RoadTest: Hioki BT3554 Battery Tester
Author: three-phase
Creation date:
Evaluation Type: Test Equipment
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: The BT3554 was compared directly to its predecessor, the 3550 HiTester. A market review was carried out against testers from 6 other competitors.
What were the biggest problems encountered?: There were no real problems encountered whilst reviewing or using the instrument that were directly attributed to it. I had issues carrying out calibration checks at low ranges, but these were due to the design of the decade boxes I possess.
Detailed Review:
This is my review of the Hioki BT3554-01 portable battery tester and will cover the concept of battery AC resistance (impedance) measurement alongside initial testing with the instrument of lithium-ion 18650 cells and a 12 V Valve Regulated Lead Acid (VRLA) battery. Following on from this I will carry out testing on some VRLA emergency battery banks with differing ages to compare the AC resistance readings with the discharge test results and investigate how the different measurement techniques compliment one another.
The BT3554-01 is specified as a Battery Tester and measures 'AC Resistance'. Unfortunately, I am a bit old school and was brought up on Megger test apparatus, where it was defined as an impedance measurement of a cell. To keep in line with Hioki, I will refer to it as an AC Resistance measurement. However, there may be occasions where the 'impedance' gets mentioned, when I meant to say 'AC Resistance'. For the purposes of my blogs, 'AC Resistance measurement' and 'impedance measurement' are interchangeable.
The BT-3554-01 and all its accessories arrived packaged in a plain brown cardboard box. For this RoadTest, the L2020 right-angled pin type leads arrived in a separate box.
I believe that if the instrument is actually purchased, it arrives with either the L2020 right angle or 9465-10 straight pin type leads and not both as it did with the RoadTest item.
The instrument is housed in a sturdy plastic instrument case. Opening the case up reveals the instrument, AA battery set, spare fuse, 9465-10 test leads, USB lead and strap all in the bottom compartment and the zero adjust board, manual and CD ROM stored in the bottom compartment.
| {gallery} BT3554 Kit Contents |
|---|
Battery Tester Case |
Battery Tester Case Opened |
BT3554-01 and Accessories |
The instrument itself has a blue protective rubber boot surrounding it. Whilst it is removable, the boot has been designed to allow access to both the battery and fuse compartments of the instrument with it still in place. These compartments each have their own access lids to allow them to be replaced without affecting the integrity of the calibration of the unit. The lid for the fuse compartment is held in place with a single screw, where as the battery compartment lid is a clip in style.
| {gallery} BT3554-01 |
|---|
Front View of Instrument |
Rear View of Instrument |
Top View of Instrument |
The LCD screen, main test lead connections and function control buttons are all immediately accessible on the front of the instrument. There is a further cutout in the rubber boot for the top of the instrument where the sockets for the USB, temperature sensor and remote hold function can be accessed.
Both sets of test leads are the concentric pin type leads. Each pin has a protective cap that clips over it to protect it when they are not in use. Spare pins are available as an extra if required. Further accessories are available in the form of a set of duplex pin test leads, kelvin clip style test leads with a built in temperature probe, large kelvin type test leads and a remote hold trigger.
The close up picture above of the concentric pin shows the surface detail, designed to make a good contact onto the cell terminals.
| {gallery} Available Accessories |
|---|
Test Leads Option 1 |
Test Leads Option 2 |
Test Leads Option 3 |
All of the accessories carry a fairly hefty price tag with them. The test leads range from £138 for the small croc clip lead set up to £253 for the right-angled pin type lead.
Link to Hioki accessories available from UK.
The meter its self ranges from £1,658 to £1,734 in the UK, across 4 different models.
Link to Battery Testers available from Farnell.
However, it would appear that whilst carrying out this Roadtest review, the current versions of the BT3554-01 and BT3554-11 are being phased out and replaced with the BT3554-50 series.
Link to BT3554-50 Series of Battery Testers.
A brief review of the manual seems to suggest that the instruments are technically the same and the vast majority of the work has been done on on the speed to the measurements and the user interface, expanding the capabilities of the connection to mobile devices. A battery profile system has also been added to the software / firmware that allows identification of measurements to a battery system rather than just memory locations. There is also the option to link a comparator setting to this profile.
A separate temperature probe now seems to be available, that allows temperature data to be recorded when using the probes and not just the Kelvin clips with the embedded temperature probe. The bluetooth connection is now an add in module, meaning any instrument in the BT3554-50 range can be upgraded for use with a mobile device at a later date.
As one of the main purposes of my RoadTest is to compare the BT-3554 to the original 3550 HiTester, this point in the blog seems like a good opportunity to bring the two instruments side by side.
The 3550 belongs to the company I work for and is between 15 to 20 years old. I could not find the price for the old unit. At the time of purchase, the unit came with the small crocodile clips test leads with the temperature probe as that was all that was available at the time. No upgrades have been purchased to enhance its use.
