Helix at Home (HaH) Processor 🎸

TODO

• updated documentation
• Scripting documentation
  • build.tcl and program_sof.tcl work but not program_jic.tcl

Overview

Helix at Home (HaH) is a synthesizable multi-effects guitar processor implemented in SystemVerilog on the DE1-SoC FPGA. It uses the on-board audio codec ADC and DAC to process live audio input from an electric guitar through a configurable chain of digital audio effects.

This project was created to explore audio DSP in hardware, with a focus on real-time streaming, pipelined processing, and FPGA-based system design.

The design was initially forked from the audio-loopback-only example in De1-SoC-Verilog-Audio-HW-FX by Navraj Kambo, which provided basic audio codec configuration and loopback functionality.

---

Audio Processing Chain

Audio is processed sequentially through the following effect chain:

Gain → Gate → EQ → Compressor → Distortion → EQ → Chorus → Gain → Delay → Reverb → Gain

---

Hardware Interface (DE1-SoC)

User Controls

• Switches & Buttons

  • Select effects and parameters
  • Increment / decrement parameter values

• HEX Displays

  • Show selected effect and parameter
  • Display current parameter value

• LEDs

  • Visual feedback for parameter values and selection state

Audio I/O

• Input: DE1-SoC LINE IN (electric guitar)
• Output: DE1-SoC LINE OUT

---

Audio & System Specifications

Feature	Value
Input	Mono (left channel duplicated for guitar input)
Output	Stereo
Audio Resolution	16-bit
Sampling Rate	48 kHz
Internal Clock	50 MHz
Max FX Slots	16
Parameters per FX	Up to 8
Latency	under 1ms

---

Block Diagram

---

Usage Instructions

The following sections outline how to use the board

Hardware + Software Needed

Hardware

• x1 DE1 SoC Board
  • x1 USB Blaster Cable
  • x1 Power Cable
• x2 AUX Cable
  • x1 whatever you connect to output, can be speacker, headphone, audio interface, etc
• x1 1/4 inch to AUX adapter
• x1 Guitar or your choice

Software

• Quartus Prime
  • Quartus Prime Version 18.1.0 Build 09/12/2018 SJ Lite Edition was used to develop this project

How to install and run on the board

1. Clone this repo into the machine with Quartus Prime
2. Open quartus prime and use the Programmer tool the program the board with the ./AudioFX.sof that is in the root directory of this repo, should not need any changes to project file or repo to do this
3. Click KEY[0] to reset the board
4. Connect one of the AUX cables to the green "LINE IN" port at the top left of the board and then connect the 1/4 inch to AUX adapter to the other end then plug that into you guitar
5. Connect the other AUX cable to the blue "LINE OUT" port at the top left of the board and then connect it to the output device
6. PLAY

How to create .jic file for power persistant design

1. Compile via quartus to generate the .sof file normally
2. Go to File > Convert Programming File in Quartus and then create the .jic file with the following settings and then click Generate

• Configuration device: EPCQ128A
• Flash Loader: 5CSEMA5
• SOF Data: .sof file generated in step 1 (should be AudioFX.sof)
• Name: Whatever you want it to be (right now, I set the latest one to be output_file.jic)

3. Plug in the DE1 SoC and open the Programmer tool then click Auto Detect and in the popup select 5CSEMA5
4. Right click on the second chip in the GUI and click Change File and select the file generated in step 2
5. Check the Program/Configure box for the selected box and then click Start

• This takes a few minutes to compile

How to control FX

HEX5-HEX2 display the currently selected FX number and parameter, with the exact mappings shown in the table below; values not listed correspond to no parameter. The current parameter value (ranging from 0 to 255) is visualized on LEDR[9:0] as a bar graph, where only LEDR[9] lit represents a value of 0 and LEDR[8:0] illuminate as the value increases. KEY[2] and KEY[3] are used to decrease and increase the selected parameter value, respectively. The active FX module is selected using SW[9:6], while SW[4:2] choose the parameter within that FX, and SW[0] mutes the output signal with HEX0 being a line if it is muted. To save a set of parameters, press KEY[1] and a b should appear on HEX1 briefly. Once the b is gone, the parameters have been saved. To load saved parameters, flip SW[1] up and down.

