Introduction
This aspect of the RoadTest is to test the Hioki BT3554-01 in a real world scenario, that the instrument is specifically designed for. At one of the sites I provide engineering assistance to, there are banks of sealed lead acid batteries that provide backup power to essential systems. I will take AC Resistance measurements from these banks with the intention of comparing the performance of the BT3554-01 to the 3550 HiTester, in terms of its ergonomics and practicality.
Using the measurement data captured, I will aim to verify the theory applied to AC Resistance measurements, with a failed VRLA battery showing a value twice that of its nominal value. I will then look to see how the data can be utilised to salvage cells from failed battery banks, in the hope of re-utilising them in other systems.
This blog is part of my RoadTest review of the Hioki BT3554-01 Battery Tester.
Hioki BT3554 Battery Tester Review
Emergency DC Battery Systems
All of the systems are 110V DC. The larger systems are made up of 55 cells, each with a nominal 2.1 V, giving an overall voltage of 115 V. In practice the cells under float charge, are around 2.2 V taking the overall voltage up to 120 V.
This system is designed to power emergency lube oil pumps and spray water systems, alongside protection and control systems, so that a turbine can be brought safely down to slow speed and slowly rotated to prevent the turbine shaft from bending in the event of a power loss to the site.
There are three of these systems, each with two battery banks. They range from 9 to 12 years old, two of the banks currently meet the discharge test requirements and the other four do not. I am looking for the BT3554-01 to determine if there is a trend between the resistance of individual cells that failed or passed the discharge test.
One of these banks is to be replaced this year and the good cells from this bank used to refurbish the other banks, so I want to use the data from the tests to determine the preferred cells to be retained.
These particular cells actually have 4 terminals on them ( 2 of each polarity). The readings with the BT3554-01 will only be taken from one pair of terminals.
Other systems are much smaller and designed to provide emergency supplies to protection systems and switchgear control. This is made up of a combination of 6V and 4V cells, again to give an overall voltage of 120 V.
As can be seen from the picture, the cells are in a stacked arrangement that limits the access space to the terminals of the cells on the bottom shelf. This is exactly what the right-angled test probes from Hioki have been designed to assist with.
One of these battery banks is over 12 years old and the other is 2 years old. I hope to pick up this age discrepancy with the BT3554-01, again I have discharge data for both of the banks for further comparison.
Below is a block diagram of one of the battery systems. These particular systems have a dual battery bank that provides DC power into a common DC distribution system. For redundancy, each battery bank has its own charger.
I do not actually have a nominal AC resistance values for these cells. In their data sheet, the manufacturer only provides nominal DC resistance values, circled on the data table below. The AC values will be expected to be below this.
I did try to verify this resistance value with the data from the discharge tests using the no load and initial load values obtained. This however, did not work out, as the initial load value during the discharge test is actually taken 30 minutes after the load has been applied.
Taking Measurements
Prior to taking the measurements from the cells, I have set up four comparator sets that will be activated for the specific cell type being measured. For the larger battery systems, this is easy enough as they consist of only one type of cell. For the smaller battery systems, I will have to change the comparator set during the measurements to reflect the correct cell type.
The comparator sets are as follows;
Post reading analysis has shown that these comparator sets are not ideal and will need to be adjusted as further testing is carried out.
I took the readings utilising the Autohold and AutoSave functions enabled. Due to the way the memory is configured on the BT3554-01, it could be set up to use a different memory set for each of the battery banks. All I had to do is remember, which memory set was allocated to which battery bank, the safest option for me being to write that down.
Here we see one of the improvements in the BT3554-01 making the job easier. The 3550 HiTester only had one memory set of 260 records. So whilst I could measure and record for multiple battery banks, I could not get around the whole site in one go. I would have to go back to the office after measuring 4 battery banks and download those results, and then return to the site to complete the readings on the remaining battery banks.
Throughout the measurements, the Autohold and Autosave functions performed flawlessly and with the comparator set up, instant feedback is received on the cell condition. The memory function also has the ability to delete an individual reading and take it again, if an error is made during the measurement. The instrument will then pickup again at the next available memory slot.
There are not really any issues with actually taking the measurements, after all, this is really just akin to taking a voltage measurement with a multimeter.
All the terminals are shielded on these particular cells with a protective cap. However, there is a small test hole in each cap that the probes will fit through and make contact with the terminal underneath and take the measurement.
Both the right-angled and the straight probes can be utilised to take readings on the larger battery systems as seen above. Some of the connections, where a cable connects and not a copper strap, are protected by a boot and the probes were more awkward to use on theses connections. The boot can easily be moved to ones side to expose the cone action to take the readings, but to do this, the straight probes had to be used.
For measurements on the lower shelf of cells for the smaller batteries, only the right-angled test probes could gain the correct access. With the straight probes, I could not get onto the terminals at the correct angle, in order to take a stable reading.
