Australian Engineer Solves 2018 Hyundai Ioniq’s Reduced Range Problem

Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News!

I have been meeting Brian O’Neill at our “Coffee, Cake and EV” mornings for the past couple of years, and recently our conversations have centred on his quest to solve reduced range problems with his 2018 Hyundai Ioniq EV. I kept saying to him: that would make a great article, write it up for me. Well, the issue has finally been resolved, and here is his story. Warning: this article contains lots of “eye glazing” technical details. One for the techno buffs out there.

Brian with his beloved, yet frustrating 2018 Ioniq. Photo courtesy Brian O’Neill.

First a bit of background on Brian. He tells me: “I am a Certified Professional Engineer (retired). I have a Master of Engineering degree in Complex Control Systems and am a Fellow of the Institution of Engineers (Australia). I have worked in the design and troubleshooting of electronic instrumentation, sensors and power systems in space research, geophysics, radar, meteorology, and electromagnetic field hardware with CSIRO, academia and manufacturing industry from 1954 until my retirement in 2020.” Wow, no wonder he was able to solve the range conundrum!

By Brian O’Neill

On 24/9/2023, I travelled from Bribie Island to Cleveland Showgrounds and returned after leaving home “fully charged” with an indicated range of 255km. This is a trip I had made numerous times over the car’s lifetime & always returned with plenty of spare range. On this occasion, at about 30km (actual) from home, I noticed that the indicated remaining range was dropping very much faster than the true distance, so I slowed to conserve energy. About 5km from home, I received the first warning at a nominal 13.5% State Of Charge. Soon after, the car switched to restricted power mode (10% SOC), then “turtle” mode (3% SOC). The indicated range ticked over to “——” as I rolled into my garage. Thus, the car went from 13.5% SOC to zero in a distance of ~5km, very slow driving. The energy used from “fully charged” to zero was 21.66kWh, as calculated from 190km x (11.4kWh / 100km) as indicated on the screenshot at the end of this report. The specified usable battery capacity is 28kWh.

As the car was almost due for a 90,000 km service, I booked it in and asked for a battery State of Health (SOH) check at the same time. When I picked up the car after the service, I was quite surprised to hear that the battery SOH was “fine”, so I purchased an OBD2 (On Board Diagnostics System) reader to obtain further information. The car scanner confirmed a battery SOH of 100%, and, at 100% SOC (Displayed), no measurable voltage difference between maximum and minimum individual cell voltages. As at December 2023, the battery SOH has fallen slightly to 99.2% — still an excellent value. Even at 20% SOC (displayed), there is negligible voltage difference between maximum and minimum individual cell voltages and no drop in battery SOH, suggesting that the battery is still in good condition.

It appears that the battery is not being charged to 100% “displayed” SOC (95% “BMS” SOC) despite the car scanner and energy meter readings. This is suggested by the fact that a fully charged battery voltage reading on my car is only 394 volts, whilst comparable cars of the same model charge to 398 volts on an AC charger. Energy meter and Remaining Range meter track reasonably well down to about 50% (displayed) SOC, but below this value both drop very much faster than the actual km travelled. It appears that the discharge voltage curve does not match that of the actual battery fitted to the vehicle, suggesting a software problem. This is confirmed by screenshots of scans of other vehicles of the same model. (Voltages are consistently lower at the same SOC in this vehicle.)

Hyundai do not publish information on the actual battery chemistry used in this model, but the subsequent 38kWh model apparently uses NMC622 batteries and I have used an NMC622 datasheet to compare readings from the later model. After allowing for the fact that the 38kWh model charges to 97% whilst the 28kWh model charges to 95%, I find that the current 100% SOC individual cell voltage settings for my vehicle and the 38kWh model both are consistent with the NMC622 datasheet. This suggests that my BMS may have been programmed for NMC622 batteries. Other (full range) 28kWh Ioniqs are programmed for a higher voltage. It seems unlikely that an NMC622 battery is actually fitted, as this chemistry was not in production when my model went into production, and my vehicle was not recalled for a battery change when a number of 28kWh vehicles were recalled in 2021. (I am unaware of the actual chemistry used for the replacement battery in the recalled Ioniqs.)

If the problem is, in fact, due to an inappropriate software update, then we would need to determine just when this update occurred. It would have to have been after September 2022, as the round trip to Cleveland showgrounds showed no problems at that time. The most likely date was November 2022 when the vehicle had its 75,000 km service. I did notice a 40 km drop in estimated 100% SOC range at that time, but, as I had replaced the tyres at the same time, I assumed incorrectly that the replacement tyres were to blame.

[Brian contacted the Hyundai dealership but …]

The Hyundai dealer service department was unable to help, simply pointing out that the SOH (State Of Health) readout for the battery was 100%, so I was on my own for finding a solution to the problem. Background research confirmed that the battery was composed of 196 individual LQ1729-A2 LG Chem NMC cells and a datasheet for these cells was obtained from LG Chem. In the 28kWh Ioniq, 192 cells are configured as a single series string of 96 parallel pairs of cells. (Note that the later version 38kWh Ioniq is configured differently as two separate sets of 88 cells in series.) The BMS (Battery Management System) is based on the Linear Technology LTC6804/LTC6811 chip series with proprietary Hyundai software. Datasheets for the LTC6804 & LTC6811 were also obtained from the manufacturer and used in the diagnosis.

