ONYX Battery Design

Intro

Battery discussions around the ONYX are typically reduced to voltage, capacity, and cell choice. In practice, these do not determine performance under load.

This post defines what 72V, 84V, and 96V batteries represent at the system level, how high-current packs are constructed using tabless cylindrical cells, and what determines whether a pack can realistically support 300A within the ONYX form factor.

This is not a general battery guide. It is a constrained analysis of a specific class of battery design.

At this level, performance is governed by system behavior under load, specifically internal resistance and current path design.

Summary

• Battery performance is defined by internal resistance, not nominal voltage
• Increasing voltage reduces current but shifts stress into other parts of the system
• Tabless cells shift the limitation from the cell to the pack architecture
• 300A capability is determined by parallel group design and interconnect strategy
• Nickel-only interconnects are not viable at this current level
• BMS, wiring, and connectors frequently become the system bottleneck
• Two batteries with identical voltage and capacity can behave differently under load due to resistance distribution
• Within a fixed enclosure, increasing voltage reduces parallel capacity and increases system stress
• Current defines component stress, while voltage determines how much power that current represents

Battery Selection Context

These recommendations are based on system constraints, not marketing specifications or nominal compatibility.

Design Constraints

• Must fit ONYX RCR 72V battery bay
• Must support 300A continuous current
• Must use tabless cylindrical cells only
• Must remain electrically stable under sustained load
• Must not rely on peak-only current ratings

These constraints are non-negotiable and define the design space.

These constraints eliminate most theoretical designs before they are ever built.

Every improvement in one part of the pack introduces a constraint in another.

Physical Pack Design and Current Paths

The enclosure, interconnects, insulation, output wiring, BMS placement, and assembly quality constrain usable current. Those details are now covered together in Battery Enclosure, Busbar, and Insulation Design. The electrical analysis below retains the voltage, current, parallel-group, sag, and sustained-load relationships. A current target is not proof of tested capacity.

System Overview

This post is primarily a system-level analysis of high-current battery design within the ONYX RCR platform.

However, where necessary, it uses a concrete design example to illustrate real constraints and tradeoffs.

That reference design is:

• a custom 20s9p tabless 21700 pack
• targeting ~45Ah capacity
• designed around a slightly increased enclosure height

This example is not the only valid configuration.

It is used to make constraints visible:

• how enclosure volume affects parallel count
• how interconnect space competes with capacity
• how thermal and electrical limits emerge in real layouts

All principles in this post apply generally.

The 20s9p configuration is used to show how those principles play out in a real design.

These values describe the target configuration, not guaranteed performance.

Most packs that claim high current capability do not sustain it under real load conditions.

Power, Current, and Voltage Relationship

Power is defined as:

• Voltage x Current = Watts

At a fixed power level:

• higher voltage reduces current
• lower voltage increases current

Example:

• 72V system at 10kW = ~139A
• 84V system at 10kW = ~119A
• 96V system at 10kW = ~104A

Lower current reduces:

• resistive losses
• heat generation
• voltage sag

However:

• all current still flows through the same system components
• resistance remains the limiting factor

Increasing voltage improves efficiency, but does not eliminate system constraints.

Reducing current reduces losses across every component in the system.

This relationship defines how voltage choices reshape the entire pack architecture.

Current Does Not Scale Linearly With Voltage

A 300A load does not represent the same operating condition across 72V, 84V, and 96V systems.

At 300A:

• 72V corresponds to ~21.6kW
• 84V corresponds to ~25.2kW
• 96V corresponds to ~28.8kW

That is not the same stress condition.

At higher voltage, the same current corresponds to higher total power, higher thermal load, and greater system stress.

Conversely, for the same power output, higher voltage reduces required current and lowers resistive losses.

This distinction is critical:

• current limits define component stress
• voltage determines how much power that current represents

In practice, higher voltage systems should be designed to reduce current demand, not maintain it.

A 96V system does not need to chase the same amperage as a 72V system to produce high power.

300A is a stress ceiling, not a performance goal.

