A method for decoupling control of audio multi-dimensional features of plasma lighting

Abstract

This invention relates to the fields of plasma discharge control and audio signal processing technology, and particularly to a method for decoupling control of multi-dimensional audio features in plasma lighting. The method includes the following steps: S1: Acquiring an external audio signal with a sampling rate of 44.1kHz or 48kHz and a quantization precision of 16bit or 24bit; S2: Performing multi-level preprocessing on the audio signal, including DC removal, power frequency notch filtering, pre-emphasis, and automatic gain control; S3: Extracting three core features in parallel from the preprocessed audio signal: instantaneous amplitude features, beat timing features, and energy distribution features of at least three different frequency bands. This invention maps the instantaneous amplitude, beat timing, and frequency band energy of the audio signal to the plasma's basic discharge power, discharge power pulse, and discharge drive frequency, respectively, achieving decoupling control in three dimensions. This allows for independent adjustment of the plasma gas motion pattern, brightness variation, and beat response, resulting in rich variations in luminous efficacy.

Description

Technical Field

[0001] This invention relates to the fields of plasma discharge control and audio signal processing technology, and in particular to a method for decoupling control of multidimensional audio features of plasma illumination. Background Technology

[0002] Plasma lighting display devices, such as plasma spheres and electrodeless plasma decorative lamps, utilize high-voltage, high-frequency electric fields to excite gas discharge and produce colored light effects. In recent years, interactive products that link plasma light effects with audio signals have gradually gained market attention. In existing technologies, the linkage between audio and plasma is usually very simple: it only detects the instantaneous amplitude (volume) of the audio signal and linearly maps it to the on / off state of plasma discharge or brightness changes. This approach has the following technical drawbacks.

[0003] The limitations of plasma lighting include: Limited control: It can only control brightness, unable to actively regulate the movement, speed, or spatial distribution of plasma gas, resulting in monotonous light effects; Lack of beat synchronization: It cannot accurately respond to the beats and rhythms in music, leading to a poor subjective experience of "sound and light asynchrony"; Lack of frequency selectivity: Different musical frequencies, such as bass, vocals, and treble, cannot produce differentiated plasma movement patterns, resulting in insufficient interactive immersion; Discharge stability issues: Dynamic changes in audio signals can cause drastic fluctuations in discharge power, easily leading to resonance lock-up, plasma extinction, or flickering. Therefore, a multi-dimensional audio characteristic decoupling control method for plasma lighting is needed to solve these problems. Summary of the Invention

[0004] The purpose of this invention is to solve the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for decoupling and controlling the multidimensional audio features of plasma illumination includes the following steps: S1: Acquires external audio signals with a sampling rate of 44.1kHz or 48kHz and a quantization precision of 16bit or 24bit; S2: Performs multi-level preprocessing on the audio signal, including DC removal, power frequency notch filtering, pre-emphasis, and automatic gain control; S3: Extract three core features in parallel from the preprocessed audio signal: instantaneous amplitude features, beat timing features, and energy distribution features of at least three different frequency bands; S4: Establish a three-dimensional feature mapping model, and independently map the three extracted core features to three control parameters of the plasma: S41: Maps the instantaneous amplitude characteristics to the plasma's basic discharge power; S42: Map the clock timing characteristics into a discharge power pulse signal superimposed on the base discharge power; S43: Map the energy distribution characteristics of different frequency bands to the discharge driving frequency of the plasma; S5: Using the audio signal processing frame as a synchronization reference, it performs adaptive resonant frequency tracking, dynamically adjusts the driving frequency to match the resonant frequency of the plasma load, and maintains synchronization between audio modulation and resonant tracking. S6: Repeat steps S2 to S5 at a period of no more than 20ms.

[0006] Preferably, the method for extracting the instantaneous amplitude feature in step S3 is as follows: the audio signal is divided into frames with a frame length of 8ms to 20ms and a frame overlap rate of 50% to 75%; the root mean square value of each frame is calculated as the original instantaneous amplitude; the original instantaneous amplitude is subjected to exponential smoothing, and the smoothing coefficient α ranges from 0.5 to 0.8.

[0007] Preferably, the method for extracting the beat moment feature in step S3 is as follows: calculate the ratio of the instantaneous amplitude of the current frame to the average amplitude of the past N frames, where N is 80 to 150; When the ratio is greater than the threshold T and the interval between two adjacent candidate beats is in the range of 250ms to 1200ms, it is determined to be a candidate beat; After detecting 3 to 5 valid candidate beats, the phase-locked loop is started to track the beat sequence and predict the time of the next beat.

[0008] Preferably, the threshold T ranges from 1.3 to 1.8.