A couple of important takeaways from a study of the battery datasheet included:

The battery degrades rapidly when it is stored at high State Of Charge (SOC), particularly at temperatures above 20°C. This suggests that ideally the SOC should normally stay under 70% and only be fully charged immediately before starting on a long trip.
The State Of Health (SOH) (which is apparently derived from the battery internal resistance in this case) is very poorly correlated with battery capacity loss, particularly at temperatures below 35°C. Thus, the SOH is no substitute for a full discharge test and should only be used as a coarse indicator.

According to the Linear Technology datasheets, passive battery balancing of groups of 6 cells (or in this instance 6 pairs of cells) occurs by discharging 0.167 amps from the highest voltage cell(s) in the group until all cell voltages are equal, while the BMS is turned on. Balancing between the 16 groups of 6 pairs of cells is thus controlled by the management software turning the cell level BMS off for all balanced groups. Individual cell voltage measurements have an error of +/- 0.5 millivolts over the temperature range. Note that the cell level balancing current is only 0.2% of the ampere hour capacity of the cell pairs so that the maximum cell level balancing rate in this case is very slow at around 0.2% per hour.

Extensive analysis of charge and discharge data confirmed that the Ioniq voltage and current sensors were approximately correct, but that the slope of the battery voltage vs kWh graph differed somewhat from the datasheet data in that the inflection points in the datasheet data had been smoothed out in the measured data. This suggested the possibility that the individual pairs of cells comprising the battery might be at different states of charge; i.e., the battery was not balanced. This could not be confirmed from the OBD2 readouts, as the 96 individual cell pair readings are truncated to a resolution of +/-2 millivolts (despite the BMS chips communicating a much higher resolution to the management software).

In order to test this hypothesis, the 7kW charger was calibrated and the charger losses calculated to be 12%. An assumption made for this test was that the BMS was switched off by the management software whilst the charging current was set to 32 amps. Monitoring the charger showed that the time period for which the current was reduced at the end of charge was very short — only a few minutes. The battery voltage at the end of charge was measured at 394 volts, and this was consistent with the values that had been measured over a 12-month period. After discharging the battery to 10% SOC, it was recharged again at a current of 8 amps. The assumption was that battery balancing would be enabled at this lower current value. At the end of this test, the apparent charger losses had risen to 20% and the “end of charge” battery voltage had risen to 398 volts. This voltage is consistent with the voltage measured on other similar 28kWh Ioniqs. The apparent rise in charger losses confirmed that more energy had been returned to the battery on the charge cycle than had been removed during the corresponding discharge cycle. This is consistent with the higher charge voltage obtained. Subsequent driving tests have shown that at least half of the apparent degradation has now been reversed.

Further study of the Linear Technology BMS datasheets suggests a possible explanation. If the battery becomes sufficiently unbalanced, the pair of cells in the pack at the highest state of charge could force the charger to stop charging before the battery voltage reaches a level which would normally reduce the charge current and turn on battery balancing. If this state is reached, battery balancing is effectively disabled. The only way to re-enable it would be to set the normal charging current at a low enough value for battery balancing to occur at lower states of charge for a long enough period of time to re-balance to below this tipping point.

To ensure that this problem does not recur, my intention is that the battery will be re-calibrated at least monthly by charging at 8 or 10 amps from a SOC of 30% or below, particularly in an environment like Queensland where high ambient temperatures are common.

As this is a problem which has not been reported by other Ioniq owners, it is likely that a particular combination of charging/discharging circumstances has been a causal factor:

High ambient temperatures.
Leaving the car fully charged ready for a possible emergency at all times.
Charging at the maximum 7kW AC always — other than DC charging 2-3 times per year.

I now store the car at 30%–70% SOC unless a run of 100km or greater is anticipated, in which case I recharge just before leaving. I re-calibrate the battery monthly by charging from ~30% at 8A AC.

The simplest check for possible battery balance problems in the 28kWh Ioniq is battery voltage at 100% (displayed) SOC. A voltage measurement under 398V indicates that re-calibration is required. Unfortunately an OBD2 reader is needed for this.

Brian tells me that there has been a worldwide study of this issue — you can see the video about this report here. He cites another Ioniq owner who has solved the problem as well.

In a follow up email, he elaborated on his analysis: “I believe the problem is more likely to occur in the Ioniq 28kWh because of a combination of very low power passive BMS (0,67W per element) and the unique battery configuration of a series of 96 parallel pairs of cells. Only one BMS element can be applied to a parallel pair which are forced to have exactly the same voltage by hard paralleling. Thus the BMS can only balance 96 individual voltages out of a total 192 individual cells. This compares with 176 individual BMS elements (2 parallel strings of 88 individual cells each) in the 38kWh Ioniq. Certainly battery balancing does not occur al a 7kW charge rate (as confirmed by my measurements) on the Ioniq 28kWh, but does occur at 2kW and 2.4kW — i.e., on ‘granny chargers’.

“With the way Hyundai sanitises the OBD2 data and uses very inaccurate (low resolution) algorithms to derive their displayed data, it would be very difficult for most Ioniq owners to detect that this problem was happening. One has to be a mathematical pedant (like me!!), or learn about the problem the hard way by running out of charge (like I did) – then doing the math to determine that it was not just unique driving conditions on that particular day. I do have a friend with the same model Ioniq on Bribie, and I have used my OBD2 reader to confirm that his Ioniq does not have the same problem (despite the fact that his battery SOH is reported as 96% while mine is reported as 100%). I believe that the small battery in this model means that most owners who mainly use the ‘granny’ charger will never see this problem occur.”

Well done, Brian. Hopefully your hard work will help other Ioniq owners. Please add your thoughts in the comments if you have had similar issues.