For both sets of battery types, the probe approach of the new BT-3554-01, offers a safety enhancement in that the termination caps do not have to be removed to take the measurements. This leaves the live terminals with their IP2X protection intact. Using the older 3550 HiTester with its Kelvin clip arrangement, the caps have to be removed to gain full access to the terminations. It does mean that the Kelvin clips can be used on any type and installation configuration, where as sometimes the straight probes have to be swapped for the right-angled ones and vice-versa.
As well as the safety improvements, a significant amount of time is taken to remove and then refit the caps, especially for the larger cells.
Admittedly, there is the option of upgrading the old meter to use the probes instead of the Kelvin clips. An external hold button would also need to be configured, and as there is no Autosave function on the 3550 HiTester, each measurement would still have to be saved manually.
The final element to the task is the uploading of the data to a computer. For the 3550 HiTester, this is a manual process of scrolling through the readings and manually entering them into a spreadsheet program. Not only is this s slow process, there is also an increased risk of incorrect data entry.
The BT-3554-01 addresses these issues with the use of the Gennect One software, which is currently only available for Windows based computers. For iOS, there is the Gennect Cross software. Both programs are free to download and install and there are no restrictions for its use.
The Windows based software uses a USB connection and is only capable of downloading data stored within the memory of the BT3554-01. The iOS uses a bluetooth connection, so requires this option to be installed on the BT3554-01. This software allows not only saved data to be downloaded, but also gives a live connection to the instrument, so data can be saved during testing. Personally, this is something I am not too comfortable with. Working on battery banks is a live working scenario and I want to keep my concentration on the dangers I am exposed to, and do not want to be messing around with a bluetooth connection making sure that the data is being recorded correctly.
The following table shows the measurement statistics of the two instruments for two of the battery banks, one large and one small.
BTB12 is one of the smaller battery systems. Using the old instrument, the caps over the connections are removed and replaced during the measurements. As can be seen, the time taken to manually download the instrument in comparison to the BT3554-01 is much longer. Overall, taking measurements on the small battery system with the BT3554-01 shows a 61% time improvement, a significant saving.
BTC12, the larger battery system, shows a similar pattern. Here I have split out the time taken to remove and replace the caps. Even with this function separated, it still takes 10 more minutes to take the readings with the old instrument and the Kelvin clips against the new instrument. Download time is significantly longer. On this battery bank, a 76% improvement in the time taken was seen, another significant improvement.
In terms of measurement accuracy, the bottom line in the tables shows the average differential of the 3550 HiTester to the BT3554-01, for all of the readings taken for that particular battery. The voltage reading shows a much lower differential than the resistance reading.
There is quite a marked difference in the AC Resistance readings, with the readings for BTB12 showing the HiTester was reading lower than the BT3554-01, that was then reversed for BTC12, which shows that the HiTester was showing higher resistance values in comparison. This difference is greater than the accuracy specifications of the instruments. The cell type and the way it reacts to the test signal may also be a contributing factor.
Absolute accuracy is not paramount with battery testing, but consistency is. This is a lesson for me to be careful about battery monitoring over a period of time. If subsequent tests are carried out by different contractors with different instrumentation, there may be an issue with trending the results supplied.
Test Results
To analyse the test results, the values obtained from the BT3554-01, are compared to the results from the 3 hour discharge tests that were carried out in November last year.
With all the results stored on the BT3554-01, they are all downloaded to the computer at the same time. As they were saved in different memory sets, I end up with a download entry for each battery.
| {gallery} Gennect Software |
|---|
Download screen showing memory slots |
Editing data in the main database screen |
When initially downloaded, the database entries only contain time and date information. Each test can be titled and comments added to it from the database screen to aid in future identification. Search tags can also be added for when the database becomes much larger and records are more awkward to find.
Analysis of both individual and multiple datasets are available within the software. The data to be analysed is selected using the check boxes and are then opened within a new analysis window.
For individual analysis, there is a basic data screen displaying the measurements. On this screen, individual comments can be added for each reading if desired. A summary is displayed at the top of this list, detailing how many cells have passed, have a warning, or have a failed reading.
The next tab displays the same data, but this time in a graph format. The graph data can be selected between, voltage, resistance and temperature, if it has been recorded. A cursor function is available that displays the relevant reading at the top of the graph as the cursor is moved.
The next tab displays the comparator threshold data selected for each of the readings. It does not appear that any data on this tab can be edited, but from my perspective, it would have been nice to have been able to fine tune these values, to provide better analysis of the measurements.
Further analysis can be carried out in the next tab labelled ‘Trend’. Primarily, this is designed for viewing the trend of cells over a period of time, which would require multiple tests of the same battery to be held with the database.
Since I do not have that amount of data in, I will use it to trend one of the old battery banks to a newer battery bank.