[0009] Preferably, the method for extracting the frequency band energy distribution characteristics in step S3 is as follows: the audio signal is divided into at least three frequency bands using at least three sets of bandpass filters, and the energy of each frequency band is calculated and normalized to obtain the energy ratio of each frequency band.

[0010] Preferably, the at least three frequency bands include: a low frequency band of 60Hz to 250Hz, a mid frequency band of 250Hz to 2500Hz, and a high frequency band of 2500Hz to 12000Hz.

[0011] Preferably, in step S41, the mapping between the instantaneous amplitude and the base discharge power is a linear mapping: P_base=P_min+(P_max-P_min)×A, where A is the normalized instantaneous amplitude, 0≤A≤1, and P_min and P_max are the minimum stable discharge power and the maximum allowable discharge power of the plasma device, respectively.

[0012] Preferably, in step S42, the parameters of the discharge power pulse signal are: pulse width 5ms to 50ms, pulse amplitude 1.2 times to 2.5 times the basic discharge power, pulse rise time ≤ 2ms, and pulse fall time ≤ 8ms.

[0013] Preferably, in step S43, the mapping relationship between the frequency band energy and the discharge drive frequency is as follows: When the low-frequency energy accounts for more than 50%, the driving frequency is set to the first frequency range; When the mid-frequency band energy accounts for more than 50%, the driving frequency is set to the second frequency range; When the high-frequency band energy accounts for more than 50%, the driving frequency is set to the third frequency range; When the energy share of multiple frequency bands does not exceed 50%, the driving frequency is the weighted average of the corresponding frequencies of each frequency band; The specific values ​​of the first, second, and third frequency ranges are determined based on the resonant circuit parameters of the plasma device, and the absolute frequency values ​​may differ for different devices.

[0014] The present invention has at least the following beneficial effects: 1. By mapping the instantaneous amplitude, beat time, and frequency band energy of audio to the basic discharge power, discharge power pulse, and discharge drive frequency of plasma respectively, decoupled control of three dimensions is achieved, making the plasma gas motion pattern, brightness change, and beat response independently adjustable, resulting in richer dimensions of light effect changes. This breaks through the limitation of the single dimension of existing technologies and significantly improves the interactive experience.

[0015] 2. This invention adopts a short-frame long parallel feature extraction architecture, combined with adaptive resonant tracking and audio control synchronization technology. The end-to-end control delay can be controlled within 30ms, which is far superior to the existing technology which is usually ≥50ms. The delay is difficult for the human eye to perceive, and the subjective sound and light synchronization accuracy is ≥90%.

[0016] 3. This invention uses audio frames as a synchronization reference, performing resonant frequency detection and adjustment once per frame. In case of instability, it prioritizes restoring resonance before restoring audio modulation, effectively avoiding arc extinction or flickering caused by dynamic changes in audio. Actual testing showed no discharge instability after more than 2 hours of continuous operation.

[0017] 4. The pulse width, pulse amplitude, and frequency band sensitivity in the mapping parameters of this invention can be flexibly adjusted, and multiple modes can be adaptively switched. The modes include beat priority mode, melody priority mode, full-dimensional mixing mode, and automatic mode, which can perfectly adapt to all common music styles such as electronic, rock, classical, and pop.

[0018] 5. It does not depend on a specific electrode structure and is applicable to all single-coil electrodeless plasma devices, such as plasma spheres, plasma light strips, plasma speakers, and commercial display lighting fixtures. The parameters in the method, such as frame length, smoothing coefficient, and frequency range, can be calibrated according to the specific hardware platform, and it has good portability. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the overall process of a decoupling control method for audio multidimensional features of plasma illumination proposed in this invention. Figure 2 This is a schematic diagram of the workflow of an embodiment of the standard plasma ball decorative lamp of the present invention; Figure 3 This is a schematic diagram of the workflow of an embodiment of the multi-mode adaptive control plasma light strip of the present invention; Figure 4 This is a schematic diagram illustrating the workflow of an embodiment of the high-power plasma lighting fixture for commercial demonstration of the present invention; Figure 5 This is a schematic diagram illustrating the workflow of an embodiment of the miniaturized portable plasma speaker of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0022] Example 1: Implementation in a standard plasma ball decorative lamp; Reference Figure 1-2Scenario Setting: This embodiment applies to a household plasma ball decorative lamp. The product is placed in the living room or bedroom, and the user plays music via Bluetooth on their mobile phone, hoping that the discharge filament inside the plasma ball will move in rhythm with the music and change shape according to the melody. The plasma ball includes: a glass shell filled with low-pressure xenon gas or a mixture of inert gases at a pressure of approximately 10 kPa; a 20 mm diameter central spherical high-voltage electrode; a transparent conductive coating on the inner wall of the shell serving as the ground electrode; a high-frequency high-voltage power supply module containing a DDS frequency synthesizer, a MOSFET power amplifier, an LC resonant boost circuit, and an audio acquisition and processing circuit board containing a MEMS microphone, an STM32F405 microcontroller, a 24-bit ADC, and a miniature current transformer for detecting the resonant current. The entire device is powered by a 12V / 3A DC adapter.