Most systems are not limited by how much current they can produce, but by how efficiently they can deliver it.

What 72V, 84V, and 96V Actually Represent

A 72V battery is a 20-series configuration. An 84V battery is a 23-series configuration. A 96V battery is a 26-series configuration. Each series group increases voltage, not capacity.

Capacity is determined by the number of cells in parallel. A 45Ah pack is not a single entity. It is the sum of parallel cell groups.

Energy is defined as:

• Voltage x Capacity = Watt-hours

Two packs labeled with the same voltage and capacity can have identical energy but different performance under load due to:

• internal resistance
• interconnect design
• cell chemistry
• thermal behavior

Increasing series count within a fixed enclosure reduces available space for parallel groups and increases system voltage stress.

Increasing voltage within a fixed enclosure trades capacity and stability for electrical stress.

Voltage class changes pack architecture, not just pack labeling.

Tabless Cell Architecture

This design uses only the following tabless cylindrical cells:

• Tenpower 50XG
• EVE 50PL
• Reliance RS50
• Ampace JP50

Tabless cells differ from traditional cylindrical cells in that current is distributed across the full electrode surface rather than a single tab point.

This results in:

• lower internal resistance
• more uniform current distribution
• reduced localized heating
• improved high-current performance

At this level, the cell is no longer the limiting factor.

Tabless design improves cell-level performance, but does not eliminate limitations in interconnects, BMS, or overall pack design.

This shifts the limiting factor away from the cell and into the pack architecture.

Bottleneck Shift

In tabless-based designs:

• cells capable
• pack architecture limiting

The limiting factors become:

• busbars
• connection resistance
• BMS current handling
• output wiring
• connector interfaces

System performance is limited by the weakest component in the current path.

What looks like a cell problem is often an interconnect, BMS, or wiring problem.

Builder Archetypes

Battery packs built with similar cells and specifications can behave very differently in practice.

Differences in performance are often less about the components themselves and more about how the system is designed and executed.

Spec-Driven Builds

These packs are typically defined by:

• focus on nominal specifications (voltage, capacity, peak current)
• heavy reliance on nickel strip for interconnects
• limited attention to resistance distribution
• minimal consideration for sustained load behavior

They often perform adequately under light or short-duration use but show:

• significant voltage sag under load
• rapid heat buildup
• inconsistent performance over time

The limitations are not usually in the cells themselves, but in how current is managed across the pack.

Transitional Builds

These designs show a better understanding of system behavior:

• improved interconnect strategy (often hybrid nickel/copper)
• more attention to parallel group sizing and current distribution
• basic consideration of thermal behavior
• higher quality assembly and layout

These packs can perform well under moderate load and may approach high-current targets, but often still exhibit:

• localized hotspots
• uneven current distribution
• performance drop under sustained load

They are functionally capable, but not fully optimized.

System-Driven Builds

At this level, the pack is treated as a complete electrical system, not a collection of parts.

These builds are characterized by:

• interconnect design centered around low resistance and current distribution
• deliberate control of current paths and symmetry
• balanced positive and negative pathways
• thermal behavior considered as part of the electrical design
• enclosure used as a structural and thermal component

These packs do not rely on peak ratings. They are designed for:

• sustained load stability
• predictable behavior under stress
• minimal voltage sag relative to load
• controlled heat generation and dissipation

At this level, differences in performance are not subtle. They are immediately apparent under load.

System-Level Reality

Two packs can share:

• the same voltage
• the same capacity
• the same cells

…and still perform completely differently.

The difference is not the specification.

It is the design.

Parallel Group Design

Current capability is determined by the number of cells in parallel.

At a 300A target:

• current per cell must remain within safe discharge limits
• thermal load must be distributed across the group

Example considerations:

• theoretical current per cell is not usable current
• derating is required for sustained load
• uneven current distribution increases localized heating

Parallel group design directly determines:

• voltage sag
• thermal stability
• lifespan

Within a fixed enclosure, increasing voltage reduces parallel count, increasing stress on each cell and raising system resistance.