Whilst the units are around the same size, the 3550 has no protective boot and is designed to be used whilst still within the case it is provided with. The 3550 has the same functional layout as the BT-3554, with the main test terminals and control buttons on the front face and the auxiliary ports on the top, accessed under small flaps.
This is where the first notable difference is between the units, whilst the temperature and remote probe connections are shared between the units, the 3550 has no USB port. Instead there is a bespoke connector for output to a thermal printer.
Whilst the temperature probe socket is the same, the remote hold socket is different on the 3550 and would not accept the remote hold button currently sold by Hioki. The old crocodile test set with the temperature probe though, is compatible between the two units.
As well as some cosmetic difference, Hioki have done a lot of work on improving the functionality of BT3554 over the 3550.
The table above summarises the functionality of the two units. The resistance function has been improved, now covering 4 ranges from 1 µΩ to 3.1 Ω, with the accuracy improving from ±1.2% to ±0.8%. The voltage is still across two ranges, but has been extended from 1 mV to 30 V up to 1 mV to 60 V. Again an increase in accuracy is seen from ±0.15% to ±0.08%.
Whilst both units have a zero ohms adjustment, only the BT3554 comes with the calibration board.
A significant improvement is seen in the memory functionality. The number of records has increased significantly and where as the 3550 only had one set of 260 records, the BT3554 has its 6000 records split across 12 sets giving much more versatility. The new unit also has an auto save function in addition to the manual and remote buttons. PC connection via USB comes as standard and tablet / phone connection via bluetooth is available as an option.
Finally, the screen has a backlight added to it, something that was missing from the old one.
The video below goes through the differences between the two instruments.
There are now a number of comparable units available from competitors as alternatives to the BT3554 from Hioki. I do also own an Applent 528 AC Ohmmeter, specifically purchased to test batteries within protection relays, that exceed the 3 Ω limit of the Hioki instruments.
The table below gives an overview of some of the units available.
In terms of functionality, the Fluke BT521 looks to be quite a way ahead, but this does come at a price. There are cheaper models available from Fluke without the temperature and current measurement functions and without the intelligent probes. The intelligent probes add a small readout to the top of one of them along with the temperature probe built in. Whilst unique, the probe design does not allow access to cells that are installed in racks stacked on one another. Only Hioki have the right angled probes designed to accommodate this setup.
The Tenmar unit is the cheapest of the bunch, but still has a reasonable level of functionality. It does have a lower level of accuracy in comparison to the other units available. The unit from Chauvin Arnoux has comparable accuracy to the Tenmar, but not all of its functionality. Given the price of it, I am not sure it would be a viable alternative. It is also available as a cloned device, from Extech as the BT100 and Amprobe as the BAT-500.
The Applent AT528 has the largest impedance measurement capability of all of the units. It is available as the AT528L, with a reduced impedance range and consequently cheaper. Whilst it does have a comparator built in, the data hold and save functions are totally manual and is therefore more awkward to use with probes. It does however, come with kelvin clips as well as the probe set.
The unit from SBS is primarily only available within the American market. Whilst it has similar functionality and accuracy to the other units, its main draw is a large database of comparable impedance values from multiple manufacturers to compare to the readings taken. It does come at quite a hefty price though.
Whilst the Megger Bite 3 unit looks to be priced out of the comparison, there is a little bit more to it. The unit has a built in harmonic analyser and none of the other units possess this or have the functionality within their software. It is also the only unit that has a function built in to measure the cell interconnecting strap resistance. Megger has a slightly different design, using a lower test frequency than all the other manufacturers and so does not produce comparable readings.
The Hioki stands out really well against these other offerings. Whilst the Fluke unit has more functionality, adding a clamp meter to the Hioki BT3554 would produce a comparable setup and would still be more cost effective.
There are also extremely cheap battery impedance meters available on the market, usually via Aliexpress or eBay. The most common unit is the SM8124. Whilst selling for around £50 in the UK, it is a fraction of the price of these other units, you get what you pay for. Whilst I would never recommend the unit for more professional use that I will put it through, it is probably ideal for a hobbyist who wants to carry out simple measurements on a few battery cells.
To some, the cost of these battery testers may seem to be excessive, so I will put some context around it based upon my work. The objective of AC Resistance measurements on cells is to determine a cell in a bad state of health within a bank and get it replaced before either the bad cell starts to affect the health of the cells surrounding it, or there is a main supply failure and the battery bank then fails to support the load for the required amount of time.
The main battery banks that I work on, range from 110 V to 220 V overall voltage and will be made up of 55 or 105 individual cells respectively. For the four battery banks that will be one of my main focus points for the testing, each bank costs around £25,000 to replace. If I can use the testing to determine bad cells early and extend the life of the bank instead of getting into a replacement scenario, then spending a few thousand pounds on a tester is a good investment.
