Introduction
In this blog, I intend to carry out testing on some individual batteries I have available to me with the BT3554-01. I aim to carry out measurements over the discharge cycle of a 12 V VRLA cell battery and a 4.2 V Lithium Ion battery. Overall I hope to be able to show how the AC resistance measurements vary for a lead acid based battery as it is discharged, which is the predominant reason as to why the AC resistance is always measured with the cells on float charge and to also show that AC resistance measurements are not the best option for testing Li-Ion batteries.
For any type of battery technology, discharge testing it from its fully charged voltage down to its discharge voltage level, is the one true accurate measurement of its capacity and performance. For a small single battery, a discharge test is not to onerous. For larger multi-cell battery banks that provide emergency power to critical infrastructure, it can be quite a burden, hence the drive to find other methods to determine the state of health of a battery bank.
This blog forms part of my RoadTest review of the Hioki BT3554-01
Hioki BT3554 Battery Tester Review
12V Battery Test Set-up
The test arrangement is relatively simple. The battery is charged to its full state by being held at 13.7 V until the charging current had dropped to zero. The battery is then transferred to a DC load and a constant current drawn from the battery until it reaches it nominal discharge voltage of 10.8 V. An AC resistance reading is taken with the Hioki BT3554-01 every 5 minutes and then the voltage and resistance values can be plotted.
12 V Battery Measurements
The table shows the measurements made during the test. The discharge current was 1.6A, which is half of its rating. The measurements were made directly off the battery terminals using the straight probes, therefore the standard zero adjustment as per the advice from Hioki was made prior to the test.
From the manufacturer's data sheet, the battery has a nominal AC resistance of 50 mOhms when measured at 1 kHz. The initial reading obtained was of 67.7 mOhms. That is 1.35 times the nominal value. The battery as a nominal life of up to 5 years when operated at 20 degrees Centigrade, and is around 2 years old now, so does seem to have already aged considerably.
A sizeable voltage drop is seen from the no load voltage to the voltage when the load is initially applied and is typical of the response from a lead acid cell. From then on, the voltage drop per step is somewhat smaller.
The AC resistance shows the opposite curve. Initially there is little increase in the AC resistance but after two readings on load, it is seen steadily increasing as the battery further discharges.
After the final reading on load, the load is removed and a further reading taken straight away. It can be seen that the voltage recovers very quickly, but the resistance stays relatively constant. This will only start to drop once the battery is placed onto a charge cycle.
The voltage after de-loading was 11.49V, which is actually reasonable for a 12V battery. The resistance on the other hand is over 4 times the manufacturers nominal value and around 3 times the original measurement prior to the test. The general guidelines for VRLA battery monitoring with AC resistance measurements, is that the battery is approaching end of life, when the AC resistance is 1.5 to 2 times that of the specification. The test shows that if the battery is measured in a state of discharge, then there is the likelihood that the battery may appear to be end of life. It is therefore important that the test engineer verifies that the battery had been on charge for the required time before carrying out AC resistance measurements.
4.2 V Li-Ion Cell Test Set-up
Unlike the VRLA battery that can be wired directly to the load bank from its terminals and any measurements made are also directly from the terminals, the 18650 cell I have does not lend itself so well to this arrangement. I have some battery clips left over from another project, that I soldered to a piece of strip-board and test wires from these clips to the load. I can take a measurement from the back of the clips with the 18650 cell in position, so that will add to the overall AC resistance value obtained.
The 18650 cell is initially charged with a bespoke charger circuit up to 4.7 V and is discharged down to 2.80 V. The discharge current this time is 2.6 A which is the nominal 0.2C rating for this cell. The manufacturer's data sheet is a bit light on AC resistance measurement, it again specifies a 1 kHz test signal, but merely states that the actual value is expected to be less then 100 mOhms for a fully charged cell.
18650 Cell Measurements
The initial measurement shows the resistance with the probes applied directly to the cell terminals, the second measurement shows the increase in resistance when the measurement is made with the cell mounted in the clips. The voltage is consistent for these two readings, as the cell is not yet under load. Once placed on load, the voltage drops down to just under 4 V, but the resistance remains relatively consistent.
During the test, a steady decline in terminal voltage is observed, but the resistance is seen to be much more consistent, staying between 46 to 50 mOhms throughout the whole test. As with the VRLA battery, the voltage from the 18650 cell is seen to jump up at the end of the test, once the load is removed.
