RokketBox
Engineering5 min read

Inside the RokketBox Simulation Engine

Under the hood of RokketBox's simulation engine: circuit-domain modelling, radiation loading, and why dense frequency sweeps matter.

Contents

Most subwoofer calculators give you a frequency response curve that looks smooth because they compute a handful of points and draw lines between them. RokketBox models the whole acoustic system as a circuit and solves it across a much denser frequency sweep.

This post explains the circuit approach, why 500 points matter, and the physical effects that go into each solve.

Circuit-domain simulation

Traditional simplified models use closed-form transfer functions from the Thiele-Small framework. They're fast and give reasonable approximations for simple cases, such as a driver in a sealed or vented box with ideal port behaviour.

RokketBox models the entire acoustic system (driver, enclosure, port and radiation load) as an equivalent electrical network. Each physical behaviour (mass, compliance, damping, radiation, losses) maps to a circuit element, and the full network is solved at each frequency point.

Composability

When a new physical effect needs accounting for, such as port turbulence, voice coil losses or radiation loading, it's added to the network without rederiving the maths. The model grows with the physics.

Accuracy at extremes

Simplified models break down at frequency extremes, near resonances and with non-ideal components. The circuit approach solves the full system of interactions, so it handles them correctly.

Multiple enclosure types

Vented and bandpass enclosures are variations of the same network topology: the network is reconfigured and solved, with no separate model per type. Sealed boxes, and the optimizer's quick passes, use the closed-form transfer function, which is exact for a sealed box and much faster.

Why so many points

RokketBox uses a dense logarithmic sweep across the full operating range. The density matters in three places.

500Frequency points per simulation
50Points that can miss narrow features entirely

Narrow features. Impedance peaks, port resonances and the impedance saddle point in vented enclosures are narrow. If the tuning frequency falls between two computed points, the saddle looks like a plateau in place of a sharp dip.

Group delay. Group delay is the negative derivative of phase with respect to frequency. Numerical differentiation needs closely spaced points to give smooth results, and coarse spacing creates noisy, jagged curves.

Port velocity peaks. Velocity can spike sharply near the tuning frequency. Missing the peak with coarse spacing under-reports the maximum, so you might not catch a turbulence problem until you build the box and hear it chuffing.

Bessel/Struve radiation loading

Simplified models treat the driver's radiation load as a constant. The radiation impedance of a circular piston (the cone) is a complex, frequency-dependent function described by Bessel and Struve functions.

At low frequencies, where the cone is small compared to the wavelength, radiation resistance is low and the driver is an inefficient radiator. It rises with frequency until the cone circumference approaches one wavelength, after which the driver radiates more efficiently.

RokketBox computes the full Bessel/Struve radiation impedance and includes it in the circuit. That gives accurate absolute SPL values (as well as relative response shapes) and correct impedance curves at the frequency extremes.

Non-linear effects

Iterative port turbulence

Port velocity and port impedance depend on each other. Turbulence adds resistance, which depends on velocity, and velocity depends on the impedance that sets how much air flows. A circular dependency like that can't be solved in one pass.

At each frequency point the solver starts from the linear result, evaluates the resulting turbulence, feeds it back into the model and repeats until the answer stabilises. This captures the compression at high drive levels, where driving the system harder gives diminishing returns as turbulence absorbs more of the energy.

BL compression

The motor force factor (BL) is at its peak at rest. As the cone moves out or in, the voice coil shifts partly out of the magnetic gap and BL drops, often by 30 to 50% at the rated Xmax.

RokketBox models this progressive loss with a smooth falloff curve derived from the driver's excursion limits. The SPL prediction then reflects real power compression and rising distortion at high excursion, where a standard Thiele-Small simulator assumes BL stays constant.

Lossy voice coil inductance

Voice coil inductance (Le) is a non-ideal component. Real coils show frequency-dependent losses from eddy currents in the pole piece and shorting rings. RokketBox uses an inductance model that captures the correct high-frequency impedance rise, which is shallower than an ideal inductor predicts and matches what an impedance analyser shows.

Most simplified models ignore Le or treat it as ideal, and both give incorrect impedance curves above a few hundred Hz.

What you see in the simulator

Every plot (SPL, impedance, group delay, excursion, port velocity and transfer function) comes from the same circuit-domain solve. There's no separate model per plot; they're different views of the same physics.

When you change box volume, tuning or port area, the engine resolves the whole sweep with the full circuit model. On modern hardware that's fast enough for real-time interaction as you move the sliders.

Common questions

How does RokketBox simulate subwoofer frequency response?

RokketBox uses circuit-domain simulation where the entire acoustic system (driver, enclosure, port, radiation load) is modelled as an equivalent electrical network. It computes a dense logarithmic frequency sweep, solving the full network at each point. This approach is more accurate than simplified closed-form models, especially at frequency extremes and near resonances.

Why does RokketBox use so many frequency points?

Dense frequency spacing captures narrow features that coarse spacing misses: impedance peaks, the tuning frequency saddle point, port velocity spikes, and accurate group delay. With only a few dozen points, critical features can fall between computed values and produce misleading curves.

What is circuit-domain simulation for loudspeakers?

Circuit-domain simulation maps each physical behaviour of a loudspeaker system to a corresponding network element. The full system is solved as an interconnected network, which is composable (new effects are added as elements rather than requiring new maths) and accurate at frequency extremes where simplified models break down.

simulationcircuit-domainengineeringtechnicalrokketbox

Related posts