⚡ electrical engineering

Design of an embedded hardware platform for cell-level diagnostics in commercial battery modules

This paper presents the design and development of an embedded hardware platform that enables non-invasive, simultaneous cell-level monitoring and balancing across all 36 modules of an Audi e-tron battery pack, facilitating accurate state-of-health assessments and supporting efficient maintenance and repurposing without module disassembly.

Original authors: Gabriele Marini, Alessandro Colombo, Andrea Lanubile, William A. Paxton, Simona Onori

Published 2026-04-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Gabriele Marini, Alessandro Colombo, Andrea Lanubile, William A. Paxton, Simona Onori

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you have a massive, high-tech backpack (the electric vehicle battery) made of 36 smaller, identical lunchboxes (the modules). Inside each lunchbox, there are three smaller sandwiches (the cells) stacked together.

For years, scientists and engineers have been great at checking the health of individual sandwiches. But when those sandwiches are packed tightly inside a lunchbox, it's been almost impossible to peek inside without tearing the lunchbox apart. If you want to know if one sandwich is stale or squished, you usually have to rip the whole thing open, which ruins the lunchbox for future use.

This paper is about building a special "X-Ray Goggles" and a "Smart Balancer" that let us peek inside these lunchboxes without ever opening them.

Here is the story of how they did it, explained simply:

1. The Problem: The "Black Box" Lunchbox

The researchers took a battery pack from a used Audi e-tron (an electric car). They wanted to see how the individual "sandwiches" inside were aging.

The Challenge: The lunchboxes are sealed tight. You can see the voltage of the whole lunchbox, but you can't see the voltage of the three sandwiches inside.
The Risk: If one sandwich is weaker than the others, it gets stressed out, heats up, and could fail. Without seeing inside, you might think the whole lunchbox is fine, only for it to fail later.

2. The Solution: The "Magic Board" (M&BH)

The team built a custom electronic board (a piece of hardware) that acts like a spy gadget. They plugged it into the lunchbox's existing ports (like plugging a charger in).

The X-Ray Goggles (Voltage Sensors): This board has tiny wires that tap into the connections between the sandwiches. It measures the voltage of each sandwich individually, 10 times a second. It sends this data to a computer so engineers can see exactly how each sandwich is doing.
The Smart Balancer (The Referee): Sometimes, one sandwich gets "too full" (overcharged) while another is "too empty." If you keep charging, the full one might burst.
- The board acts like a referee. If it sees one sandwich getting too full, it opens a tiny "leak" (a resistor) to let a little bit of energy drain out of that specific sandwich.
- This keeps all three sandwiches at the same level, ensuring they don't get stressed out during the test.

3. The Test: The "Stress Workout"

Once the board was attached, they put the lunchboxes through a rigorous workout routine:

The Long Run (Capacity Test): They drained the battery slowly to see how much "fuel" each sandwich could hold.
The Sprints (HPPC Test): They gave the battery quick jolts of power (like sprinting) to see how much "friction" (resistance) was inside each sandwich.
The Temperature Check: They did this in a room that could be cold, warm, or hot to see how temperature affected the sandwiches.

4. The Big Discoveries: What They Found Inside

Because they could finally see inside without breaking the lunchbox, they found some surprising patterns:

The "Weak Link" Pattern: In most lunchboxes, the first sandwich in the stack was usually the weakest (held the least energy). It was the one most likely to fail first.
The "Middle Child" Effect: The middle sandwich was always the "coolest" (had the lowest electrical resistance). It seemed to run more efficiently than the ones on the outside.
- Why? The researchers suspect the middle sandwich is slightly warmer than the outer ones because it's surrounded by other sandwiches. Just like how your hands get warmer when you hold them together, the middle sandwich runs hotter, which lowers its resistance.
The "Stale Sandwich" Problem: They found that even though the lunchboxes looked identical from the outside, the sandwiches inside were aging at different rates. Some were much older and weaker than their neighbors.

5. Why This Matters: The "Second Life"

This is the most important part. Electric car batteries are expensive. When a car battery is "too old" for a car (maybe it only holds 80% charge), it's often thrown away. But that's a waste!

The Second Life: These batteries are perfect for stationary storage (like storing solar power for a house or a factory).
The Sorting Problem: Before you can reuse a battery pack, you need to know which lunchboxes are safe and which ones are ticking time bombs.
The Impact: This new "Magic Board" allows companies to check the health of every single sandwich inside a used battery pack without destroying the pack. They can sort the good lunchboxes from the bad ones, ensuring that when these batteries get a "second life," they are safe and reliable.

In a Nutshell

Think of this paper as inventing a way to check the health of every apple inside a sealed crate without opening the crate. By using a clever electronic board that listens to the apples and gently nudges them to stay even, the researchers proved they can find the rotten apples and the healthy ones, making it safe to reuse these batteries for new jobs.

1. Problem Statement

While battery aging is frequently studied at the individual cell level, evaluating aging and performance within fully assembled commercial battery modules remains a significant challenge.

The Gap: Traditional methods for accessing cell-level data (voltage, current, resistance) require disassembling the module, which is destructive, time-consuming, and compromises the mechanical and thermal integrity of the pack.
The Limitation: Existing non-destructive methods are rare. Most commercial modules (like those in EVs) do not provide direct electrical access to individual cells within a series string without opening the casing.
The Need: There is a critical need for a non-invasive hardware solution to screen individual cells within commercial modules to assess heterogeneity, ensure safety, and facilitate reliable "second-life" repurposing of battery packs.

