Nuvoton Puts EIS Inside a 26-Cell Battery Monitor, Samples Start January 2027

Nuvoton Technology Corporation Japan (NTCJ) announced on September 29, 2026 that it will begin sampling the KA85010UA in January 2027. The part is a sixth-generation battery monitoring IC that carries electrochemical impedance spectroscopy (EIS) on the same die as its cell monitoring front end: up to 26 series-connected cells, a maximum stack voltage of 135.2 V, ±2.0 mV voltage accuracy at end of life, in a 128-pin LQFP. The number Nuvoton is leading with is thermal — its proprietary inductor excitation is claimed to cut PCB temperature rise and power loss during EIS measurements by approximately 93% compared with the conventional resistor method.

For teams building EV and grid-scale storage BMS hardware, the announcement lands on two levels. Technically, impedance spectroscopy — until now a bench-instrument technique performed with dedicated equipment — becomes something the BMS can run on board, continuously, which is what makes state-of-health estimation, internal temperature inference and early detection of cell anomalies practical rather than occasional. Commercially, it confirms that EIS has moved from a research topic to a competitive theme in battery management front ends.

Reading the Spec Sheet

The product page and the KA85010UA Product Brief (Rev 1.00, dated September 29, 2026) list the following:

Channels and voltage: up to 26 series cells, 135.2 V maximum stack voltage, ±2.0 mV accuracy (EOL); every cell voltage is measured over two independent paths — data acquisition ADCs and monitoring ADCs — forming a redundant measurement system
EIS: inductor excitation with quadrature detection-based V/I complex analysis, performed inside the chip
Communication: SPI and a bidirectional daisy chain; 4 MHz serial interface with packet error checking, up to 55 ICs chained over twisted-pair differential signaling
Functional safety and quality: ISO 26262 ASIL-D Ready, AEC-Q100, OV/UV/OT/UT alarms, 10 FIT market failure rate
Other: 16-bit parallel delta-sigma ADC, 14 GPIOs, cell balancing at 300 mA with internal MOSFETs or external N-channel MOSFETs, low-power intermittent operation, LQFP-128 (14 × 14 mm²)

Two details deserve a careful read. The interfaces listed are SPI and daisy chain only — no isoSPI, no CAN anywhere in the material — so teams whose architecture already commits to an isoSPI ecosystem should budget the isolation and harness changes rather than assume a drop-in. And the safety wording is "ASIL-D Ready" and "designed for ISO 26262-compliant systems," which is a positioning statement about the system design target, not a certificate for the device alone. Nuvoton also does not publish the EIS measurement frequency range; the release only says that high, mid and low bands map to different parts of the cell.

Where the 93% Comes From: Recycling Excitation Energy Instead of Burning It

EIS works by superimposing an AC current onto the battery and deriving impedance from the resulting voltage and current response. The conventional resistor method pushes several amperes through a resistor, and most of that energy leaves as heat — which is precisely why EIS has struggled to leave the lab. The KA85010UA uses two FETs to steer current direction and circulates energy stored in a coil as the AC excitation current instead.

Nuvoton published its comparison conditions along with the result: power loss was compared at a total voltage of 72 V with 1.5 A superimposed current, and temperature rise at a pack voltage of 60 V with 0.7 A. Both cases came out at roughly 93% lower than the resistor method.

The practical meaning is thermal headroom. A BMS board sits next to the cells in an already hot environment; an excitation stage that adds tens of degrees will often be rejected at thermal review outright. Suppress that contribution to near-negligible and you can justify taking EIS data across operation, charging and rest, which is exactly what Nuvoton claims the inductor approach enables.

Why Everyone Is Betting on EIS — and Who Is Already There

Conventional BMS designs estimate SOC and SOH from voltage, current and temperature plus coulomb counting and OCV correction, and they simply cannot see inside the cell. EIS resolves the cell's interior by frequency: the high-frequency band reflects terminals, wiring and electrolyte, the mid band the negative electrode, the low band the positive electrode and internal reactions (Nuvoton's own description). The link between impedance and SOH is well established in the literature, including a 2023 review in Protection and Control of Modern Power Systems on EIS-based state-of-health estimation.

Nuvoton says that under its own evaluation conditions the KA85010UA's EIS results agreed closely with dedicated EIS equipment and with the 1 kHz AC internal-resistance measurement specified in IEC 62620, and that the verification was published in an IEEE paper (DOI 10.1109/TIM.2026.3661703) as joint work with Japan's National Institute of Advanced Industrial Science and Technology (AIST). The two sides previously reported joint results on equivalent-circuit fitting for SOH estimation and, later, on internal cell temperature estimation.

Demand-side numbers explain the investment: MarketsandMarkets puts the global battery energy storage system market at USD 50.81 billion in 2025 and USD 105.96 billion by 2030 (15.8% CAGR), with the automotive BMS market at USD 6.53 billion in 2025 rising to USD 15.65 billion by 2030 (19.1% CAGR).