[0023] The MEMS microphone is soldered onto the processing circuit board and connected to the SAI interface of the STM32 via the I2S bus. The MEMS microphone model is INMP441. The microphone's pickup hole faces outward from the spherical shell, about 2cm away from the bottom of the shell, to avoid picking up power supply noise.

[0024] The STM32's SPI1 interface connects to the frequency control word register of the DDS chip. The DDS outputs a sine wave, which is sent to the gate driver chip of the MOSFET power amplifier. The DDS chip is model AD9833, the peak-to-peak value of the DDS output sine wave is 0.6V, and the frequency range is 1.5MHz~7MHz. The MOSFET power amplifier is model IRF510, and the gate driver chip is model TC4420.

[0025] The drain-series LC resonant circuit of the MOSFET power amplifier consists of: L, a primary coil (4 turns, 0.8mm wire diameter) wound on a nickel-zinc ferrite core; and C, a high-voltage mica capacitor (470pF / 2kV). One end of the secondary coil (200 turns, 0.2mm wire diameter) is connected to the center electrode, and the other end is grounded through a current transformer. The secondary output of the current transformer (turns ratio 1:50) is rectified and filtered by an operational amplifier before being connected to the ADC (12-bit) of the STM32 microcontroller.

[0026] Power regulation is achieved by changing the supply voltage of the MOSFET power amplifier: the PWM output (20kHz) of the STM32 controls a synchronous buck DC-DC converter (TPS5430), with an output voltage range of 5V~30V, corresponding to a discharge power of approximately 5W~60W. The supply voltage and discharge power are calibrated to have a linear relationship: P(W)=2×V+0.5.

[0027] In the above structure, the microphone picks up ambient music, the STM32 performs audio processing and feature mapping, the DDS generates the drive frequency, the DC-DC converter controls the base power, the MOSFET power amplifier amplifies the DDS signal and applies it to the electrodes via LC resonance boost, and the current transformer provides feedback on the resonant state. The entire control loop forms a closed loop: audio → digital processing → power / frequency adjustment → discharge → feedback → drive frequency adjustment.

[0028] Work process: First: The user turns on the plasma ball power, and the system initializes. After the STM32 starts, it first performs a self-test: driven at the default frequency of 3.5MHz and the default supply voltage of 12V, the effective value of the secondary current is detected through the current transformer. Then, it scans from 3.0MHz to 4.5MHz in 100kHz steps, recording the current value at each frequency point, and finds the resonant frequency f0 corresponding to the current peak (measured at approximately 3.6MHz). The DDS frequency is locked to f0, with the deviation controlled within ±1%. This process lasts approximately 200ms, completing the initial resonance tracking. After the self-test is completed, the plasma ball enters standby mode, the base power is reduced to 10W, and it maintains a dim light.

[0029] Then: The user starts playing music, and the system enters its main loop. The STM32 performs the following operations in a 10ms cycle: Read 480 sampling points from the microphone (48kHz sampling rate, 10ms frame length) and store them in the DMA buffer.

[0030] Preprocessing includes: DC removal (subtracting the average value), 50Hz notch filtering (second-order IIR, Q=5), pre-emphasis (first-order high-pass filter, cutoff frequency 50Hz, coefficient 0.97), and automatic gain control (dynamic range 50dB, response time 100ms, release time 300ms).

[0031] Parallel computing has three characteristics: Instantaneous amplitude: Calculate the intra-frame RMS, and after exponential smoothing with α=0.7, we get A_n=0.7×RMS_n+0.3×A_{n-1}, which is normalized to [0,1].

[0032] Beat Time: Calculate the ratio R of the current frame's RMS to the average RMS of the past 100 frames. If R > 1.5 and the interval between the current time and the previous candidate beat is between 300ms and 1000ms, it is marked as a candidate beat. After detecting 5 consecutive candidate beats, start the PLL: calculate the average BPM based on the intervals of the last 8 beats, and predict the next beat time. If the prediction error is < 20ms, the predicted beat is accepted.