In order to utilise this function, you have to go back to the main database screen and select the relevant records from the database list.
Having selected the multiple records, the individual cell measurements are added to the trend plot by selecting them in the data list screen, right clicking and then selecting the ‘Add Trend(A)’ function that pops up on the screen.
The trend display is initially then setup for all the readings selected. You can see in the far left pane that both BTB11 and BTB12 are available.
Beside the trend plot are all the individual readings available to be displayed. Due to the way in which the data was saved whilst taking the measurements, the label ‘Data No. 1’ corresponds to Cell 1 out on plant, with ‘Data No. 2’ corresponding to Cell 2 and so on.
The graph is a bit busy, but the check boxes next to the data labels allows data points to be enabled and disabled as desired.
In the image above, there are only about half of the cells displayed, leading to a much clearer picture.
The trend graph shows what was expected. The cells from the newer battery, BTB12 on the left, all show a resistance less than the cells from BTB11 on the right that is around 12 years old now and due for replacement. This isn’t immediately obvious from the trend graph, because the individual batteries are not easily identified. In fairness, this isn’t the purpose of the trend graph, it is really only designed to display data from one battery over a period of time, which it does perfectly well.
The software provides the ability to assemble all of the data screens into a single PDF report. Pictures can also be added into the report, but first have to be imported into the Gennect software via the file import function. I did have some issues with this function, as it did not want to import all files I tried, coming back with a bad format error. I presume that this is down to the camera, or the editing software used to resize the image. I have attached an example PDF report to the blog, for those that are interested in seeing one.
One final note on the software is that it does not appear to provide any control functionality over the BT3554. Once the data has been extracted from the instrument, I then had to delete all the data from within the instrument itself, which is easy enough to do, but the option to download and then delete is not available, as it is in some manufacturer’s software.
Analysis in Excel
In order to compare the AC resistance values to the discharge test results, I need to get all the data into a spreadsheet and the BT3554 offers two methods to get the data into Excel. The first is the option to import direct into a CSV file instead of importing the data into the Gennect database. This is the option that I tend to use and have had no problems with it. The second is to export the data from the Gennect software once it has been saved to the database.
The particular battery banks I have tested have an annual 3 hour discharge test in accordance with the British Standard and the manufacturer's specification. To do this test, the battery bank is disconnected from the plant and reconnected to a constant current source. The battery bank passes the discharge test if none of the cells drop below 1.80 V within the 3 hours. If a cell does drop below this, then the discharge test is stopped.
If a battery bank is less than half-life, then a new cell can be installed in place of the cell(s) that fails the test. After a battery bank has aged beyond half-life, it becomes more tricky. Where I have multiple banks of the same cell type, it is possible to replace one complete battery bank with new cells and then use the best of the cells from this bank to replace the worst cells in the remaining banks. This is the situation I am in with the four battery banks comprising of the DDm 85-21 cells, so I hope to use the AC resistance values to help decide on the best cells to retain.
I have taken the AC resistance values and compared them to the voltage value at the end of the 3 hours discharge test. In theory, the cells with the lowest discharge voltage would be expected to be in the worst condition and therefore should have the highest AC resistance values. This is backed up by the trend graph of the two BTB battery banks where the 12 year old bank has higher AC resistance values than the 2 year old bank.
Overall analysis, shown in the table below, has revealed that this theory is not clear cut. For all the battery banks, the cell with the lowest discharge voltage is not the cell with the highest impedance. Looking a the opposite end, it is also clear that the cell with the highest discharge voltage is not the cell with the lowest impedance.
The table below is for BTC21 battery, that failed its discharge test. The base data columns are the data listed by cell number. In the centre columns, the data has been reordered with the resistance from higher to lowest and in the last four columns, it has reordered with eh discharge voltage from lowest up to highest. Cell 5 is seen as having the lowest discharge voltage and was actually the cell that stopped the test. In terms of impedance, it does not have the highest, as expected, but it is very close.
Looking at the other end of the chart, there are two cells with the highest discharge voltage of 1.93 V and on the face of this, would appear to be in better condition than all the others. This is backed up by cell 3, that has one of the lowest AC resistance values. However, for cell 38, whilst it has a high discharge value, it has one of the highest AC resistance values, which does not correlate to what was expected.
Overall the spread across the AC resistance values is relatively low, showing that the cells have all aged and probably don't have too much life left in them.
Taking a look at the battery that passed the discharge test. This battery is 9 years old, so is slightly younger. The range of resistance readings, is slightly larger than that of the previous battery, which would not have been expected. Laying out the results in the same manner as the first battery, reveals and even less clear status. The one cell that has the lowest discharge voltage only just creeps into the upper quartile of AC resistance values.
Again there are two cells that share the highest discharge voltage of 1.93 V. In terms of their AC resistance values, they both appear to have relatively high values and are not the lowest as would be expected. As with the previous battery, one of the cells that responded best to the discharge test, responds as one of the worst for the AC resistance reading.
