02 – Analog to Digital

Translating the circuit into DSP.

Rather than recreating the effect as a black box, the CE-2 circuit was broken down into its functional building blocks. Each signal-processing stage was analyzed individually and translated into a corresponding digital module in RNBO.

Functional block diagram of the CE-2 signal path:

Analog circuit stages and their corresponding RNBO-implementation:

03 – Circuit Simulation

From analog components to digital filters.

The analog filter stages were recreated and simulated in KiCad to derive validated frequency responses for the digital model. The Pre-Emphasis stage serves as an example of how the analog circuit was translated into four sample-rate-aware digital filters.

Pre-Emphasis Filter

Although the circuit initially resembles a shelving filter, simulation revealed an additional high-frequency roll-off caused by C4 and R6 in the op-amp feedback path. The digital model therefore combines a shelving filter with a second-order low-pass to more closely reproduce the simulated analog response.

Sample-Rate Adaptation

The analog roll-off extends beyond the Nyquist frequency at common sample rates such as 44.1 and 48 kHz. To emulate the Pre-Emphasis response as precisely as possible at each sample rate while remaining aliasing-free, the RNBO implementation automatically selects the appropriate filter structure and recalculates its coefficients whenever the host sample rate changes.

Other Filter Stages

The remaining filter stages were implemented using the same simulation-based approach. Their digital responses closely reproduce the frequency characteristics derived from the analog circuit simulations.

Anti-Aliasing

Reconstruction

De-Emphasis

04 – Measurement

Measuring the modulation behavior.

While the filter stages could be derived from circuit analysis and simulation, the CE-2’s actual modulation behavior was determined experimentally. A dedicated measurement setup was used to derive the delay range and LFO rate directly from the hardware.

Measurement Setup

A periodic impulse signal was sent through a reamping box into a CE-2W. In stereo mode, the pedal outputs the processed wet signal and the unprocessed dry signal separately. The instantaneous delay was then determined from the sample offset between corresponding impulses in both recordings.

Tracking the BBD delay over time.

Repeating the measurement across multiple LFO cycles reveals the time-varying delay produced by the BBD. Its extrema and period were used to derive the modulation depth, LFO rate and the pedal’s base delay.

Derived Modulation Parameters

Notably, the measured base delay of 6.41 ms is shorter than commonly associated with chorus effects, indicating that the characteristic sound of the CE-2 depends on more than delay time alone.

05 – Digital DElay Structure

Recreating the CE-2’s delay structure.

At the core of the CE-2, the MN3101 clock generator controls the delay produced by the 1024-stage MN3007 bucket-brigade device. In the digital model, this clock-driven structure is reduced to its signal-processing function: a continuously modulated delay line controlled by the measured behavior of the original pedal.

LFO – Generating the modulation signal

A triangular LFO recreates the modulation behavior measured from the hardware. Rate controls its frequency, while Depth scales the modulation range. Filtered noise is added to introduce subtle irregularities into the otherwise ideal waveform.

MN3101 – Translating modulation into delay

In the analog CE-2, the MN3101 converts the LFO voltage into the clock signal driving the BBD. In the digital model, the measured modulation range and base delay are used directly to derive the instantaneous delay time. The corresponding BBD clock frequency is retained as a derived quantity rather than serving as the mechanism that physically clocks the delay line.

MN3007 – Applying the variable delay

The MN3007’s 1024-stage bucket-brigade structure is replaced by a variable digital delay line controlled by the preceding modulation stage. Rather than simulating each individual BBD stage, the model reproduces its time-varying delay behavior directly. Filtered noise is added at the output to approximate the measured noise floor of the analog device.

07 – Validation

How close is the digital model?

The completed model was validated against measurements of the CE-2W at three representative parameter settings. Frequency responses were compared under matching conditions to evaluate how closely the digital implementation reproduces the behavior of the analog hardware.

Minimum Settings

Mid Settings

Maximum Settings

Result:
The model reproduces the CE-2W's characteristic frequency-dependent comb-filter structure across the tested parameter range. Remaining deviations primarily concern individual notch positions, phase behavior and nonlinear distortion.

Further Validation

Beyond the frequency-response comparison shown here, the full evaluation also covered phase response, nonlinear distortion (THD), and operation at 44.1 kHz.

08 – Listen

Hear the reconstruction.

The final RNBO model runs directly in the browser. Explore six reconstructed CE-2W settings, adjust Rate and Depth, and compare the model directly with the dry signal and the original hardware recordings.

09 – Documentation

Want the full technical details?

This project was developed as my Bachelor’s thesis in Audiovisual Media at Hochschule der Medien Stuttgart. The full thesis, including circuit analysis, DSP implementation and evaluation, is available upon request.

Max / RNBO · KiCad · MATLAB · JUCE · C++ · AU/VST3