[0033] Bandwidth energy: The energy of each band is calculated using three sets of fourth-order Butterworth IIR bandpass filters (80~200Hz, 200~2000Hz, 2000~8000Hz), and the normalized ratios L_ratio, M_ratio, and H_ratio are obtained.

[0034] 3D mapping: Base power: P_base = 10W + 50W × A (linear mapping).

[0035] Beat pulse: If the current time is a beat moment (detection or prediction), then generate a pulse: P_instant=P_base×1.5, pulse width 20ms (that is, maintain high power during the duration of this frame, and resume in the next frame).

[0036] Drive frequency: If L_ratio>55%→f_drive=2.2MHz; if M_ratio>55%→f_drive=3.5MHz; if H_ratio>55%→f_drive=5.0MHz; otherwise f_drive=L_ratio×2.2+M_ratio×3.5+H_ratio×5.0(MHz).

[0037] Execution control: The DDS frequency register is updated via SPI; the DC-DC output voltage is adjusted via PWM to correspond to P_base (or P_instant if there is a clock pulse in the current frame). The PWM duty cycle is updated at the beginning of the frame, ensuring that the pulse rising edge is ≤1ms.

[0038] Resonance Tracking: In the last 1ms of each frame, the DC voltage V_fb after rectification by the current transformer is read via ADC. The difference ΔV between V_fb and the previous frame is calculated. If ΔV < -15% and persists for 2 frames, it is considered a resonance offset, and the DDS frequency is adjusted in 20kHz steps towards f0. If V_fb recovers after adjustment, adjustment continues until the deviation is <2%; if V_fb continues to decrease after adjustment, adjustment is reversed. During adjustment, if a drop in V_fb exceeding 25% is detected, the beat pulse superposition is paused, and only the base power is maintained until V_fb recovers to the normal range (≥80% of the original value) before resuming.

[0039] Then: Repeat the above cycle. When the user plays different music, the color, speed, and rhythm of the discharge filament inside the plasma sphere change in real time with the music. For example, when playing heavy bass electronic music, the low-frequency energy of 80~200Hz often accounts for more than 60%, the driving frequency is stable at 2.2MHz, the discharge filament is a relatively thick wire (about 2-3mm in diameter), rotates slowly (about 0.5 revolutions / second), and at the same time, each pulse makes the discharge filament brighten instantly and "spray" outward, creating a strong visual impact.

[0040] When playing pop vocals, the mid-frequency range of 200-2000Hz dominates. The frequency then shifts to 3.5MHz, where the discharge filament becomes finer (approximately 1mm in diameter), moves faster (approximately 1.5 revolutions per second), and the pulse intensity decreases (by 1.5 times), highlighting the smooth variations in the melody. When playing classical string music, the high-frequency energy is higher, the frequency rises to 5.0MHz, the discharge filament becomes extremely fine (<0.5mm), vibrates rapidly (approximately 3 revolutions per second), and there are no pulses or extremely weak pulses, resulting in an elegant and delicate light effect.

[0041] The current transformer is installed on the secondary coil grounding circuit, and its output is a weak sine wave (mV level). After rectification and filtering by the operational amplifier, a DC voltage of 0~3.3V is obtained, which is proportional to the effective value of the discharge current. The STM32 reads this voltage every frame and compares it with the previous frame. The instability judgment criterion is: a voltage drop of more than 25% lasting for 2 frames. When instability occurs, the STM32 expands the DDS frequency step search range to 100kHz to quickly find a new resonant point, which is usually recovered within 3 frames. At the same time, in order to prevent power supply overshoot caused by sudden changes in DC-DC output voltage, the STM32 adopts incremental change limit when updating the PWM duty cycle: a single change does not exceed 5% of the duty cycle. If a larger change is required, such as a clock pulse, it is executed in two steps: first rising to the intermediate value, and then rising to the target value after 1ms.

[0042] In this embodiment, actual testing on the prototype showed an end-to-end control delay of approximately 18-25ms. From microphone pickup to discharge power changes, subjective evaluation indicated excellent audio-visual synchronization with no visible delay. No arc extinction occurred during 2 hours of continuous operation. Differences in discharge filament morphology were visually discernible under different music styles. User feedback indicated a significantly superior user experience compared to commercially available single-dimensional products.

[0043] Example 2: Plasma LED strip with multi-mode adaptive control; Reference Figure 3 Scenario Setting: This embodiment applies to a flexible plasma light strip, 1 meter in length, containing 8 independently controlled plasma discharge units for smart home ambient lighting. Each unit includes a small single-coil transformer and a pair of needle electrodes. The light strip is powered by a 24V / 5A power adapter. Users can select different music response modes via a mobile app: beat priority, melody priority, and automatic. The light strip can dynamically display light effects in space in sync with the music, such as the beat flowing from left to right. The main control chip uses the ESP32-S3, which supports Wi-Fi / Bluetooth. Audio acquisition is achieved through an onboard MEMS microphone, and a 3.5mm external audio input interface is also provided.