In comparison to the 12V battery, the 18650 shows twice the voltage increase when the load is removed, indicating a faster recovery. The AC resistance reading is slightly lower than when the test started, unlike the 12V battery, where it is over 4 times its initial reading.
It seems that whilst the AC resistance value of the VRLA cell is affected by its state of charge, it is not affected by the state of charge for a Li-Ion cell.
To further investigate AC resistance measurements on Li-Ion cells, I have a couple of old battery packs from Ryobi power tools, one of these battery packs has failed and will not charge up in the charger, so seems an ideal candidate to take a look at and investigate with the BT3554. One thing to note is that all of these battery packs are the smaller Ah packs with a single stack of cells within them, I can therefore measure the AC resistance of each cell. Larger Ah packs have multiple cell stacks and I would be reading the AC resistance of cells in parallel and not an individual cell. One slight disadvantage, is that in order to carry out the measurements, I have to break open the battery packs to gain access to each of the individual cells.
For these measurements, I utilised the Bluetooth function of the BT3554-01 to send the results straight to my iPad 4. For the second set of measurement, I sourced the data sheet for the specific cell within the Ryobi battery pack. The information on impedance measurement of this cell indicated that it should be below 30 mOhms with a 1 kHz test signal. I therefore set up limits within the BT3554-01 of 45 mOhms for a warning and 60 mOhms for a fail. These values are based upon the 1.5 and 2.0 times limits established for VRLA cells.
| {gallery} 18650 cell limits |
|---|
45 mOhm warning limit |
60 mOhm fail limit |
With the limits set up in the BT3554-01, there is a beep and the screen turns red momentarily when a measurement exceeds one of the limits. This function can be turned off if desired or altered so that the BT3554-01 beeps for a cell that passes the measurements. Note that the reading in the picture on the BT3554-01 is different to that on the iPad due to the slight delay in sending the readings to the iPad. By the time the iPad is updated, the screen is no longer showing the red warning.
The test results are analysed after the failed battery pack is manually recovered using a power supply to charge the cells directly. The results are shown in the table below. The oldest battery pack, 02-2009, whilst it will still recharge, shows 3 cells that are below the nominal 4.2 V cell voltage. Two of those cells have AC resistance readings that exceed the alarm limit. In comparison the next battery pack, which is the youngest, shows all the cells at around 4.1 V and the AC resistance values are all less than 30 mOhms. Note also the low voltage and AC resistance differentials across these cells in comparison to the cells from the 02-2009 battery pack.
The 03-2009 battery pack that will not charge, shows very low cell voltages that are actually below the 2.5 V minimum discharge value from the manufacturer's specification sheet. One cell is above the minimum level and is the culprit fir the high differential across the voltages. Interestingly, the AC resistance readings for these cells are only slightly above the 30 mOhm limit for a new cell. The cell with the higher 2.77 V is below the 30 mOhm limit. It would appear then that the BT3554-01 is saying that the cells within the 03-2009 battery pack are in a reasonable condition despite the low voltage charge.
The final reading in the table, shows the cell measurements of the 03-2009 battery pack after the voltage on them was raised with a power supply and then the battery successfully charged on the Ryobi charger. All 5 cells are close to a 4.1 nominal voltage and the spread is down to 0.97%. The AC resistance values have also dropped to below the 30 mOhm level, but the spread remains similar to well the battery pack was discharged. It would appear then that the battery pack has been recovered and is potentially in a better condition than the 02-2009 battery pack.
The tabulated data is displayed in some bar graphs for those that prefer a more visual concept. The plots contain an extra set of values for the 02-2009 battery pack that I made to confirm the original readings and that there had been no test measurement errors.
| {gallery} Ryobi Battery Plots |
|---|
Comparison of cell voltage measurements |
Comparison of Cell AC Resistance measurements |
Comparison of Voltage and AC Resistance Differentials |
As with all battery packs, the ultimate proof is a discharge test, so one was carried out on the to older battery packs. A discharge current of 1.4 A was used, which should have given a 1 hour duration for the test.
The 03-2009 battery pack shown in blue, shows a reasonable discharge curve that tails off to the minimum voltage of 14.8 V. This is the voltage that the Ryobi battery pack battery management system (BMS) cuts-off at and is not actually the lowest the cells can go to based on the manufacturer's data, which would have been 12.5 V. Overall the battery lasted 49.5 minutes which is 82.5% of its capacity, quite respectable for a 12 year old battery pack that has been recovered.