FX Mapping Table

FX	Parameter	FX Number	Parameter Number
Gain1	fx_gain	F0	P0
Gate	fx_threshold	F1	P0
Gate	fx_attack	F1	P1
Gate	fx_release	F1	P2
Gate	fx_knee	F1	P3
Gate	fx_depth	F1	P4
EQ1	fx_sub_gain	F2	P0
EQ1	fx_low_gain	F2	P1
EQ1	fx_mid_gain	F2	P2
EQ1	fx_high_gain	F2	P3
Compressor	fx_threshold	F3	P0
Compressor	fx_ratio	F3	P1
Compressor	fx_attack	F3	P2
Compressor	fx_release	F3	P3
Compressor	fx_input_gain	F3	P4
Compressor	fx_makeup_gain	F3	P5
Compressor	fx_mix	F3	P7
Distortion	fx_drive	F4	P0
Distortion	fx_makeup_gain	F4	P1
Distortion	fx_mix	F4	P7
EQ2	fx_sub_gain	F5	P0
EQ2	fx_low_gain	F5	P1
EQ2	fx_mid_gain	F5	P2
EQ2	fx_high_gain	F5	P3
Chorus	fx_rate	F6	P0
Chorus	fx_depth	F6	P1
Chorus	fx_mix	F6	P7
Gain2	fx_gain	F7*	P0*
Delay	fx_time	F8	P0
Delay	fx_feedback	F8	P1
Delay	fx_mix	F8	P7
Reverb	fx_size	F9	P0
Reverb	fx_damping	F9	P1
Reverb	fx_mix	F9	P7
Gain2	fx_gain	F10	P0

NOTES:

• You should see the FX Number values on HEX5-HEX4, with HEX5 always being F
• You should see the Parameter Number values on HEX3-HEX2, with HEX3 always being P
• The order of FX numbers represents the order the FX are connected in series, with the output of F0 being connected to the input of F1, and so on
• Gain2 should be modulated by the expression pedal
• Descriptions for what the parameters do are in the next section

Implemented Effects & Parameters

Gain

Applies a gain multiplier to the signal.

Parameters

• fx_gain: Gain multiplier (0–255), where 32 = unity. Values above 32 amplify; values below 32 attenuate.

---

Gate

Noise gate with a soft-knee and adjustable floor gain. Signals above the threshold pass through at unity; signals below are attenuated toward the depth floor. The knee region smoothly transitions between open and closed rather than switching abruptly.

Parameters

• fx_threshold: Gate open level (scaled to the full 16-bit signal range)
• fx_attack: Speed at which the gate opens after the signal exceeds the threshold (0 = instant, 255 = slowest)
• fx_release: Speed at which the gate closes after the signal falls below the threshold (0 = instant, 255 = slowest)
• fx_knee: Soft-knee half-width — widens the transition region around the threshold (0 = hard switch, 255 = widest knee)
• fx_depth: Gain floor when the gate is fully closed (0 = full mute, 255 = unity — effectively bypasses gating)

---

EQ

4-band EQ that splits the signal into sub, low, mid, and high frequency bands and applies independent gain to each.

Parameters

• fx_sub_gain: Sub-band gain (128 = unity)
• fx_low_gain: Low-band gain (128 = unity)
• fx_mid_gain: Mid-band gain (128 = unity)
• fx_high_gain: High-band gain (128 = unity)

---

Compressor

Peak-envelope-based compressor with lookahead that reduces signal levels above a threshold while leaving lower levels unchanged. Includes independent input and output gain stages and a parallel dry/wet blend for New York-style parallel compression.