[0044] Structural connections and fit relationships: The difference from Embodiment 1 is that the main controller is an ESP32, whose I2S interface is connected to a microphone and its SPI interface is connected to a DDS (AD9833) to generate a common drive frequency. All eight discharge units share the same DDS frequency source because frequency affects the gas motion pattern, and a unified frequency ensures visual consistency. However, each unit has an independent power control MOSFET (AO3400) and a DC-DC buck circuit, both of which are controlled separately by the eight PWM channels of the ESP32.

[0045] The power control loop for each discharge unit: The PWM output (10kHz) of the ESP32 controls a simple step-down circuit with an output voltage range of 0~24V, corresponding to a discharge power of 0~15W for each unit. Since each unit is independent, spatial light effect choreography can be achieved.

[0046] An audio feature preprocessing module has been added: within the ESP32, in addition to extracting 3D features, low-frequency energy (20~100Hz) and instantaneous BPM are calculated for automatic mode switching. Simultaneously, the ESP32 receives mode setting commands from a mobile app via Bluetooth.

[0047] To reduce overall power consumption, the system automatically enters standby mode when no audio input is detected for more than 30 seconds: all unit power is reduced to 2W and the drive frequency is fixed at 3.0MHz.

[0048] Work process: First: The user selects beat priority mode via a mobile app. Upon receiving the command, the ESP32 adjusts the mapping parameter table: The pulse width has been increased from the default 20ms to 40ms, and the pulse amplitude has been increased from 1.5 times to 2.0 times.

[0049] The basic power mapping range is compressed to 5W~12W (originally 5W~15W), making the relative intensity of the pulse more prominent.

[0050] The sensitivity of the band energy to frequency mapping is reduced, meaning that the frequency only switches to 2.5MHz when the low-frequency energy is >70%, otherwise it remains at 3.5MHz, so as to avoid frequent frequency switching from interfering with the beat perception.

[0051] Simultaneously, a spatial flow effect is enabled: at each beat, the ESP32 sequentially (at 5ms intervals) triggers pulses from unit 1 to unit 8, creating a light effect that flows from left to right.

[0052] Then: The user plays electronic music. The system analyzes in real time: the low-frequency proportion consistently exceeds 70%, and the instantaneous BPM remains stable at 125-130. In beat-priority mode, at each beat, all eight units simultaneously generate a 40ms wide pulse with twice the power, causing an instantaneous jump in the overall brightness of the light strip; simultaneously, due to the longer pulse width (40ms), the plasma gas within each unit exhibits a noticeable jet-like motion, with the discharge filament expanding outward and then contracting, creating a visual effect that perfectly matches the drumbeat. In addition, the flowing effect causes the light to sweep from left to right, enhancing the sense of rhythm.

[0053] Finally: If the user switches to classical music, they can select "Melody Priority Mode" through the app. The parameters become: pulse width 10ms, amplitude 1.2 times, and frequency band mapping sensitivity doubled. This means that a change in mid-frequency energy from 45% to 55% can cause a linear change in frequency from 3.2MHz to 3.8MHz. At this time, the discharge filament shape changes continuously with the rise and fall of the violin melody. Because the beat of classical music is not significant, the beat pulse is almost imperceptible, highlighting an elegant gradual effect. Simultaneously, spatial flow effects are disabled, and all units change synchronously to avoid distraction.

[0054] Automatic mode details: In automatic mode, the ESP32 calculates the low-frequency energy percentage (20-100Hz) and instantaneous BPM (using an autocorrelation algorithm) every 500ms. If the low-frequency percentage is >60% and the BPM is >110 for 2 seconds, it switches to beat priority mode. If the mid-to-high frequency ratio (>800Hz) is greater than 65% and the BPM is less than 90 for 2 seconds, switch to melody priority mode; otherwise, use full-dimensional mixing mode with a pulse width of 20ms, an amplitude of 1.5 times, and normal frequency band sensitivity. The switching process is smooth: the pulse width and amplitude are changed linearly within 200ms to avoid sudden parameter changes that could lead to unstable discharge or abrupt auditory sensations.

[0055] The independent control of each discharge unit requires precise synchronization. The ESP32 uses a hardware timer with a 5ms baseline interrupt. In each interrupt, the PWM duty cycle of each unit is updated. The beat timing is calculated by the audio processing task and sent to the timer interrupt service routine via a queue. To ensure the timing accuracy of the flow of the eight units (5ms interval), a flow counter is maintained in the timer interrupt, and the flow sequence is started once for each beat. In addition, to prevent electromagnetic interference between units, the transformer cores of each unit are spaced at least 20mm apart, and ferrite beads are installed on the power lines.

