Battery Performance Tester

Stuff I am working on
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

With the next test, using Sanyo Ni-MH 2500's, I put the matching pairs of cells in CH1 and CH3, instead of CH1 and CH2 as I normally do. We know that CH1 and CH3 should be matched more closely. The terminal voltages were also better matched. The following chart proves to be rather interesting on several counts, but first the chart:-
Sanyo-2700-1.jpg
Sanyo-2700-1.jpg (61.39 KiB) Viewed 16930 times
We see that CH1 and CH3 terminates nearly at the same time but the discharge profile is a little different. In fact, on CH3, you can see that the voltage actually rises for a while before declining. This is very odd and cannot be explained through battery chemistry changes. Since it only occurs on one channel and not the others, it cannot be attributed to sampling cycle issues either. I'm still puzzling over this one.

It may be that our sampling isn't all that far off. What we need now are matched batteries with same starting capacities. Nevertheless, it does look like battery chemistry does impact discharge profiles somewhat.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

Okay, I got the batteries tested with the MAHA C-9000 analyzer and that reported the following:-

GP-1600-NiMH cells
BATT1 1197mAh
BATT2 1255mAh
BATT3 1180mAh
BATT4 1217mAh

The same batteries were put through the existing test setup and the following graph was the result:-
GP-1600-1234ABCD-3.jpg
GP-1600-1234ABCD-3.jpg (65.55 KiB) Viewed 16929 times
There seems to be some agreement with the results in general although our load estimation is probably way off at this point. Assuming a 1ohm load, our readings are:-

BATT1 1154mAh
BATT2 1280mAh
BATT3 1196mAh
BATT4 1180mAh

We really need to sort out the PWM load business.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

Yet another test - this time with the Sanyo-NiMH-2700 cells, which was analyzed by the MH C-9000 to have the following capacitites:-

BATT1 1626mAh
BATT2 1421mAh
BATT3 1643mAh
BATT4 1418mAh

Note that BATT1 and BATT3 is better matched, while BATT2 and BATT4 are better matched. The resultant chart is as follows:-
Sanyo-2700-1234ABCD-2.jpg
Sanyo-2700-1234ABCD-2.jpg (66.56 KiB) Viewed 16926 times
The chart seems to match the analyzer results but there still seems to be a greater voltage drop on channel-1. This could well be due to the cell itself though.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

Right, in this next test, we revert to PWM-duty-cycle derived load calculations instead of depending on what the DMM read for us. Previously, the results didn't reflect what the DMM was reporting and we expect the duty-cycle approach to be more accurate. This was done using the same GP1600 batteries that were charged since the last test. The charger reported the following charge capacities:-

BATT1 1408mAh
BATT2 1480mAh
BATT3 1530mAh
BATT4 1335mAh

The result of the test is as follows:-
GP-1600-1234ABCD-4.jpg
GP-1600-1234ABCD-4.jpg (70.88 KiB) Viewed 16925 times
As you can see, the cells in this test are much better behaved, probably due to the correct lower load now and so it didn't cause strange cell chemistry. BATT1 and BATT3 did start with higher terminal voltages. The software reported the following tested capacities:-

BATT1 1397mAh
BATT2 1303mAh
BATT3 1434mAh
BATT4 1277mAh

It should be noted that I terminated a start after a few minutes to make some changes so I probably lost about 2mAh to 4mAh there. These results are a little odd though, not tracking the amount of charge it took but broadly following the trend. BATT3 still has the highest reported capacity, and BATT4 the lowest. BATT1 is pretty close but BATT2 is off.

At the same time, I've made some code changes to stabilize operations and to make the LCD display more useful. We may have to re-measure the load resistors again, and perhaps find a way to measure it more precisely.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

With more testing done using the duty-cycle based load computation, it's starting to look like we're on the right track. We tested the Sanyo2500's out of the charger this time. The starting capacities as reported by the charge intake were:-

BATT1 1849mAh
BATT2 1679mAh
BATT3 1846mAh
BATT4 1671mAh

For this particular test, I put BATT1 and BATT3 in channels 2 and 4 respectively in order to see if there are systemic errors from these channels since it looked like BATT1 and BATT3 always came out on top. So, in the interpretation of the chart, bear this in mind.
Sanyo-2700-1234ABCD-3.jpg
Sanyo-2700-1234ABCD-3.jpg (70.97 KiB) Viewed 16924 times
Note how BATT1 and BATT3 still comes out on top, reflecting the higher initial charge. There appears to be a strange "kink" around the 9700-second mark on the discharge curves of BATT1 and BATT3 and this could be due to channel crosstalk as BATT2 and BATT4 both end their test and the voltage on their channels rise suddenly.