The following table summarises the findings.
Comparing BTB11, the 12 year old failed battery to BTB12, the 2 year old battery, shows that it has an average AC resistance of 1.25 times BTB12. Whilst the actual differential would have been expected to be higher, this is consistent with the theory of the testing.
For the other battery banks, that are roughly the same age. There is no real meaningful differential between those banks that pass the discharge test and those that fail it. This is likely to be due to the similar age of the cells and it would appear that I may be expecting too much from the AC resistance data in terms of identifying the best cells from an ageing bank.
Effects of Bolted Connections on AC Resistance
I will add this short experiment in here, as it is an issue I have come across when carrying out AC Measurements on VRLA battery systems. There are two main methods of connecting to the terminals of the cells. The first is a standard bolt connection, that then has an insulating cap over the connection, or a complete lid over the cell. The insulating covers usually have a small hole above the bolt head to give access for the test probes. This is the connection that has been utilised on the battery banks I have tested for this blog.
| {gallery} Bolted connection arrangements |
|---|
Standard bolt connection on a cell |
Insulated bolt connections on cells |
The second method is a bolt with an insulated head and an insulated connection strap. The insulated head of the bolt is permanently fixed into position. Versions are available with or without the test point. The first issue is that the test point is not big enough for concentric probes, but this is easily resolvable by drilling out the plastic head with the appropriately sized drill bit. The second issue is that the test point under the plastic head is actually a small lead ball inserted into the steel head of the bolt, and this creates a number of issues for low resistance measurements.
The first is that there is a dissimilar metal connection between the lead and the steel, that adds resistance to the joint. The second is that the lead ball can become dislodged under the cap and leads to high and erratic readings. This occurs over time as the probes from multimeters and impedance testers are pressed into the test point to take readings. The third is that the lead is soft and the concentric probes are almost teeth like, this can churn up the lead ball and give erratic readings, but you can also end up with flakes of the lead jammed into the probe tips.
To show this is, I set up a little plate with two M10 bolted connections and took a few readings using different kinds of bolts.
The table below details the results of the measurements. Testing straight from the straps of the test block gives a nominal reading of 12.92 V and 79.2 mOhms. Fitting two standard bolts and torquing them both down to 15 Nm, and repeating the test gives an extra 0.2 mOhms to the AC Resistance reading. This can be lived with as it does not affect the overall resistance in this scenario.
Removing one of the steel bolts and installing the red insulated bolt takes the reading up to 83.7 mOhms with the voltage remaining as 12.92 V. Again. this is not so much of an issue and if it was consistent, could be catered for by the zeroing function of the instrument. If you have a cell with a much lower AC Resistance value, this could start to become an issue though. A few other insulated bolts are then tested all with readings above 400 mOhms, which is way above the initial reading of the battery and may lead someone to thing the battery is faulty. Again, the voltage reading remains unaffected, staying at 12.92 V.
I finalised the test by re-testing with two steel bolts and then replaced both steel bolts with two insulated ones. This returned readings of 78.7 mOhms and 876 mOhm respectively. It should be noted that this issue exists with all AC Impedance measurements and is not just an issue with the Hioki Instrument or its operation.
After having a think, I wondered what effect of drilling out the lead ball within the bolts would have on the readings. This is easily achieved with a 3 mm drill bit. The readings obtained then dropped down to around the 80 mOhm level and inline with measurements when using standard bolts. This reading was also stable and did not have the erratic nature seen before. This therefore, could be a solution to installation where these insulated bolts have been used and the AC Resistance readings are erratic.
Conclusions
Whilst the data captured from the cells has not been able to agree completely with the theory for the measurements, nor has it been decisive in determining the best cells to salvage from the banks, the performance of the BT3554-01 over the older 3550 HiTester has been substantial.
The ease of the right-angled probes to gain access to battery cells on the lower stacks improves both the safety, speed and the overall convenience of taking measurements is quite clear to see. A average 68% increase in performance is substantial for larger battery banks and is even more worth while in locations where multiple battery banks are located. The improved memory allows for measurements to be made on up to 12 different battery banks before needing to be downloaded, where as the old 3550 HiTester was really restricted to 4 battery banks.
The Auto Hold / Auto Save function worked very well during the measurements. I never had any issues with missed readings.
An overall slight misgiving of the system is that the comparator function must be set up first on the instrument to be able to utilise it within the software. The comparator must then be enabled during the measurements. There is no option to add or modify comparator thresholds within the software once results have been downloaded, which would have been beneficial.
Under this work scenario, the BT3554-01 certainly offers multiple advantages over the older 3550 HiTester.
 | BTB12 Hioki Report Trend PDF.pdf |
 | BTC31 Hioki Report PDF.pdf |
Top Comments