❓ Frequently Asked Questions

Common questions about this study, the methodology, and DIY battery systems.

Contents

System Design

Why mixed brands? Isn’t that risky?

The architectural immunity hypothesis posits that parallel-connected cells self-balance through the shared bus connection. When cells are connected in parallel, Kirchhoff’s laws force them to the same voltage. Any cell at higher voltage will discharge into lower-voltage cells until equilibrium is reached.

This study provides empirical evidence supporting this principle: 94+ days of monitoring show no divergence at the bus potential, and no growing instability signatures.

However: This doesn’t mean you should deliberately mismatch cells. The claim is that if you have mixed-brand cells (common in budget builds), parallel connection provides inherent balancing.

Do I need a BMS?

Yes, always. Architectural immunity does not replace cell protection.

A BMS protects against:

The architectural immunity concept addresses voltage matching between parallel cells, not any of these safety functions.

What’s the advantage of 4S (12V) over 8S (24V) or 16S (48V)?

This study uses a 4S configuration (12.8V nominal) for several reasons:

  1. Compatibility — Works with common 12V appliances, inverters, and RV/marine systems
  2. Simplicity — Fewer series connections means fewer potential failure points
  3. BMS availability — 4S BMS units are common and inexpensive
  4. Safety — Lower voltage is inherently safer for DIY work

Higher voltage systems (24V, 48V) are more efficient for high-power applications due to lower current for the same power (P = V × I), but require more sophisticated BMS and wiring.

Can I add more capacity later?

Yes, with caveats:

  1. Same chemistry — Only add LiFePO₄ to LiFePO₄
  2. Similar SOC — Charge new cells to match existing bank before connecting
  3. Adequate wiring — Ensure bus bars and cables can handle increased current capacity
  4. BMS capacity — Verify your BMS can handle the total capacity

The architectural immunity principle suggests mixed-brand additions will self-balance, but starting with similar SOC minimizes initial current flow between new and existing cells.

Data & Methodology

Why time-based rolling mean instead of sample-count?

Variable sampling cadence (from state-change logging) means a fixed sample count represents different time windows depending on how frequently the voltage changes.

For example, with ~3-second median cadence:

Time-based rolling (rolling('60s')) ensures consistent smoothing regardless of sampling irregularity. This is critical for fair comparison across different time periods.

How accurate is the parasitic draw estimate?

The 13–20 mA range is inferred from voltage drift, not directly measured. This introduces several uncertainties:

  1. OCV-SOC relationship — LiFePO₄ has a very flat voltage curve in the mid-SOC range, making voltage-to-SOC conversion imprecise
  2. Temperature effects — Voltage varies with temperature (~1 mV/°F observed)
  3. Equilibration time — Voltage continues to stabilize for hours after current changes
  4. System draw variability — Actual draw varies with Wi-Fi polling, sensor activity, etc.

The highest-value improvement would be direct bus-current measurement with a calibrated shunt, which would collapse this uncertainty.

Why are there multiple drift rates reported?

Drift rates are window-dependent on a non-linear relaxation curve. The battery voltage follows an exponential-like approach to equilibrium, not a linear decline.

Window Rate What It Represents
Full stasis (70 days) −0.67 mV/day Average decline over entire monitoring period
Last 30 days −0.17 mV/day Rate near equilibrium
Last 7 days ~0.0 mV/day Approaching stable storage voltage

This isn’t a contradiction—it’s expected behavior for a system approaching equilibrium. The 75% rate reduction from full-period to last-30-days is the clearest evidence of stabilization.

What caused the spread increase on Dec 23?

The spread increase correlates precisely with enabling Eco Mode on the Shelly Plus Uni sensor at Dec 23, 2025 ~15:40 local time.

Eco Mode reduces device power consumption but triggers a reboot and may change sampling behavior. The spread increase is a measurement-regime artifact, not electrochemical divergence.

Evidence:

How do I know the data is trustworthy?

Several factors support data quality:

  1. Reproducibility — Analysis scripts are provided; results can be regenerated
  2. Raw data access — All source data is publicly available
  3. Evidence mapping — Each claim links to specific data, code, and figures
  4. Methodology documentation — Methods are explicitly described
  5. Known limitations acknowledged — Caveats and uncertainties are stated clearly

The Evidence Map provides full traceability from claims to supporting data.

Safety & Best Practices

What fuse should I use?

Class T fuses are recommended for lithium battery banks:

Bank Capacity Fuse Rating Notes
100 Ah 100–150A Size for expected max load + margin
200 Ah 150–200A Consider wire gauge limits
500 Ah 200–400A Multiple fuses may be needed

Fuses should be placed as close to the battery terminals as possible—ideally within 6 inches of the positive terminal.

Can I charge in cold weather?

Never charge LiFePO₄ below 0°C (32°F).

Charging below freezing causes lithium plating on the anode, which:

Most quality BMS units have low-temperature charging cutoff. If yours doesn’t, add external temperature protection.

Discharging in cold weather is generally safe, though capacity is reduced.

What’s the fire risk?

LiFePO₄ is one of the safest lithium chemistries:

However, any high-capacity battery can be dangerous:

Always use proper fusing, BMS protection, and follow safe handling practices.

Replication

What hardware do I need to replicate this study?

Minimum requirements:

Component Purpose Example
LiFePO₄ battery bank Subject of study Any capacity
Voltage sensor Measurement Shelly Plus Uni, INA226
Data logger Recording Home Assistant, Node-RED
Temperature sensor Environmental correlation Any digital sensor

See Replication Guide for detailed setup instructions.

Can I use different sensors?

Yes, with considerations:

Sensor Type Pros Cons
Shelly Plus Uni Easy setup, Wi-Fi, Home Assistant integration ~10mV resolution, ESP32 ADC limitations
INA226/INA219 High precision, current + voltage Requires microcontroller, more complex
Victron SmartShunt Professional quality, Bluetooth Expensive, proprietary ecosystem
Multimeter logging Very accurate Manual or expensive auto-logging

Higher-resolution sensors will show less quantization noise but may reveal more high-frequency variation.

How long should I monitor?

Minimum useful monitoring periods:

Goal Duration Notes
Basic health check 1 week Sufficient for gross problems
Drift characterization 30+ days Captures relaxation behavior
Seasonal effects 3+ months Temperature variation
Long-term storage viability 6+ months Full discharge curve

This study’s 94+ days provides good drift characterization and storage viability data.

Technical Details

Why is the temperature coefficient positive?

The observed +1.0 mV/°F coefficient is system-level, not pure LiFePO₄ electrochemistry.

It includes:

Pure LiFePO₄ OCV temperature coefficient is typically negative (voltage decreases with increasing temperature) at moderate SOC. The positive observed coefficient suggests measurement-chain effects dominate in this system.

What’s the ADC resolution limit?

The Shelly Plus Uni uses an ESP32 ADC with:

This means voltage changes smaller than ~10mV are not reliably detectable with this sensor. The MA-60s smoothing helps extract trends from quantized data.

Why state-change logging instead of fixed interval?

Home Assistant’s default behavior logs new records only when values change. This:

Advantages:

Disadvantages:

For future work, fixed-interval logging would be preferable for advanced signal analysis.

Still Have Questions?

See Also