Nuvoton is not arriving at an empty table:

ADI: the ADBMS6842, ADBMS2970 and ADBMS6822 chipset embeds the EIS measurement chain in BMIC and AFE silicon and is already in a production XPENG vehicle; ADI describes XPENG as the first OEM to put its EIS architecture into mass production
NXP: announced an EIS-capable BMS chipset (BMA7418, BMA6402, BMA8420) on October 29, 2025, built around hardware-level nanosecond synchronization and in-chip discrete Fourier transformation
TI: launched the BQ79826Z-Q1 on June 9, 2026 — also 26 series cells, cell voltage accuracy better than 1.7 mV from -40 to 125°C, EIS impedance accuracy of 1% at 1 A excitation into a 200 µΩ impedance, with preproduction units available and production expected by the end of 2026
Nuvoton: the inductively excited KA85010UA differentiates on excitation-stage power and heat plus algorithm support; samples arrive in January 2027

Notably, each vendor excites the pack differently. TI's published material describes a dual-active-bridge topology that reuses existing pack balancing circuits; NXP uses the DC-link capacitors as secondary energy storage to generate the excitation signal; Nuvoton circulates energy in an inductor. The goal is shared — make the excitation stage cheap enough in power and heat to ship — but the implementations have diverged, which usually means another round of competition over accuracy, algorithms and tooling.

Samples, Roadmap, and What the Lineage Suggests

Per Nuvoton, KA85010UA samples are available from January 2027; no mass-production date has been published. EIS analysis algorithms come with the part, along with support spanning measurement setup, data collection and organization, and implementation of estimation and anomaly-detection functions. Evaluation hardware is by request — the product page only states that Nuvoton provides information and evaluation tools, and no evaluation board part number is published.

The lineage makes the direction clear. The fourth-generation automotive battery monitoring chipset that entered mass production in September 2023 (KA84950UA and siblings) handled up to 25 series cells, 132 V absolute voltage, 1.5 mV accuracy, SPI and daisy chain with up to 55 devices chained, in QFP-100. On July 29, 2026 Nuvoton launched the 16-cell KA49703A and KA49713A for AI-server battery backup units at ±2.9 mV, in mass production since September. The KA85010UA now folds 26 cells and EIS into one LQFP-128. EIS is the theme of generation six, and at 26 cells Nuvoton has matched the same-count part TI brought to market in June.

HSY Perspective

The question we actually get from customers is not whether EIS is useful — battery teams already accept that — but whether to start now or wait a year. Our honest read is that EIS for SOH has always been blocked by two things: the heat and power of the excitation stage, and the fact that somebody has to turn impedance data into an algorithm. The 93% addresses the first, and the analysis algorithms plus "measurement environment to product implementation" support address the second. That combination, not the pretty spec table, is what makes this part worth tracking — our customers are never short of ADCs, they are short of diagnostic capability.

The January 2027 sample date is what projects need to plan around. A passenger-car platform that locks its BOM in the second half of 2027 can put this part on the candidate list now and evaluate the EIS data chain and diagnostics in parallel. A storage cabinet or PCS program that needs samples before the end of this year will most likely ship on an existing device and migrate a generation later. From what we see across customer projects, the timeline for these algorithm-backed BMICs is rarely set by the silicon; it is set by whether the battery model exists on the customer side. The same chip takes one team three months and leaves another still calibrating half a year in.

A few spec details are easy to skim past, and we flag them early for purchasing and hardware teams. The interface is SPI and daisy chain, so anyone with isoSPI already in the architecture should price out isolation and harness changes separately. The 300 mA balancing figure assumes internal MOSFETs; higher balancing current means external devices and a fresh thermal and layout pass. On functional safety, the official wording is ASIL-D Ready — confirm the documentation boundary and deliverable list with the factory directly instead of treating marketing language as an acceptance criterion. And the EIS frequency range for this part has not been published at all, so ask for the measurement conditions rather than assuming a benchtop analyzer's spectrum.

HSY has carried Nuvoton's line for a long time and keeps its documentation close at hand, and the KA85010UA is the kind of part we would sample early. If you are designing a new EV or storage BMS platform, it is worth having us send the samples and Product Brief now so you can get the "how do we take the impedance data, and who computes it" loop running. Getting the data early tends to pay off more than getting the chip early.

Sources: Nuvoton Technology, "Nuvoton introduces 26-Cell EIS-Enabled Battery Monitoring IC Samples" (link, 2026-09-29); KA85010UA product page and Product Brief Rev 1.00; IEEE, DOI 10.1109/TIM.2026.3661703; Analog Devices XPENG EIS case study; NXP news release (2025-10-29); Texas Instruments news release and BQ79826Z-Q1 product page (2026-06-09); Protection and Control of Modern Power Systems, EIS-based SOH estimation review (2023); MarketsandMarkets battery energy storage and automotive BMS market reports.