Parameters

• fx_threshold: Compression onset level (0–255, scaled to full 16-bit range)
• fx_ratio: Compression ratio — 0 = 1:1 (no compression), 255 = maximum ratio
• fx_attack: Time before gain reduction begins after exceeding the threshold (upper nibble controls slew rate)
• fx_release: Time before gain reduction is released after falling below the threshold (upper nibble controls slew rate)
• fx_input_gain: Pre-compression input gain (64 = unity)
• fx_makeup_gain: Post-compression output gain (64 = unity)
• fx_mix: Dry/wet blend (0 = fully dry, 255 = ~99.6% wet)

---

Distortion

Amp-style distortion that models the full signal chain of an overdriven guitar amplifier through seven processing stages:

1. Pre-emphasis — A first-difference high-shelf filter (emph = x + (x − x_prev) >> 2) boosts the presence band (~3 kHz) before the signal hits the clipping stage, adding bite and pick attack to the distorted tone.

2. Drive — Multiplies the pre-emphasized signal by a gain in the range 1× to 32.875× (drive_gain = 256 + fx_drive × 32). Higher drive pushes more of the signal into the clipping region.

3. Asymmetric bias — A fixed +5% full-scale DC offset is added before clamping. This shifts the clipping threshold so positive and negative half-cycles clip at different levels, introducing even-order harmonics that give the distortion a warmer, more tube-like character.

4. Soft clip (tanh approximation) — The biased signal is passed through a 3rd-order polynomial approximation of tanh(x), which smoothly compresses peaks rather than hard-squaring them:

           {  +2/3              , x ≥ +1
   f(x) =  {  x − x³/3          , −1 < x < 1
           {  −2/3              , x ≤ −1
   

   Division by 3 is approximated as (x + x>>2 + x>>4 + x>>6) >> 2 (error ≤ 0.39%, inaudible).

5. Wet/dry mix — Blends the clipped signal with the original dry input (0 = fully dry, 255 = ~99.6% wet), enabling parallel or blended distortion tones.

6. Makeup gain — Compensates for the level reduction caused by clipping (128 = unity).

7. Cabinet simulation — Two cascaded one-pole IIR low-pass filters (fc ≈ 4.4 kHz @ 48 kHz) roll off the harsh ultrasonic content produced by hard clipping, approximating the frequency response of a guitar speaker cabinet.

Latency: 6 samples.

Parameters

• fx_drive: Amount of gain into the clipping stage (0 = 1× / unity, 255 ≈ 32.875×)
• fx_makeup_gain: Output level after clipping and cabinet simulation (128 = unity)
• fx_mix: Dry/wet blend (0 = fully dry, 255 = ~99.6% wet)

---

Chorus

Creates a thicker sound by duplicating and delaying the input signal, then modulating the delay time with a triangle-wave LFO before mixing it back with the original. Two quadrature voices (0° and 90°) produce a lush stereo spread.

Parameters

• fx_rate: LFO frequency (0 = very slow, 255 = fastest)
• fx_depth: LFO modulation depth (0 = no modulation, 255 = maximum)
• fx_mix: Dry/wet blend (0 = fully dry, 255 = ~99.6% wet)

---

Delay

Implements an echo effect using delay lines with feedback.

Parameters

• fx_time: Delay time (larger value = longer delay, scaled across the full delay buffer range)
• fx_feedback: Number of echo repeats (0 = single echo, higher values = more repeats; internally capped to prevent runaway feedback)
• fx_mix: Dry/wet blend (0 = fully dry, 255 = ~99.6% wet)

---

Reverb

Implements a Schroeder Reverberator using parallel damped feedback comb filters followed by series all-pass filters to simulate room reverberation.

Parameters

• fx_size: Room size / decay time — scales all comb filter delays (0 = smallest room / shortest tail, 255 = largest room / longest tail)
• fx_damping: High-frequency damping of the reverb tail (0 = bright / full HF content, 255 = dark / heavily damped)
• fx_mix: Dry/wet blend (0 = fully dry, 255 = ~99.6% wet)

---

Tools, Technologies & Platform

• FPGA Board: DE1-SoC
• HDL: SystemVerilog
• Audio Codec: DE1-SoC on-board codec
• Synthesis & Programming: Intel Quartus
• Simulation: ModelSim