The orange line, showing the discharge of the 02-2009 battery pack shows a different story. At the start of the test, a much large voltage drop is seen from the no load voltage and the initial voltage with the load applied. Whilst a Li-Ion cell does drop off more rapidly as it reaches its discharge voltage, the test comes to an abrupt halt at just over 13.5 minutes. The battery pack voltage was 16.3 V, but must have then dropped rapidly below the 14.8 cut-off voltage and the BMS stopped the test. This is only 22.5% of the battery capacity, showing it is in a much worse condition that the other battery pack.
Looking back at the measurements made with the BT3554-01 on the 02-2009, shows that there are three cells with a voltage around 3.9 V, two of those cells have an AC Resistance value around the 50 Ohm mark. This is over the 1.5 times warning value applied to VRLA cells. I decided to carry out a bit more work on the 02-2009 battery pack to see if it could be rejuvenated. A technique often used on a bank of Nickel Cadmium cells, when there is mismatch across their voltage values, is to over charge them for a duration. This is known as a boost or equalising charge. The same technique is also used on VRLA cells during commissioning after they have been installed.
I do not know if this equalising charge would work with Li-Ion cells, but since the battery pack is bordering on useless, I have nothing to loose. The danger with Li-Ion cells is overheating them, so I opted for a low charging current limited to 250mA for 4 hours. This was applied directly across the chain of cells, bypassing the battery management system built into the Ryobi pack. At the end of the charging cycle, the current had dropped down to just over 5 mA.
A sequence of AC Resistance measurements between successive discharge and re-charge cycles was carried out to see if the performance of the battery pack has changed. Initially, the discharge test result was good, with the pack lasting for nearly 30 minutes. Whilst the cell voltage values had risen across all the cells, the differential across them had gone up to just over 9%. The three cells that had been below 4 V, when initially measured, were now up at 4.2 V, but the other two cells, had risen further up towards 4.7 V. AC Resistance wise, there had not been too much change, other than for the very last cell, whose resistance had risen up by 9 mOhms to 40.5 mOhms. As this cell previously had the lowest AC Resistance, this had the effect of lowering the overall differential across the resistance values.
The initial discharge test after the equalising charge showed that there was a significant improvement in the battery packs performance. The battery lasted 29 minutes, up from the initial duration of 13.5 minutes. The question of course, is would this be sustainable, and further recharge and discharge cycles, produced a lower duration now down to around 20 minutes. Although this was still a 45% improvement on the original 13.5 minute discharge test.
A blip was seen on the discharge test after the equalising charge, seen as the blue line in the plot above. The initial application of the load, caused a voltage drop, as it always does, but for some reason the battery voltage then recovered, before going into its usual discharge curve shape. Whilst there is an improvement over the duration, the discharge curve, still has an abrupt stop, with the battery pack never discharging down to the 14.8 V of the other packs.
With regard to the voltage values across the cells, the differential had dropped considerably since the equalising charge, and at 4.06%, this was even lower than the 5.74% differential when the battery pack was first tested.
For the AC Resistance values, the plot shows a very slow increase in the differential as subsequent test cycles are completed. No real difference is seen with Cells 2 and 3, that were the cells that identified with a high AC Resistance. Cell 5 now seems to have been permanently damaged by the equalising charge, now showing a higher, but stable, AC Resistance from its original test.
It will be interesting to see how the battery pack continues to perform. My suspicion is that it will relatively quickly deteriorate back to the original test results. It may be that a slightly lower equalising charge, probably at 21 V instead of 22, would be sufficient to improve the battery, but not damage the cells that were in better condition.
NiMH AA Cells
I have a batch of Nickel-Metal Hydride AA cells that I use for instruments and wireless keyboards and mice, and decided to give some of these the same treatment as the 18650 cells. They are of various different makes and ratings. The youngest cells are a batch from EBL, I purchased back in November and have 4 discharge/charge cycles accredited to them. The other main batch are Energizer cells and are in the region of 5 to 6 years old, I have no idea how many cycles they will have undergone. Thrown in between these are a set of FreeLoader cells are purchased with a minim USB charger, these are 3 to 4 years old, again the number of cycles is unknown.