This isn't taking into account the affect of a battery back up failure when required. Re-metalling a white metal bearing because a DC lube oil pump failed will be in the region of £5,000. But it takes two to three weeks to mobile crews and get the actual bearing removed, sent away, repaired, returned to site and then re-installed. Even if a spare bearing is available on site, you may still take up to a week to replace it and may be longer if it is in a more restricted area. As well as the cost of the work, the down time and loss of production, will also incur significant costs.
Large data centres that utilise these type of battery banks face a similar issue with their losses incurred due to not providing services to customers. Some industries may also face regulatory fines if they fail to recover from issues with prescribed times.
The alternative of a discharge test exists. For the one battery bank mentioned above, the test is around £2,000 and takes all day including disconnection and reconnection times. However, the battery then has to be recharged. For large VRLA battery banks, recharge to 95% of capacity takes around 10 to 12 hours, it will take a further 5 to 6 days to reach the full capacity. Issues also arise if a cell fails during a test. IEEE standards, allow a discharge test to be stopped for 15 minutes to bypass a faulty cell. However, if this is not achieved, or possible, the test is aborted. The bad cell would have to be replaced and then another test carried out, to ensure there were no other bad cells within the bank.
For these large installations, battery testing using AC resistance measurements is a normal routine activity for preventative maintenance schemes and a viable alternative to discharge testing.
Battery AC resistance is one of the number of ways that the condition of a battery can be assessed and monitored over its life time. Whilst an AC resistance value can be obtained for all batteries, the technology works best with lead acid batteries and is a well established practice for large battery banks utilised for emergency backup supplies in process industries and data facilities.
Table of main methods of battery condition assessment
For a long while the main methods for monitoring battery banks would be via measuring and recording the voltage, temperature and specific gravity of the cells. This was backed up with routine discharge testing of the bank. However, whilst the basic measurements can be made with the battery bank in service, a discharge test requires the battery bank to be removed from service, disconnected and connected to a constant current load for a specific time. This can be up to 10 hours, but can be reduced if a powerful enough load bank is available.
DC resistance measurement is predominantly a method used in the automotive industry to test starter batteries for combustion engines. In theory the battery can be on load during the test, but it should not be on charge. As the DC Resistance test measures all aspects of the battery that form the DC Resistance, it has its limitations for condition monitoring purposes on lead acid cells.
This leads on to AC Resistance, or impedance, measurement for determining the condition of the cells. This is a methodology that can be used with the cell either online or offline, but it should be fully charged. Generally, it is carried out with the battery bank on float charge.
To understand the concept of what the AC Resistance measurement achieves, we need to look at the equivalent electrical circuit for a lead acid cell, which is usually depicted by Randles Model.
The main charge of the cell is contained within CBAT. This charge will decay on its own through internal leakage resistance represented by RD. As the cell has internal conductors and connectors, they will have their own resistance, shown as RL in the model. The electrodes within the cell, have capacitance and resistance where they interact with the electrolyte, represented by CPL and RCT respectively. It is this value of RCT that is the most helpful in determining that health of a lead acid cell. There is a small inductive element within the cell, however, it is considered to be so small that it is often omitted from the model.
A standard DC test will measure, RL and RCT together. However, when an AC signal is applied to the cell, this has the effect of lowering the impedance of CPL and effectively shorting out RCT, leaving a value for RL.
Ideally, if a DC resistance test is also carried out, an individual value for RCT can be obtained by subtracting the AC Resistance value from the DC Resistance value.
The following picture shows Randles Model applied to the construction of a lead acid cell.
Naturally, in an actual cell, there are multiple positive and negative plates, but these have been omitted for clarity.
This is where the Hioki BT3554 sits in terms of battery testing. As far as I know, it is a straight impedance meter, testing at 1 kHz, so the values it displays are more representative of RL rather than RCT.
An extension of AC Resistance measurement is impedance spectroscopy, where the AC Resistance is measured over a spectrum of frequencies, typically from 10 Hz to over 100 KHz. This is a highly sensitive technique and whilst there is some field based test apparatus, it is predominantly a laboratory based research technique. It is not something that I have seen carried out on a site.
The final test method, for a battery banks is the discharge test method. This is another old school method for testing that the capacity of the battery bank is acceptable. It is the ultimate proof that the battery will perform.
Whilst it is routinely carried out, it does require the battery to be out of service and connected to a load bank for the duration of the test. It must then be recharged, before it is effectively back in service. Below is a picture from a Megger webinar that depicts a discharge test being carried out on a battery bank. This is the automated version of the test, not the number of connections across the top of the battery bank, each one have a transmitter that sends the voltage to the computer during the test. The load bank is just by the side of the box that the computer sits on.
This is a relatively small battery bank, for larger battery banks a number of load banks can be wired inparallel to obtain the required discharge current, or the test can be carried out for an extended duration.