The reported capacities are as follows:_

BATT1 1697mAh
BATT2 1457mAh
BATT3 1617mAh
BATT4 1504mAh

The trends agree although the actual capacities don't. It is important to remember that charge capacity may not equate to the charge intake as there will be inefficiencies. In this case, the results actually agree with the earlier analyzer profile of the cells, which was:-

BATT1 1626mAh
BATT2 1421mAh
BATT3 1643mAh
BATT4 1418mAh

At this point, we need to deal with four issues:-

1. The voltage reading spike when LCD backlight comes on during the start of the test
2. Channel crosstalk
3. Why terminal voltages on channels-2 and -4 always seems lower?
4. Why we lost data on CH1 and CH2 after the data dump?

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

In the fourth case, removing a battery from the holder will result in the channel turning off, and if so - no further readings were given since there is no reference battery. I suppose we could live with this scheme although it is easy enough to change. At least this isn't due to a code glitch or stack overflow.

Also, preliminary tests on the ADC's show that they are all reading within <0.001V error. Re-measurement of the load resistor yielded a new set of results though:-

0.97+ohms
0.98+ohms
0.97ohms
0.97ohms
0.97ohms

As for the first issue, it turned out to be a simple problem - I was setting the current limit on the power supply too low so that whenever the LCD backlight came on, it was pulling down the supply voltage enough to cause a voltage sag on the 3.3V supply, which resulted in an apparent voltage rise for the ADC. Sometimes the simplest oversight can create puzzling effects.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

Back to the GP1600 batteries. The charge they took were:-

BATT1 1410mAh
BATT2 1493mAh
BATT3 4027mAh *probably a fault here
BATT4 1235mAh

The problem with the voltage rise with LCD backlight is completely resolved. It looks like there might still be some crosstalk but it is not so evident in this test. The results are as follows:-
GP-1600-1234ABCD-5.jpg
GP-1600-1234ABCD-5.jpg (70.64 KiB) Viewed 16923 times
As you can see, the curve is very well behaved with BATT1 showing it's characteristic sharper knee. The reported capacities are:-

BATT1 1320mAh
BATT2 1217mAh
BATT3 1438mAh
BATT4 1125mAh

Notice that BATT3 took a lot more charge due to the fault and this is being reflected in the reported capacity. Other than that, the whole thing is pretty well-behaved now. Just need to verify the reported capacities. We're almost done with the hardware and software design. Once done, we can get on with testing of cells.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

We're testing the GP1600's again, this time charged through the Re-Cycko charger and put into the holders in the reverse order.
GP-1600-1234ABCD-6.jpg
GP-1600-1234ABCD-6.jpg (56.32 KiB) Viewed 16920 times
The reported capacities are as follows:-

BATT1 1357mAh
BATT2 1428mAh
BATT3 1408mAh
BATT4 1485mAh

Notice that now batteries 2 and 4 seem to be showing higher capacities. This suggests that the channels 2 and 4 readings might be biased or that the charger put in different charges. This test needs to be repeated. We will have to measure the MOSFET's Rds to see if it is having an impact here as well.

*update: Tests with the DMM is showing a full on Rds of (ohms) 1.26, 1.26, 1.25, 1.26, 1.26 respectively. This means that the Rds is behaving quite consistently - certainly not enough to account for the problems we are seeing thus far with the lower voltages. We are going to use these new measurements for the duty-cycle calculations but at these resistances, the quality of the connection can make a bit difference. Our earlier computations did not take the Rds into account and this could mean that earlier capacities are over optimistic. It is, however, possible that the Rds gets lower with higher bias voltages but the characteristic graphs so far suggest that it tracks more or less linearly.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

We re-did the test with our modified resistive load coefficients and as expected, the reported capacities are lower.
GP-1600-1234ABCD-7.jpg
GP-1600-1234ABCD-7.jpg (72.52 KiB) Viewed 16919 times
The reported capacities are:-

BATT1 1261mAh
BATT2 1168mAh
BATT3 1284mAh
BATT4 1220mAh

Well, it appears that cells 1 and 3 have now re-gained their improved performance. This looks suspiciously like the result of un-matched channels. I will repeat this test again to confirm this.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Battery Performance Tester

Post by Daniel Wee »

This next set is the Sanyo 2700's after 2-days of reconditioning (break-in) on the C-9000. The end report of the reconditioning are as follows:-

BATT1 1625mAh
BATT2 1389mAh
BATT3 1563mAh
BATT4 1431mAh

This seems to match quite well with our test results:-
Sanyo-2700-1234ABCD-4.jpg
Sanyo-2700-1234ABCD-4.jpg (71.44 KiB) Viewed 16918 times
The reported capacities are:-

BATT1 1927mAh (1529)
BATT2 1440mAh (1143)
BATT3 1833mAh (1466)
BATT4 1530mAh (1214)

There was some computational oversight in the code which resulted in higher than expected results. The numbers in brackets show what the correct calculations should have been. Those numbers seem somewhat lower than the analyzer reported so we may have to re-check what the Rds+load resistance is. In general, however, there is agreement in terms of the ranking of the cell capacities. The "kink" in the channel-4 curve near the start was probably due to my trying to adjust the seating of the battery after the test started.

Daniel
Post Reply