Manufacturer information on these cells is also hard to come by. Only Energizer produce specific information on AC Resistance measurements and capacity. The cells appear to be tested at a nominal 0.2C and then the data extrapolated to give a 1C rating. They do provide a nominal 12 mOhm figure for impedance measured at 1 kHz. However, it does specify a test current of 1A for this measurement, where as the BT3554-01 will only provide a maximum of 160 mA.
I carried out a number of discharge tests, measuring AC Resistance prior to the test, after the test and then after the recharge. All of the cells were charged on a universal charger from EBL. All cells were charged prior to the initial test and measurement.
The Energizer cells being the older, quite clearly had the higher readings and none were even close to the nominal 12 mOhm, although there seemed to be a good spread across the values for testing purposes. The EBL cells showed the lowest AC Resistance at the start of the tests, which was to be expected and the FreeLoader cell sat in between the EBL and Energizer cells. The EBL cells were twice the expectation of the Energizer cells but I do not have any information to say that the 12 mOhm is also applicable to other manufacturers cells.
All of the cells failed to reach their potential discharge rates. On top of that the Energizer cells with the high AC Resistance values could not be tested at 1C. At that load, the cell voltage instantly dropped down to 0.5 V as the load was applied, I didn't even get time to switch the voltage logging function on. The cells with the lower AC Resistance values could take a higher test current.
Some correlation can be seen between battery capacity and the AC Resistance, with the EBL cells with their lower AC Resistance, predominantly showing a better capacity, especially when tested at 0.2C. I am not sure that the practice of testing at a lower capacity and then extrapolating the data out to give a rating is particularly accurate, given the age of these cells. I may choose to get hold of a new set of Energizer cells at some point and test them at 0.2C, to see if the results improve.
Below are the discharge curves for the cells where a 0.2C discharge rate was applied. As can be seen none of them reached the 5 hours (300 minutes) specification. The EBL cell with the lowest AC Resistance shows a more gradual discharge. The two Equaliser cells with the high AC Resistance show a much higher initial voltage drop, due to a combination of the higher AC Resistance and the load current. They then show a more flat response until the end of the test.
The FreeLoader cell, has a low AC Resistance and so shows the small initial voltage drop similar to the EBL. It is then quite flat, until a sudden drop-off, that actually dropped all the way down to zero without giving me a chance to switch the load off. I think that this likely to be due more to the quality of the cell manufacture and is a bit of a red herring within the results.
Interestingly, all the cells that had a very high AC Resistance after charging, all saw a drop in their AC Resistance values after the discharge test. This is opposite to the VRLA and 18650, that all saw an increase in AC Resistance when discharged.
When the cells were recharged, the AC Resistance increase again, but not up to the same levels as previously seen. It could be that these older cells are suffering from only being partially discharged and a full discharge and recharge cycle has improved their state of health. The Freeloader and EBL cells, saw very little change across their AC Resistance values irrespective of their state of charge.
It would seem that the NiMH technology does respond to AC Resistance measurements, although the information from manufacturers to verify the readings does not seem to be as progressive as it does for other cell technologies.
Conclusions
This has been an interesting couple of experiments for me. The tests on the VRLA battery have been predictable and show the importance of carrying out AC Resistance measurements on VRLA cells when they are on float charge.
Te Li-Ion cells tests were more of a learning experience. Previously, I have been led to believe that AC Resistance measurements on Li-Ion cells are not as effective as DC Resistance testing. The state of health curve of a Li-Ion battery shows that the AC Resistance over the life time of a Li-Ion cell stays relatively flat until the very end of its life, when it will rise quickly. This gives less opportunity to detect a failing cell, which is the intention of AC Resistance measurements. Whilst I have picked up faulty cells within the Ryobi Battery Pack 02-2009, using the BT3554-01, it looks like those cells have pretty much reached the end of their life and that was obvious from the discharge test prior to carrying out the measurements.
NiMH cells were also seen to respond to AC Resistance measurements, with those cells showing an initial high AC Resistance, showing a poor performance during a discharge test and in some cases being unable to sustain full rated current outputs, due to the high AC Resistance values measured with the BT3554-01.
The BT3554-01 has proved that it can identify the faulty cells in the pack, which is its ultimate goal and should in theory allow the battery pack life to be extended cheaply by just replacing the faulty cells. The downside of course is that the battery pack has to be stripped down to get to the individual cell terminals for this type of testing, which isn't always possible.
Top Comments