Measuring AC Resistance offers a compromise between the simple methodologies of voltage and temperature measurement, that offer limited assessment data and an expensive and time consuming discharge test, that offers ultimate proof of battery bank capacity. It works well with VRLA technology due to the way the internal resistance behaves as a VRLA cell ages. Below is a pictorial representation of internal resistance plots for different cell technologies against their state of health (SOH), which is the numerical valuation of the cell condition. Note that the diagram is a representation and the plots for the different cell technologies have been separated for clarity, but in reality would likely overlap.
Over the life time of a VRLA cell, displayed in blue on the plot, there is a steady rise in the internal resistance that gets towards double the initial value when the cell reaches end of life. For a Li-Ion cell, the rise in internal resistance dose not occur until the cell is already near the end of its life, this limits the effectiveness of the measurement for this technology. For Nickel Cadmium cells, there is not as much rise, and again what does occur, tends to be more towards the end of life. NiMH technology also responds well to internal resistance measurements. A rise will be detected as the cell passes half life and by the end of life the reading can be many times that of the original value.
The instrument does not arrive with a calibration certificate, so I carried out a few checks to verify its performance. The manual provided details a basic calibration procedure, but does not provide any specific test values.
The test voltage produced should be 1 kHz with a 30 Hz tolerance, there is no specification for the peak to peak voltage. An oscilloscope was used to take a look at the signal.
The peak to peak voltage was measured as 3.08 Volts with a frequency of 979.2 Hz. The signal is captured off the source connections of the instrument.
The instrument has a 6 V and 60 V range. I applied test voltages at 10, 50 and 95% of the full range values using a Keithley SMU. The manual shows the appropriate test connections to be made to read the DC voltage applied.
To get a voltage reading, I did have to follow the procedure to disable the disconnection function of the instrument.
All the values tested, were well with the expected tolerance bands.
For the resistance function, there are 4 ranges, 3 mΩ, 30 mΩ, 300 mΩ and 3 Ω. At the moment, I do not have all of the appropriate resistance values to carry out too many tests. I have a shunt resistance box set up of 4-wire resistance measurements that can cover off the first 3 ranges. The final range, I will have to use a low ohm decade resistance box. This is not suited for 4-wire measurements, but I do have bespoke test leads that will help with the tests.
The only values, I have for the 3 mΩ range are a 500 µΩ in the shunt resistance box and a 60A current shunt with a nominal 1 mΩ resistance.
For the 30 mΩ range, I only have a 10 mΩ resistor available.
For the 300 mΩ range, I can utilise the decade box set to 30 mΩ and the 100 and 257 mΩ resistors from the shunt box. For the 3 Ω range, all of the test values came from the decade box. I did find a problem when using the decade box, as the residual resistance measured was 47 mΩ, which is much greater than the zero adjust function of the instrument can cater for. Adjustment of the values therefore became a manual process and means that whilst the decade box can go down to 10 mΩ, the addition of the 47 mΩ limits its use for testing in the lower ranges.
| {gallery} Decade box issue |
|---|
Measuring residual resistance of decade box |
Starting the zero ohm adjustment procedure |
Zero ohm adjustment failure |
All of the resistance values were found to be within acceptable tolerances as per the data table below.
The final element was to verify the test current. Each resistance range of the instrument has its own test current value. This proved to be awkward to measure for the lower resistance ranges, as the resistance of the shunt within the Fluke 8846A forced the BT3554 into its higher resistance range. Initially, the current was therefore verified, by measuring the voltage across a resistor under test.
The current for the 300 mΩ and 3 Ω ranges should be 16 mA and 1.6 mA respectively. The table above shows that these two ranges had a test current close to these expected values. For the two lower ranges, the test current should be 160 mA, but was only measuring around 130 mA when using the Fluke test setup.
I then decided to remove the Fluke 8846A from the circuit and utilise a Megger DCM305E Earth Leakage Clamp to check the current. This is obviously not the desired approach for a calibration check, due to the inherent inaccuracies of using a clamp meter to measure current. It did record 157.2 mA for both ranges, which is comparable by ratio to the 15.7 mA and 1.57 mA for the two higher resistance ranges.
Overall, the calibration checks have worked out well despite the limitations I have. I hope to build up a resistance board with a set of applicable values to be able to verify the instrument in a more professional manner.
This will give me, 10, 50, 90 and 97% test points for each resistance range. I have also included a straight through trace, to carry out the zero adjust on prior to taking any measurements. My intention was to have the board made up for use with both the probes and the kelvin clips. Hence the large pads at the sides of the board that I intended for use with the clips. To do this I need a pad on the opposite side of the board with a further track to the resistor and then likely a via through to join the tracks together.
This is where I became stuck as I needed to add these pads and tracks in manually. Whilst I managed to draw the tracks and pads in, I could not convince kicad to add in the visa. I think Iwill have to go back to the schematic and add in the duplicate connections to see if that solve the issue, instead of creating them manually within the PCB layout.
Overall the calibration has worked out well despite some of the limitations of my test setups.