Increasing series count does not increase total energy unless capacity also increases. It reduces available parallel groups within a fixed volume.

For a fixed power target, higher voltage reduces required current, which changes what an optimal parallel count looks like.

Series Configuration

A 20s, 23s, or 26s configuration defines voltage behavior:

• 20s full charge: ~84V, nominal: ~72V, cutoff: ~60V
• 23s full charge: ~96.6V, nominal: ~83V, cutoff: ~69V
• 26s full charge: ~109.2V, nominal: ~94V, cutoff: ~78V

Under load:

• voltage drops due to internal resistance
• controller behavior changes based on real-time voltage
• performance decreases as sag increases

Voltage is not static. It is load-dependent.

Higher series configurations increase voltage stress on BMS components, wiring insulation, and connectors.

Higher voltage increases electrical stress across all components in the system.

The choice of voltage class changes the load case for the entire pack.

BMS Constraints

A BMS rated for 420A does not guarantee stable operation at 300A.

All pack current passes through the BMS, making it a critical point of resistance and heat generation.

Limitations include:

• MOSFET thermal limits
• trace and internal path resistance
• voltage drop across the BMS
• thermal throttling under sustained load

Failure modes:

• sudden cutoff under load
• heat-induced shutdown
• progressive performance loss under sustained current

At high current, the BMS often becomes the primary system bottleneck.

The BMS is often the highest resistance component in the discharge path.

A 300A condition at 96V is a much higher total load state than 300A at 72V.

Thermal Behavior

Heat is generated by:

• internal cell resistance
• interconnect resistance
• BMS losses

Thermal behavior determines:

• sustained output capability
• safety margins
• lifespan

A pack that can deliver 300A briefly may not sustain it without overheating.

Thermal saturation occurs when:

• heat generation exceeds dissipation
• internal temperatures continue rising under constant load

As temperature increases, resistance increases, further accelerating heat generation.

Thermal behavior is also influenced by enclosure design.

A rigid metal enclosure:

• spreads heat more evenly
• reduces localized hotspots
• improves structural stability under thermal expansion

Poor enclosure design traps heat and accelerates thermal saturation.

Thermal behavior feeds back into electrical performance through resistance.

Heat is the physical manifestation of electrical inefficiency.

Common Failure Archetypes

These failure modes are system-level issues, not isolated component failures.

These failure modes are all expressions of resistance, heat, and current imbalance.

Spec Sheet Pack:

• strong nominal specs
• poor real-world performance due to resistance

Nickel Bottleneck Pack:

• limited by interconnects, not cells

BMS-Limited Pack:

• cutoff under load despite capable cells

Thermally Saturated Pack:

• performance drops after short high-load periods

Diagnostics

Common symptoms and causes:

Voltage sag under load:

• high internal resistance
• insufficient parallel groups

Sudden cutoff:

• BMS limit reached
• voltage drop below threshold

Heat buildup:

• interconnect resistance
• poor current distribution

Performance inconsistency:

• uneven cell loading
• connection variability

These symptoms translate directly to riding behavior:

• voltage sag -> reduced acceleration and top speed
• sudden cutoff -> loss of power under load
• heat buildup -> performance drop over time
• inconsistency -> unpredictable throttle response

Electrical issues present as performance issues.

Diagnosis should focus on identifying where resistance is introduced in the system.

All diagnostic symptoms are manifestations of resistance and heat.

Diagnosis is the process of locating resistance within the system.

Buy vs Build

Buying:

• consistent
• validated design
• limited visibility into construction

Building:

• customizable
• potentially higher performance
• increased risk if improperly designed

Building is only justified when:

• performance requirements exceed available packs
• design constraints are fully understood
• construction quality can be maintained

Most performance issues in custom packs come from execution, not design intent.

Voltage Sag Under Load

Voltage sag is one of the clearest indicators of real-world battery performance and total system resistance.

Voltage sag is the drop in pack voltage when current is drawn.