[0056] This embodiment achieves adaptive matching of music style and spatial lighting effect arrangement. User tests show that in beat-priority mode, the drumbeats of electronic music have a strong impact, and the flowing lighting effects enhance the fun; in melody-priority mode, the melody changes of classical music correspond continuously to the movement of the discharge wires. The automatic mode switching accuracy is approximately 90% (based on tests with 30 different music styles), requiring no manual user intervention.

[0057] Example 3: High-power plasma lighting fixtures for commercial displays; Reference Figure 4 Scenario Setting: This embodiment is applied to large-scale plasma lighting installations in commercial venues, such as bars and stages, with a total discharge power of up to 500W, used for dynamic light effect displays. This device requires a balanced audio signal (line level, +4dBu) from an external professional mixing console, rather than a built-in microphone. Due to the high power, the Q value of the resonant circuit is high (approximately 50), and the resonant frequency is very sensitive to load changes (temperature, gas pressure), requiring higher adaptive tracking capabilities. The device includes: a large spherical plasma lamp filled with xenon gas; a high-voltage power supply using a full-bridge LLC resonant converter, with an input of 220V AC and an adjustable high-frequency high-voltage output; and a main control system using an STM32F407+FPGA (for high-speed resonance detection).

[0058] The audio input uses a balanced XLR interface, which is converted into a single-ended signal by a differential amplifier (INA137) and then sent to the STM32's ADC (24-bit, 48kHz sampling).

[0059] The power amplification stage employs a full-bridge LLC resonant converter with the following topology: four MOSFETs (IXFH26N50) form a full bridge, driving the primary winding of the transformer. The secondary winding is rectified to obtain a DC bus voltage, which supplies the Class E power amplifier. The control strategy adopted in this invention is as follows: the audio amplitude is mapped to the target output power P_target, and then the driving frequency f_base is obtained by looking up a table in reverse according to the gain-frequency curve of the LLC converter; the offset Δf of the frequency band energy mapping is then superimposed on f_base to obtain the final driving frequency f_drive = f_base + Δf. This achieves partial decoupling between power and frequency.

[0060] A temperature sensor (PT100) is added to the core of the LLC transformer. When the temperature exceeds 85°C, the maximum power limit is automatically reduced (linearly from 500W to 300W) as a protection device.

[0061] Resonance detection: The phase difference between the voltage across the LLC resonant capacitor after attenuation by the high-voltage differential probe and the primary current of the current transformer is calculated by the FPGA to obtain the resonance deviation. The FPGA reports the deviation value to the STM32 once every frame (10ms).

[0062] Work process: First: After the system powers on, the STM32 controller controls the FPGA to perform a resonant frequency scan: from 1.5MHz to 7MHz in 50kHz steps, pausing for 5ms at each frequency point. The FPGA records the voltage and current phase difference, finding the frequency f0 (resonant point) where the phase difference is zero. The rated operating point of the LLC converter is set at f0 + 100kHz, as LLC converters typically operate slightly above the resonant frequency to achieve ZVS soft switching and a better adjustment range. Gain-frequency curve calibration: Within a range of ±300kHz around f0, the relationship between output power and frequency is measured and stored in a lookup table.

[0063] Secondly: The main loop period is set to 10ms. The audio processing flow is the same as in Example 1, but the mapping relationship is adjusted: The basic power mapping range is P_min=50W to maintain the minimum stable discharge, and P_max=450W to leave a 50W margin.

[0064] The pulse amplitude is limited to 1.3 times the base power to prevent damage from overpower. The maximum instantaneous power does not exceed 500W.

[0065] Frequency band energy is mapped to a frequency offset Δf: when the low-frequency band (60-250Hz) dominates (>60%), Δf = -150kHz; when the mid-frequency band (250-2500Hz) dominates, Δf = 0; when the high-frequency band (2500-12000Hz) dominates, Δf = +200kHz. When multiple frequency bands are mixed, Δf = (L_ratio×(-150) + M_ratio×0 + H_ratio×200)kHz.

[0066] Final drive frequency: f_drive = f_base(P_target) + Δf, where f_base is obtained by looking up a table.