2. Methodology and Experimental Setup

The authors designed and implemented a custom Monitoring and Balancing Hardware (M&BH) platform to test 36 modules from a 95 kWh Audi e-tron battery pack.

A. Hardware Design (M&BH)

The platform is a low-power embedded system that interfaces with the module's external connectors (G-, C1+, C2+, C3+) without opening the casing. It consists of:

Analog Circuit: A resistor/capacitor (RC) network that scales cell voltages (4:1 divider) and filters high-frequency noise (low-pass filter at ~113 Hz).
Digitization: An ADS1115 16-bit ADC sampling at 10 Hz.
Microcontroller: A Raspberry Pi Pico that processes data, rescales voltages, and manages logic.
Balancing Circuitry: Passive balancing switches connected to bleeding resistors ( $R_b = 67.5 \Omega$ ).
Communication: A CAN adapter to communicate with the battery cycler (Arbin LBT22013).

B. Balancing Strategy

To ensure safety and prevent voltage imbalances during testing, the system employs a bang-bang control strategy:

If a cell's voltage ( $v_j$ ) exceeds the minimum voltage in the module by a threshold ( $v_{th} = 2.5$ mV), the corresponding switch closes.
Current is diverted through the bleeding resistor, dissipating excess energy and slowing the charge rate of the overcharged cell.
This ensures the maximum voltage imbalance ( $\Delta v_{max}$ ) remains within the defined threshold.

C. Testing Protocol

The study utilized a specific diagnostic profile on all 36 modules (configured as 4P3S, meaning 3 series groups of 4 parallel cells):

CC-CV Charging: To initialize all cells to a uniform state (4.2V) with active balancing.
Capacity Test: C/3 discharge to 3.3V (balancing disabled to capture natural heterogeneity).
HPPC (Hybrid Pulse Power Characterization): Current pulses at 90%, 65%, and 40% State of Charge (SOC) at three temperatures (15°C, 25°C, 35°C) to measure internal resistance.

D. Data Processing

Capacity/Energy: Calculated via Coulomb counting over a common voltage window defined iteratively across all modules to ensure comparability.
Internal Resistance: Calculated from the ohmic voltage drop during HPPC pulses, corrected for a known 100ms communication latency between the cycler and the M&BH.

3. Key Contributions

Non-Invasive Hardware Platform: The first reported custom hardware solution capable of non-destructively measuring individual cell voltages and performing balancing in commercial, series-connected modules (specifically 4P3S topology) without disassembly.
Quantification of Heterogeneity: A methodology to extract cell-level metrics (Capacity, Energy, Resistance) from module-level tests, enabling the identification of "weakest cells" within a module.
Thermal and Electrical Correlation: Demonstration of how internal resistance varies with temperature and SOC, and the identification of structural intra-module patterns (e.g., central cells behaving differently than outer cells).
Second-Life Enabler: A practical tool for sorting and qualifying used battery modules for repurposing in stationary storage, addressing the safety risks of unknown cell health.

4. Experimental Results

The platform was tested on 108 "Cells" (where a "Cell" represents a 4P parallel group) across 36 modules.

Capacity and Energy Heterogeneity:
- Significant variations were observed across the pack.
- Pattern: "Cell 1" (the first series group in the module) was identified as the weakest cell (lowest capacity) in 16 out of 36 modules, suggesting a systematic intra-module gradient.
- The module with the lowest overall capacity (Module 12) contained the weakest individual cell (Cell 34).
- Strong correlation ( $\rho = 0.924$ ) was found between discharge capacity and discharge energy.
Internal Resistance Heterogeneity:
- Resistance increased as temperature decreased and as SOC decreased.
- Pattern: A distinct periodic pattern was observed where the central Cell (Cell 2) consistently exhibited lower resistance than the outer cells (Cell 1 and Cell 3).
- Hypothesis: This suggests the existence of intra-module thermal gradients, where the central cell operates at a slightly higher temperature than the outer cells, reducing its resistance. Surface thermocouples showed minimal temperature differences ( $\Delta T < 1^\circ$ C), implying that surface measurements may mask internal thermal variations.
Balancing Effectiveness: The M&BH successfully maintained voltage imbalances within the 2.5 mV threshold during charging, preventing overcharging of individual cells.

5. Significance and Conclusion

This work bridges the gap between academic cell-level research and industrial module-level testing.

Safety & Reliability: By enabling non-invasive screening, the platform allows for the detection of underperforming or failing cells before they cause thermal runaway or premature pack failure.
Second-Life Applications: The ability to quantify internal heterogeneity is crucial for the "second life" of EV batteries. It allows for accurate sorting of modules, ensuring that only healthy modules are reused in less demanding applications (e.g., grid storage).
Future Directions: The authors note that while surface temperature sensors showed uniformity, the resistance data suggests internal thermal gradients. Future work will focus on direct cell-level temperature sensing to validate these thermal hypotheses.

In summary, the paper presents a robust, scalable, and safe engineering solution that transforms how commercial battery modules are diagnosed, moving from destructive sampling to comprehensive, non-invasive internal monitoring.