I took a quick look at the current drawn by the BT3554-01 by replacing the battery pack with a DC power supply. What was of particular interest to me is the current drawn when the instrument is off, as it appears to be a soft-start on/off button. Pleasing the current drawn when off was down to 5 uA.
| {gallery} Current Drawn by Instrument |
|---|
Current drawn with instrument off |
Current drawn for high resistance measurement |
Current drawn with backlight on |
Current drawn for low resistance measurement |
Current drawn with no measurement in progress |
With the instrument in operation, the current drawn ranged from 104 to 144 mA when taking a measurement. With no measurement being made the current was around 80 mA.
I have created two separate blogs to cover the testing and analysis carried out on VRLA and Li-Ion battery banks / packs, so I will only provide a summary of the work / results on the RoadTest.
Link to Discharge Testing VRLA and Li-On Blog
Link to VRLA Battery Bank Testing Blog
AC Resistance measurement on cells should only be carried out on a fully charged cell, usually, it is done whilst they are also on float charge, but can be done will the cell offline. This is especially important for VRLA cells, that the technology is primarily aimed at. For these cells as its state of charge (SoC) gets lower, its AC resistance will increase as a natural effect. This is seen from the plot of the 12 V VRLA battery below.
The battery manufacturer gives a nominal AC resistance value of 50 mOhms for this kind of battery. You can see that the initial value is just over 60 mOhms, indicating that some ageing has occurred. Research of AC Resistance measurements have shown that a cell has reached the end of its useful life when the AC Resistance measurement is twice that of the manufacturer's specification. or the initial reading when the battery was new, if a manufacturer's value is not available.
On the plot above, the AC Resistance reaches twice that value just after the halfway point of the test. At the end of the test, the AC Resistance is over three times the initial value. Therefore if an AC Resistance reading is taken on a partially discharged cell, it may give the impression that it is faulty. As AC Resistance measurements, do not load a cell in the same manner as a DC Resistance test would, the cell voltage can be relatively high and may not indicate that the cell is discharged.
Moving this concept onto the large battery banks, I had a number of banks across a wide age range to carry out measurements on. There were three main intentions. The first to verify the theory of a double AC Resistance value showing end of life by comparing the AC Resistance Measurements from a 12 year old battery against a 2 year old battery of the same cell type.
The second intention was to look at the AC Resistance values across a three banks of the same cell type, that had all failed discharge tests, with the hope of identifying the better cells that could be utilised to make up a good battery bank or two.
There final element was to compare the performance of the new BT3554-01 to the old 3550 HiTester whilst carrying out the measurements.
The overall results of the tests are tabulated below.
The two top rows show a slightly disappointing result for verifying the two times AC Resistance theory. On average, the older battery bank had a cell AC Resistance of only 1.25 times the two year old battery bank. This isn't even above the 1.5 times warning limit. Sadly, I do not have original AC Resistance readings from 12 years ago, so I am unable to use that figure for further validation. Generally speaking, VRLA battery technology has not changed too much over the years, but when dealing with such low resistance values anyway, it is possible that a technical change within the cell is a contributing factor to the poor result. As the manufacturer of the cells does not publish AC Resistance values, I am unable to verify this. The only good aspect is that the AC Resistance increase is relatively consistent across all the cells as seen in the trending snapshot front Hioki Gennect software below.
The three battery banks, that I wished to evaluate are labelled as BTC11, BTC21 and BTC22 in the summary table above. These have all failed a discharge test and as a further check, I have battery banks BTC12, that has passed the discharge test. These battery banks are all of a similar age.
Looking at the actual average of each battery bank, nothing can really be determined. None of the failed battery banks are showing a significant AC Resistance above that of the BTC12 battery bank that passed. A deeper look into the data is required.
For an AC Resistance reading, a cell with a lower reading is better, conversely, for the discharge test, a cell with a higher voltage at the end of the test is better. I therefore reorganised the data to be able to determine which cells had the lower AC Resistance readings and if they were the cels that had the higher discharge voltage. The could not be determined in absolute terms. As the table above shows, none of the cells with the lowest AC resistance had the highest discharge voltage across any of the battery banks. The opposite hypothesis, of the cells with the highest AC Resistance having the lowest discharge voltage, could also not be determined from the data.
I did look more in depth into the data, which can be found in the VRLA battery testing blog, and in some instances cells with high AC Resistance appeared in the bottom quartile of discharge voltage readings, indicating that there is some correlation with the expectations. However, there were also a few cells with high AC Resistance that also had quite high discharge voltage values. It would appear that the data collected, is not going to be able to provide a definitive list of cells that are in a better condition than others. One of the battery banks is due for replacement later this year, so I intend to capture AC Resistance values shortly after commissioning and compare these readings to see if these older cells have followed the ageing principle.
I tested a 18650 Li-Ion cell across its discharge curve in the same manner as the VRLA cell. As expected this produced a much flatter AC Resistance curve over its discharge cycle.
