using .75C to evaluate the amp hour rating of your batteries?

- The test to measure internal resistance, using a 400 amp load, is highly unusual. Could you explain the science you are trying to achieve.

I am asking the above because I want to assist.

5. ### MrChips Moderator

Oct 2, 2009
12,648
3,458
I'm with Les on C/10 or C/20 discharge rate. Discharging at C rate is rather severe.

6. ### Julia C Thread Starter New Member

Oct 15, 2014
9
0
The batteries are LiFePO4, I should have mentioned that, sorry.

The reason I am using a discharge rate of 0.75C is because that is how the batteries will be loaded in the vehicles they power.

As for why a 400A spike for an IR test - I don't have a good answer. Can you please explain the issue with this? I thought I could do the math (ΔV/ΔI = Ri). Would the high load change the properties/give me an inaccurate Ri?

I have no idea how to size a heatsink. Ronv, how did you calculate "290 square inches each"? We have an existing 80/450A load and it's heat sink isn't that gigantic. Photos attached.

Thank you!

File size:
1.9 MB
Views:
39
File size:
1.5 MB
Views:
42
7. ### Julia C Thread Starter New Member

Oct 15, 2014
9
0
A bit more about the 400A: that's on the higher end of Amperage the cells could experience in the vehicle, for about 10 seconds. So it's also to see if the cell can handle that (and rule out bad cells).

8. ### Lestraveled Well-Known Member

May 19, 2014
1,957
1,218
Julia, you have quite a task in front of you. Yes, your battery type and application explains everything.

- From my experience, the chassis mount type resistor you indicated seem to have a high failure rate. I suspect it has something to do with heat transfer from the element to the case and shock. I have had much better luck with this style resistor.
http://www.digikey.com/product-detail/en/FSOT5509ER2500KE/FSOT55-.25-ND/953865
It does not require a heat sink but does require a high air flow. I have yet to have one of these fail and I run them at 3 times their rated wattage.

- Consider copper buss bars for your resistor connection/mounting.

- Account for your cabling resistance.

- Check e-bay for resistors. No kidding.

Internal resistance
I need to make up some pictures/sketches. I'll address that later. (kind of busy) Something to think about. The internal resistance of a battery is independent from the amount of current used to measure it.

Last edited: Oct 16, 2014
Julia C likes this.
9. ### Lestraveled Well-Known Member

May 19, 2014
1,957
1,218
I would like to make another point.

Below on the left is a 1500watt space heater, on the right is my multi resistance load bank.

There are no heat sinks and they weigh about 6-7 pounds each.

There are 18 40 watt resistors that can dissipate 500 watts, standing vertically, and 1500 watts with the fan on.

Julia, you do not need big heavy heat sinks to dissipate power, just the right resistors and a good air flow.

File size:
378.5 KB
Views:
206
• ###### DSCF1351 small.jpg
File size:
386.4 KB
Views:
216
Last edited: Oct 16, 2014
10. ### nsaspook AAC Fanatic!

Aug 27, 2009
3,017
2,376
That's the two-tier DC load battery test. With a reasonable load ratio of about 10 for I2 from I1 it will provide accurate data but your 400A load will likely cause the battery chemistry to go into overdrive and skew the results.

http://www.cnlumos.com/lumos/Battery Knowledge/Battery Standards/IEC61951-2-2003.pdf
Page 45.

11. ### ronv AAC Fanatic!

Nov 12, 2008
3,404
1,478
I got the square inches from the resistor specification. (see attachment) But having said that it is the size needed to get the full 50 watts from the resistor on a flat plate (not a good heat sink). I suspect your current design does not need 50 watts each from the resistors. I don't know how you arrived at the number of 39 needed. I come up with 6.25 foe each 75 amps. So maybe 6 to allow some resistance in the wiring. That would be 37 watts each for the resistors. It might work on your heat sink with the fans.

File size:
117.3 KB
Views:
30
12. ### Lestraveled Well-Known Member

May 19, 2014
1,957
1,218
Measuring the internal resistance of batteries keeps getting harder as the battery technology improves. It appears simple, inside a battery is a perfect battery in series with an unknown resistor. The task is to discover the value of the unknown resistor. OK, simple enough, measure the voltage with no load and then with a known load, a little math and you are done. Nope. Lets look at the things that complicate such a seemingly simple measurement.

First, most batteries, without a load, will float to an artificially high voltage. This is sometimes called a surface charge. So a more accurate test will be to go from load "A" to load "B".