[0067] Finally: When playing rock music, the low-frequency energy is high, Δf is negative (-150kHz), and the drive frequency is reduced. Due to the gain-frequency characteristic of the LLC converter: as the frequency decreases, the gain increases, and the output power naturally increases. At the same time, the low-frequency drive makes the discharge wire thicker and moves more slowly (about 0.2 revolutions / second), resulting in a strong bass impact. The beat pulse (1.3 times the power, 15ms width) further brightens the lamp instantaneously during strong beats, causing the discharge wire inside the entire spherical lamp body to expand outward. Adaptive resonance tracking uses the FPGA to provide the phase difference for each frame, and the STM32 adjusts the frequency to maintain the phase difference within ±5°. If the phase difference exceeds ±15°, it is considered unstable, and pulse superposition is paused, adjusting towards f0 at a speed of 5kHz / step until the phase difference is restored.

[0068] Due to the high power output, thermal management is crucial. During continuous high-power output (>400W), temperature sensor feedback causes the STM32 to gradually reduce P_max: when the temperature >80℃, the maximum power is reduced by 5% every 10 seconds; when the temperature >85℃, it is reduced by 10% every 5 seconds; when the temperature >90℃, it is forced into protection mode, reducing the power to 100W and stopping audio modulation. Simultaneously, if the FPGA reports a phase difference >20° for 10 consecutive frames, the system automatically enters protection mode: fixing the drive frequency at f0+100kHz, reducing the power to 100W, maintaining basic discharge, and attempting recovery after 30 seconds. Furthermore, to meet the requirements of long-term operation in commercial settings, all power devices are mounted on forced-air-cooled heatsinks, with fan speeds controlled by PWM.

[0069] This embodiment achieves audio synchronization with high-power commercial plasma lighting fixtures, providing a strong visual impact while ensuring safety. Testing showed that at a peak power of 450W, the audio-visual synchronization error was less than 25ms (oscilloscope measurement: the rise time difference between the audio pulse and the light intensity pulse). After 4 hours of continuous operation, the maximum temperature remained stable at 82℃ without any faults. On-site user experience feedback: During bass frequencies, the entire space felt like it was vibrating; during treble frequencies, the discharge filament flickered like snowflakes.

[0070] Example 4: Miniaturized Portable Plasma Speaker; Reference Figure 5 Scenario Setting: This embodiment applies to a portable Bluetooth speaker with a small plasma ball (8cm in diameter) integrated on its top. It is battery-powered (5V / 2A USB, built-in 18650 battery pack, 5000mAh capacity). Due to power limitations (maximum 8W), specially optimized algorithms are needed to reduce power consumption and extend battery life. Users expect the product to provide ambient lighting effects during outdoor gatherings and have a battery life of at least 4 hours. The main control chip uses a low-power Cortex-M0+ (STM32G031) with a main frequency of 48MHz. The audio processing algorithm is simplified: frame length 20ms, only instantaneous amplitude and beat are calculated, and frequency band energy is an optional function that users can choose to turn on or off via the app; turning it off saves 30% of power. The power amplification uses a Class E amplifier (efficiency up to 85%), with the DDS frequency fixed at 2.8MHz (when frequency band mapping is off). Power is adjusted by changing the amplifier's supply voltage (0~5V). Add a photosensor to automatically adjust the maximum power according to the ambient brightness. It can operate at full power (8W) during the day and reduce to 2.5W at night to avoid glare and save power.

[0071] The MEMS microphone connects to the STM32 via a PDM interface, saving I2S pins. The STM32 internally converts PDM to PCM, with a sampling rate of 44.1kHz and a frame length of 20ms (882 sampling points).

[0072] The DDS uses the chip's internal programmable clock generator (STM32G031's TIM1 + lookup table DAC) to output a 2.8MHz square wave, which, after low-pass filtering, drives a Class E power amplifier. The Class E power amplifier includes: MOSFET (AO3400), choke inductor, resonant capacitor, and load coil.

[0073] Power regulation: The STM32's DAC output of 0~1.2V controls the output voltage (0~5V) of a low dropout linear regulator (LDO, RT9013), which serves as the power supply voltage for the Class E power amplifier. Since the output power of the Class E power amplifier is proportional to the square of the power supply voltage, the mapping relationship is P(W) = 0.32 × V^2 (calibrated value).

[0074] The photosensor is connected to the STM32 via I2C and reads the ambient illuminance every 10 seconds.

[0075] Battery power monitoring uses an ADC to read the voltage divider resistor, and automatically switches to power-saving mode when the battery level is below 20%.