I have some aged 18650 cells in some Ryobi power tool battery packs that I open up in order to obtain further test data. These cells were from 2009 to 2012 and already had depleted performance when used with the tools. One of the battery packs will not charge up on the Ryobi charger, so I also investigated the condition of the cells within this pack with the BT3554-01. The video below is a bit long, but goes through the strip down and initial testing the packs.
This battery pack was interesting, whilst the voltage levels of the individual cells were well below acceptable values, the AC Resistance measurements actually indicted that the the cells were acceptable. The plot below on the left shows the depleted voltages, with only one cell being above the minimum 2.75 V discharge voltage. The plot on the right shows that all the AC Resistance values are around 30 mOhm, which is specified as being acceptable in the data sheet.
With this knowledge, I attempted to bypass the on board Ryobi battery management system and charge the cells manually. The subsequent retest of the cells with the BT3554-01 showed that the the voltage levels were now all at acceptable values and a slight decrease in the AC Resistance values was also seen, withthem all now being below 30 mOhms.
The Hioki Gennect software on the Windows computer also allows for trends to be established and two plots were produced to show the differential between the voltages and resistances prior to and after rejuvenating the battery pack.
The ultimate proof of recovery is with a discharge test that I set up and produced an adequate result for a battery pack that is around 12 year sold and has sat depleted for a while. The test should have lasted for one hour, but as the graph below shows itcut-off at 45 minutes. However it did cut-off above the manufacturer's minimum discharge voltage, this was the battery pack management system operating, and preventing full depletion of the cells.
The plot above shows the data collected by the BT3554-01 during the rejuvenation of the battery pack. The end result is a voltage differential across the cells that is significantly lower than the initial measurements, whilst the AC Resistance values have remained stable.
One of the other battery packs that was functioning was shown by the BT3554-01 to have a couple of faulty cells, so a similar procedure was carried out with this pack to see if it too could be rejuvenated.
Whilst some success was achieved, the BT3554-01 AC Resistance data showed that the cells were still in a poor condition, which was again verified with a discharge test.
Whilst an initial improvement was seen in the discharge curve, this deteriorated but was still better than the initial curve. This curve also never followed the pattern of the other battery, showing a much sharper cut-off at the end of the test, as the battery pack voltage collapsed and the BMS shut it down.
This testing was much more. successful than the VRLA battery testing, allowing a battery pack perceived as being bad, to be recovered and also identifying bad cells within a functioning battery pack.
I carried out readings of the single VRLA and the VRLA battery banks with both the BT3554-01 and the 3550 HiTester for a performance comparison. The BT3554-01 is marketed on a faster measurement capability than the old instrument, so I verified this using the single cell VRLA and the Kelvin Clips. As seen in the video below the 3550 HiTester takes around 12 to 15 seconds for the reading to settle. In manual mode the BT3554-01 to around 5 seconds to settle. When using the Auto Hold / Auto Save function, the BT3554-01 took 3 seconds to take the reading.
It was going to be interesting to see how this would benefit when taking readings on actual battery banks. For comparison I carried out a full set of measurements on one of the larger 55 cell systems and one of the smaller 19 cell systems. In the table below, I have split the measurement task up into specific actions for better clarity.
In terms of voltage accuracy, there isn't much to choose between the two units. There is a substantial difference between the AC Resistance readings, a -12% difference for the smaller battery bank and a +13% difference for the larger battery bank. The BT3554-01 does offer a lower 3 mOhm range, which was used for the readings against the 30 mOhm range of the 3550 HiTester, that may be a factor in the discrepancy. I did not record the test signal from the 3550 HiTester, so this could be different to the signal from the BT3554-01 and also contribute to the differential. The largest contributing factor could well be the connection methodology, Kelvin Clips for the 3550 HiTester, that are also over 15 years old, and brand new concentric pin probes for the BT3554-01. The differential is not too much of a concern, for battery testing, consistency is the key, but from the results, if different instruments are used, then there could be an issue for data trends. It is probably therefore better to stick with the same instrument for the measurements.
The main purpose of this test was to compare the overall timing aspect. For both battery banks, a significant improvement is seen when utilising the BT3554-01 over the 3550 HiTester. Even when looking just at the measurement aspect on its own, it was around twice as quick when using the BT3554-01. A further improvement is also seen over using probes instead of the Kelvin Clips of the 3550 HiTester, that requires the insulating caps to be removed. Not only is this quicker, it is also safer as no live terminals are exposed when utilising the BT3554-01 and the probes.
There is absolutely no comparison when downloading the data captured by the BT3554-01, against the manual transfer of data from the 3550 HiTester to the computer, there is also less likelihood of data errors, that can occur when manually typing in numbers read from another screen.
An overall 61% and 76% time saving is well worth having, especially for those with multiple battery banks to test.