It is caused by resistance across the entire system:

• cell internal resistance
• busbar and interconnect resistance
• BMS resistance
• wiring and connector losses

Sag is not a defect. It is a measurable behavior of the system under load.

What Sag Looks Like in Practice

Using fully charged packs under a high load (e.g. ~300A), sag can be roughly interpreted as:

These are not absolute values, but they provide a practical reference for evaluating pack behavior.

• Low sag -> strong pack, low resistance, stable output
• Moderate sag -> acceptable performance, but not optimized
• High sag -> excessive resistance, heat buildup, and reduced usable power

At higher voltage, the same percentage sag represents a larger absolute voltage drop, but a lower percentage loss relative to total system voltage.

What Causes Sag

Every component in the current path resists flow.

At 300A:

• even small resistance values create measurable voltage drop
• losses compound across the system

Higher resistance results in:

• larger voltage drop
• increased heat
• reduced usable power

Why Sag Matters

Performance is based on real-time voltage, not nominal voltage.

Under load:

• controller sees reduced voltage
• power output decreases
• acceleration and top speed are affected

Two identical packs on paper can behave differently due to sag.

Parallel Groups and Sag

Increasing parallel count reduces sag:

• more cells share the load
• lower current per cell
• reduced voltage drop

Insufficient parallel groups result in:

• higher per-cell stress
• greater voltage collapse under load

Voltage vs Sag

Higher voltage systems reduce current for the same power.

This results in:

• less total voltage drop across the system
• improved efficiency under load

However:

• poor interconnect design will still cause sag
• resistance remains the limiting factor

Increasing voltage does not eliminate sag. It reduces its impact.

Thermal Link

Sag and heat are directly related:

• voltage drop = energy lost as heat
• higher sag = higher temperature

Thermal saturation increases resistance further, creating a feedback loop.

Real-World Behavior

Sag is most visible:

• during hard acceleration
• at low state of charge
• in poorly designed packs

A well-built pack:

• maintains stable voltage under load
• recovers quickly after load is removed

A poorly built pack:

• drops voltage aggressively
• feels inconsistent
• limits usable performance

System Constraint

Sag is not determined by one component.

It is the result of:

• cell selection
• parallel design
• interconnect quality
• BMS capability
• wiring and connectors

The pack performs as well as its total resistance.

What a Well-Designed Pack Feels Like Under Load

A well-designed pack does not just look better on paper. It behaves differently in use.

Under repeated acceleration and sustained load, a strong pack feels:

• consistent rather than variable
• stable rather than abrupt
• predictable across state of charge
• resistant to sudden power collapse

A key characteristic of a strong pack is how it behaves as state of charge decreases.

At high state of charge (100% -> ~75%):

• peak performance is available
• voltage remains high under load
• acceleration is strong and immediate

At mid state of charge (~75% -> ~50%):

• performance remains consistent
• slight reduction in peak output
• minimal change in overall feel if the pack is well designed

At lower state of charge (~50% -> ~35%):

• voltage sag becomes more noticeable
• peak power is reduced
• the system becomes more sensitive to load

Below ~35%:

• available power drops more rapidly
• voltage collapses more easily under load
• performance becomes increasingly limited

A well-designed pack maintains usable, predictable performance through most of its discharge range and degrades gradually.

A poorly designed pack may feel strong at full charge but becomes inconsistent or unstable much earlier, often due to resistance and poor current distribution.

That difference is not subtle under real use. It is the result of resistance, current path quality, and thermal behavior across the entire system.

Energy, Runtime, and Range

Watt-hours define how much energy a battery stores, not how far the bike will travel.

Energy is calculated as:

• Voltage x Capacity = Watt-hours

A ~72V 45Ah pack:

• ~3240Wh

An ~84V 45Ah pack:

• ~3780Wh

A ~96V 45Ah pack:

• ~4320Wh

This is total stored energy.

However, stored energy is not the same as usable energy under load.

A portion of that energy is always lost as heat due to resistance throughout the system:

• cell internal resistance
• interconnect losses
• BMS losses
• wiring and connector losses

Higher current increases these losses.