Second, the internal resistance of modern batteries, especially LiPO, is very very small, as in single digit Milli-ohms. Imagine a 3 volt perfect cell with a 0.002 ohm resistor in series with it. The delta V voltage due to the internal resistance is so small, the discharge rate of the battery degrades your measurement. You might think you can increase the current load to yield a larger, more readable, delta but you have proportionally increased the discharge rate and your measurement is still degraded. What is the key to making an accurate internal resistance measurement? Time. You must take the Va measurenent just before the load change, and Vb just after, before the discharge rate can degrade the measurement. Time for the graphs.

In the graph below, figure A, is an idealistic representation of what is happening during the internal resistance test, all inductive effects are removed. The battery voltage is changing before and after the load transition, so, the measurements must be taken as close to the transition as possible. The only way you can make this measurement is with an oscilloscope.

The trouble is, is that the world, and batteries, have inductance. Below, figure B, is what you will likely see on a scope. You will have to project, interpolate, a line, to evaluate the real Vb.

Figure C shows what you have to do.

See, it is not a simple test if you want to do it right.

File size:
31.2 KB
Views:
190
File size:
25.5 KB
Views:
189
• ###### internal resistance c.jpg
File size:
27.8 KB
Views:
188
Last edited: Oct 17, 2014
Julia C and nsaspook like this.
13. ### nsaspook AAC Fanatic!

Aug 27, 2009
3,017
2,376
A wise man you are.

I have some notes about the ESR testing routines in my BMS system. I'm testing GC-2 FLA types in series on two 12V strings (with different SOCs), the ESR is in milliohms. I'm not worried about the exact value in this case but I need consistent and stable values to help compute the battery health condition from the data.

ESR

Last edited: Oct 17, 2014
14. ### Julia C Thread Starter New Member

Oct 15, 2014
9
0
Thank you, I did forget about the inductance. After giving it some thought I agree with what nsaspook mentioned, that I won't need such precise data - I think that as long as I am measuring all of my cells consistently (past all of those spikes) I should be satisfied with my data. As well, 400A for 10 seconds is also a way for me to measure "can the battery survive short bursts of heavy load without setting off the Low-Voltage alarm".

The resistors you recommended Les, they don't list any heat/heatsink specs on their datasheet. How did you go about sizing your fan? I'll be having 37 resistors running, all with 35-40W of energy dissipating. That's double your setup.

15. ### Lestraveled Well-Known Member

May 19, 2014
1,957
1,218
The resistors I indicated do not utilize a heat sink, they use air flow. (reference Post #9) You do not need the mass of a heat sink to do what you need to do.

As far as cooling, I am attaching an app note from Ohmite, look at figure 14.

The BMS ( battery management system) that nsaspook referenced, appears to measure the discharge slope and correct for that error source. You, being a programmer, can easily automate the process and also correct for slope.

File size:
559.2 KB
Views:
21
16. ### Lestraveled Well-Known Member

May 19, 2014
1,957
1,218
nsaspook, your BMS seems pretty smart. Who makes it?

Oct 9, 2007
2,301
339
18. ### nsaspook AAC Fanatic!

Aug 27, 2009
3,017
2,376
I 'designed' it to maintain a small off-grid solar installation at the house to monitor and control the charging of 2 banks of 225Ah energy storage batteries with redundant 7Ah controller power batteries.

The experimental code for the original battery monitor is here: https://github.com/nsaspook/mbmc
The swm8722 folder has the main controller C program files.
The hardware is mainly from old scrap machines and leftover parts from work.
Old pictures of the system.
https://farm8.staticflickr.com/7049/6884674473_5a90daa5f2_b.jpg
https://farm8.staticflickr.com/7141/6848168603_751c114d7b_b.jpg
Web Interface:
https://farm9.staticflickr.com/8087/8430398032_e7ef490ca7_b.jpg

I'm currently designing a new system that uses CAN and Ethernet to link PIC controllers to a RPi for a simplified version of that monster.
https://www.flickr.com/photos/nsaspook/sets/72157638932852706/

Last edited: Oct 17, 2014
19. ### Lestraveled Well-Known Member

May 19, 2014
1,957
1,218
Bravo nsaspook Bravo . Do you compensate for the discharge droop in your internal resistance measurement?

20. ### nsaspook AAC Fanatic!

Aug 27, 2009
3,017
2,376
In the original software I don't because the calculated droop factor as it's usually under a few percent due to my use of smaller load resistors. This requires a precision measurement so I use a external Vref on the ADC channels and some software oversampling to generate a measurement resolution of only a few mV with a 10-bit ADC. The next system will use 12-bit ADC channels for even greater precision so droop will be included in the IR measurement.