It is also worth mentioning the ergonomics of the BT3554-01 against the competitors. I am not usually a fan of the box style arrangement seen with some of the other test apparatus from Hioki, in comparison to the more usual multi-meter style tester. For battery bank testing though, the arrangement of the Hioki is well suited. The instrument is hung around the neck using the supplied strap and allows the user to move along the battery and take measurements, whilst viewing the screen. The beep conformation also aids with ensuring data has been captured.
The Applent 528 tester I have is more of a multi-meter styled instrument, similar to the vast majority of the other battery testers out there. This makes taking measurements whilst holding the meter more of a challenge. The meter can be stood next to the bank, or hung from the framework, but when you have a battery bank that is 5 metres long, the instrument has to be repeatedly moved. With the meter away from the user, you constantly have to keep moving to verify data is captured on the screen. Offerings of remote probes form the likes of Fluke can overcome this to some extent, but then you can get into problems of gaining access to cell terminals, that the right-angled probes offered by Hioki access with absolute ease and user comfort.
Another aspect I like about the BT3554-01 is its use of standard 4 mm input jacks with a standard spacing of 19 mm. This allows third party accessories to be utilised with the instrument, unlike something like the Fluke BT520 series, that uses a bespoke connector, limiting the user to the manufacturer's accessories.
Throughout the use of the BT3554-01, results from the tests have been stored within the instrument and downloaded to the software for further analysis and longterm storage. Hioki offer two versions of their Gennect software, Gennect One for Windows based computers and a Gennect Cross for iOS and Android devices. As far as I know, there is not a version for Mac based computers. Initially trials were with the Windows based software.
The software maintains a database of all the readings and is structured based around the date that the results are imported into the database. Data is imported either directly from the BT3554-01 or from a file, which can be a number of formats from Hioki's bespoke file to a standard csv file. The software also allows image files to be imported, so that they can be included in reports. With data loaded, test titles, comments and search tags can be added, to further identify the systems being worked on. This is required as when out on site, the user needs to record which memory set the tests are being saved to as the instrument cannot record this. This information is then entered into the software for future reference.
Analysis can be carried out on individual data sets or on multiple data sets for trending purposes. Individual data sets can be displayed as either a list or graphic format. If the comparator limits are set, then the software will utilise these in the displays. The discharge data of the 12 V VRLA battery shows a good example of this.
| {gallery} 12V VRLA Battery Analysis |
|---|
Data list of 12V Battery test |
Graphic plot of 12V Battery test |
In the data list window, all the test values are displayed, along with comparator limits. Test values that hit the warning limits are. displayed in yellow and those that hit the fail limits. are shown in red. A summary of the pass, warning and fail is provided at the top of the table. The graph option displayed either voltage, resistance or temperature readings and these are manually selected. The same key is used to show readings that exceed the warning and fail limits as can be seen in the screen shots above.
The next tab displays the comparator threshold values if they have been entered. This must be done from within the BT3554-01 and enabled at the same time as taking the readings. If this is not done, no data is displayed in this tab. I would have liked the option to be able to edit the values in this tab, so that analysis can be adjusted if required.
The final tab is the trend tab, that is only activated when multiple entries from the database have been selected and then the individual measurement points added to the trend from the list tab. It is possible to turn on or off individual measurement points in the display, so more points can be added initially to save from recreating the trend list.
Data can be exported from the Gennect software as a CSV file for further analysis in a spreadsheet if required. A PDF report of the readings and graphs can also be created and saved to be sent out to a client if the battery testing was being carried out on behalf of someone else.
I did comment in the VRLA Battery Testing blog, that the Windows software could only import data, and there were no options to manage the BT3554-01 from the software. Whilst writing these blogs, Hioki have released an update to the Gennect software, that has added this option in, along with other options to work with the threshold tables and adjust the instrument settings. This gives it the same options as the iOS based software.
| {gallery} New Gennect Software |
|---|
New software cell overview |
New software cell analysis list |
New software profile tab screen |
New software plot analysis |
New software trend analysis |
New software memory management function |
The software also adds in a profile tab to allow information to be added in regarding the battery bank under test, and appears to allow measurement memory locations to be allocated to the profile. This would allow further flexibility of the instrument and allow more battery banks to be tested prior to downloading the instrument. Unfortunately , this only works with the new BT3554-50 series so is redundant for the BT3554-01 and in some ways, it seems now that it was not such a good idea to update the software and I should have stuck with the previous version. The vast majority of the software functionality remains the same as can be seen above.
Later testing was then carried out using the Gennect Cross software on an iPad 4. As the BT3554-01 has the bluetooth option install, this arrangement allows for data to be collected on the iPad at the same time as the readings are being taken. I only used this option when testing on the bench, for me personally, I would find it as a distraction when measuring on the large battery banks that I work on. This is classified as a live working scenario, some of the battery banks are 110 V and some are 220 V. They can have either centre-tapped or negative rail earthing arrangements that present an electric shock risk, so I would rather concentrate on the task in hand and not be distracted by checking that the iPad is capturing the data. This is obviously just me, and others may be comfortable with recording at the same time as making the measurements.