Real-world runtime is always lower than theoretical capacity, and the gap increases as load increases.

Advertised capacity describes stored energy. It does not describe how efficiently that energy is delivered.

Energy vs Power

Energy determines how long the system can run.

Power determines how quickly that energy is used.

At higher power levels:

• energy is consumed faster
• losses increase due to higher current
• usable energy decreases due to heat

This is why high-power riding reduces both runtime and efficiency.

A battery does not just “drain faster” at higher power. It also becomes less efficient while doing so.

Runtime

Runtime is determined by average power draw, not peak power.

Example (72V ~3240Wh pack):

• 3000W average -> ~1 hour
• 1500W average -> ~2 hours
• 6000W average -> ~30 minutes

These are idealized values.

In practice:

• higher current increases losses
• voltage sag reduces usable output
• thermal behavior further reduces efficiency

As a result, actual runtime is always lower than calculated values, especially under sustained high load.

Real Power Usage

Actual riding power is highly variable.

Typical ranges:

• light cruising: 1500W to 3000W
• normal riding: 3000W to 6000W
• aggressive riding: 6000W to 10000W+

Short bursts may exceed these values, but sustained power is what determines runtime and thermal behavior.

Higher sustained power:

• reduces runtime
• increases heat
• amplifies system inefficiencies

Range Is Not a Battery Spec

Range is not determined by voltage or capacity alone.

It depends on:

• average power draw
• speed
• terrain
• rider weight
• aerodynamic drag
• throttle behavior

Two riders using the same battery can see significantly different range because they are using energy at different rates.

Range is a usage outcome, not a battery specification.

Voltage, Current, and Efficiency

Higher voltage systems:

• reduce current for the same power
• reduce resistive losses across the system
• improve efficiency under load

However:

• total stored energy still defines total runtime
• higher voltage does not increase runtime unless total watt-hours increase

Efficiency gains from higher voltage come from reduced current, not increased energy.

Poor system design can eliminate these gains entirely.

Load, Losses, and Usable Energy

As load increases:

• current increases
• resistive losses increase
• more energy is converted to heat instead of motion

This means:

• high-power riding consumes more energy
• and wastes more energy at the same time

Usable energy decreases as load increases.

This is why two identical batteries can produce different real-world range depending on how they are used.

Practical Interpretation

A battery does not provide a fixed distance.

It provides a fixed amount of energy, delivered with varying efficiency depending on load.

Higher power:

• increases energy consumption
• increases system losses
• reduces total usable output

Higher voltage:

• reduces current for the same power
• improves efficiency
• reduces loss per unit of power

But:

• energy is still finite
• and system design determines how efficiently that energy is delivered

The difference between a good pack and a poor one is not just how much energy it stores.

It is how much of that energy actually reaches the controller under load.

Thermal Imaging and Hotspot Detection

See Battery Thermal Imaging and Hotspot Detection for ONYX RCR for the complete reference.

Final Advice

At this level, battery design is not defined by voltage, capacity, or cell selection.

Those are starting points.

Performance is defined by what happens under load:

• how current flows
• where resistance is introduced
• how heat accumulates
• how the system behaves over time

Tabless cells increase potential, but they do not remove constraints.

They shift the limitation away from the cell and into the system.

Two packs with identical specifications can behave completely differently because:

• resistance is distributed differently
• current paths are not equivalent
• thermal behavior diverges under load

At high current, performance is not defined by peak capability.

It is defined by what the system can sustain.

Small inefficiencies scale into system-level limitations:

• resistance becomes voltage drop
• voltage drop becomes heat
• heat reduces usable performance

The system does not fail instantly.

It degrades under load.

And that degradation defines real performance.

The pack behaves as a single system, limited by its weakest path.

Design is the process of deciding where those limits occur.

You cannot eliminate constraints.

You can only control:

• where resistance exists
• where heat accumulates
• how the system reaches its limits

Performance is not a specification.

It is the result of those decisions.

A battery is not defined by what it can do once. It is defined by what it can continue to do under load.