The Gennect Cross software is a multi-instrument platform and can record data from a number of the instruments Hioki offer. The first task therefore, is to select the desired instrument. from the list. This will enable the bluetooth search function and any instruments of the selected type will appear ready for connection. Connecting to the BT3554-01 was simple and consistent, I never had any issues with it throughout my testing.
| {gallery} Gennect Cross Software |
|---|
Gennect Cross Home Screen to select instrument |
Bluetooth Connection to BT3554-01 |
Measurement Capture Screen |
List of Measurements Captured |
Plot of Measurements Captured |
Threshold values applied to measurements |
Measurements are captured on the screen as they are taken. The option exits to record the measurements and they will be saved within the Gennect Cross. software database, in format similar to the Windows software. The process must be started and stopped manually.
With the readings captured, they can be viewed in either a list or a graphic format. The same key from the Windows software is applied for identifying the readings that have exceeded warning or fail limits. The Third tab allows the limits to be displayed. A save screenshot function has also been added to each of these tabs, that will save the screen into the database and allow it to be added into a report.
Measurements, reports and screenshots can be shared as PDFs, CSVs or Gennects internal format and sent out from within the software.
| {gallery} Gennect Sharing Data |
|---|
Data Table Selected can be shared as CSV or Gennect Format |
Data file emailed as CSV |
Plot from PDF Report showing bunched up data |
The reports created follow a similar format to those created on the Windows software with the exception of being able to create a trend between different sets of readings. I could not find out how to do this on the iOS version. I did also note, that whilst the plots fill out the screen when viewing the data, within the PDF report, the plots seem to be set to a standard format, resulting in plots with small amounts of data bunched up into the left corner of the plot instead of spaced out like they are on the screen.
The Gennect Cross software adds in some options to maintain the threshold tables, memory and date and time settings.
The threshold tables can be imported from the instrument and edited. Alternatively, a new table can be created. With modifications made, they can then be sent back to the BT3554-01. I do find this option to be slightly easier and faster than editing the threshold values directly in the instrument. Although it does take a few seconds to import and download as all 200 threshold entries are transferred, whether or not they have values added to them. You also need to maintain a list of what cell types the thresholds are for, as the instrument only stores them as a identifier from C1 to C200.
| {gallery} Gennect Cross Maintenance |
|---|
Gennect Cross options for maintaining BT3554-01 |
Options for Threshold Tables |
Importing Threshold Tables from BT3554-01 |
Threshold Table List |
Memory Management Options |
The memory management option allows data saved on the BT3554-01 to be downloaded and the used memory to be cleared. On the instrument itself, individual readings can be deleted, but the software only allows for each of the 12 memory slots to be cleared, effectively deleting 500 individual readings at a time.
The Hioki BT3554-01 has shown itself to be a very capable instrument. I can honestly say that nothing has really gone wrong during this RoadTest that can be attributed directly to the instrument itself. Certainly, not all of the test results have followed expectations, particularly in regard to the measurements on the VRLA cells, that have not been able to be correlated with the discharge test data and and be used to ascertain the best cells for retention.
For battery bank testing though, the set-up of the Hioki BT3554-01 showed great safety and time improvements over the older instrument. Using the probes there was no need to remove terminal safety covers and expose live voltages. The right-angled probes offer much better access to cells that are installed on a lower racking with limited space in which to take the measurements.
Tests on Li-Ion cells was more productive and matched the expectations much more closely, with those cells showing a higher than expected AC Resistance showing a much shorter discharge time. The BT3554-01 identified a number of cells, that whilst they were showing very low voltages below the manufacturer's specification, the AC Resistance reading indicated that they had good health, and these cells could be recovered using a manual charging process.
Even during this RoadTest review Hioki have continued to extend the capabilities of their battery testers, adding more to the user interface and software modules. If you have an old 3550 HiTester, then I would say that you would make significant gains by upgrading to the BT3554. Unless you have a specific requirement for cells above the 3.1 Ohm limit of the BT3554, then it offers accuracy, performance and user friendliness over both the Applent and Tenmar offerings. In terms of cost benefit, although I was unable to get hold of a Fluke unit for direct comparison, on paper I would say that the Hioki would be a better offering and add a clamp meter to cater for the extra test functions the Fluke offers. This package would still be cheaper than the cost of the Fluke setup.
This concludes my review of the Hioki BT3554-01 battery tester. I hope that, whilst it is not a subject of prominence within the element14 community, there are some that find it interesting and useful to read. Many thanks to Hioki and element14 for allowing me to review the instrument.
Top Comments
Very comprehensive road test report.
Well done.
DAB
Top shelf review DL! I missed lunch watching one of the videos:)
The video product were excellent. What was your setup that provided the stability and great viewing angle. I suspect the camera was strapped…
Hi Donald,
Great review! The knowledge and research that went into this review is really impressive.
One minor thing: The resistance values you used for your calibration tests - are these DC values or AC…