[0076] Work process: First: The user plays music from their mobile phone via Bluetooth. The system defaults to full-feature mode (amplitude + beat + frequency band). With the battery at 100% and ambient light at 300 lux (daytime), the system operates at maximum power of 8W. Frequency band mapping is enabled: the audio is divided into two frequency bands (low frequency (60-200Hz) and high frequency (2000-8000Hz)) using a fourth-order IIR filter (simplified calculation). The drive frequency is set to 2.5MHz when low frequency dominates and 3.2MHz when high frequency dominates, achieved by switching DDS to look up a table. The beat detection algorithm is simplified: only the RMS ratio is calculated, without PLL, with a threshold of 1.6 and an interval of 300-800ms.

[0077] Secondly, during music playback, when the battery level drops to 18%, the system automatically switches to power-saving mode: frequency band mapping is turned off (fixed frequency 2.8MHz), the frame length remains at 20ms, but automatic gain control is disabled to save computation, and beat detection is changed to once every two frames, reducing the processing load. Actual power consumption is reduced by approximately 40%, and battery life is extended from 5 hours to 7 hours. In power-saving mode, the amplitude mapping power range is changed to 2W~5W, the beat pulse width is 10ms, and the amplitude is 1.2 times. The discharge wire movement is smooth, and the beat is weaker but still perceptible.

[0078] Finally: If the user manually disables the beat pulse (via the app), the system further reduces power consumption: only amplitude mapping is performed, and the pulse is disabled. In this case, the plasma brightness only changes with the volume, and the gas movement remains stable, making it suitable as an ambient light at night. Combined with a photosensor: when the ambient illuminance is <100 lux, the maximum power is automatically limited to 2.5W, further saving power.

[0079] To extend battery life, the system employs dynamic voltage regulation: when power demand is below 2W, the Class E amplifier supply voltage is reduced to 1.5V (the minimum output voltage of an LDO), while efficiency remains above 80%. Simultaneously, the STM32 enters sleep mode (waking up only when an audio frame is interrupted), reducing standby power consumption to below 1mA. A photosensor reads the sensor every 10 seconds, automatically adjusting the maximum power factor: factor = 1.0 for ambient illuminance > 300 lux, factor = 0.3 for < 100 lux, with linear interpolation in between to avoid excessive brightness and glare at night.

[0080] This embodiment demonstrates the feasibility of the method of the present invention on low-power, miniaturized devices. Prototype testing: Under full charge, the battery life is 5.2 hours in full-feature mode and 7.5 hours in power-saving mode. The audio-visual synchronization error is approximately 30ms (due to a 20ms frame length), which is subjectively acceptable. Users can flexibly select feature dimensions according to their scenarios to balance performance and battery life.

[0081] The audio multi-dimensional feature decoupling control method for plasma lighting displays provided by this invention is implemented through software algorithms, does not depend on specific hardware structures, and can be widely applied in plasma decorative lights, plasma speakers, commercial display lighting, smart home ambient products, and other fields. The parameters in the method, such as frame length, smoothing coefficient, and frequency range, can be calibrated according to specific hardware platforms, exhibiting good portability.

[0082] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for decoupling and controlling the multi-dimensional audio features of plasma illumination, characterized in that, Includes the following steps: S1: Acquires external audio signals with a sampling rate of 44.1kHz or 48kHz and a quantization precision of 16bit or 24bit; S2: Performs multi-level preprocessing on the audio signal, including DC removal, power frequency notch filtering, pre-emphasis, and automatic gain control; S3: Extract three core features in parallel from the preprocessed audio signal: instantaneous amplitude features, beat timing features, and energy distribution features of at least three different frequency bands; S4: Establish a three-dimensional feature mapping model, and independently map the three extracted core features to three control parameters of the plasma: S41: Maps the instantaneous amplitude characteristics to the plasma's basic discharge power; S42: Map the clock timing characteristics into a discharge power pulse signal superimposed on the base discharge power; S43: Map the energy distribution characteristics of different frequency bands to the discharge driving frequency of the plasma; S5: Using the audio signal processing frame as a synchronization reference, it performs adaptive resonant frequency tracking, dynamically adjusts the driving frequency to match the resonant frequency of the plasma load, and maintains synchronization between audio modulation and resonant tracking. S6: Repeat steps S2 to S5 at a period of no more than 20ms.

2. The method for decoupling and controlling the multi-dimensional audio features of plasma illumination according to claim 1, characterized in that, The method for extracting instantaneous amplitude features in step S3 is as follows: the audio signal is divided into frames with a frame length of 8ms to 20ms and a frame overlap rate of 50% to 75%; the root mean square value of each frame is calculated as the original instantaneous amplitude; the original instantaneous amplitude is subjected to exponential smoothing, and the smoothing coefficient α ranges from 0.5 to 0.

8.

3. The method for decoupling and controlling the multi-dimensional audio characteristics of plasma illumination according to claim 1, characterized in that, The method for extracting the beat moment features in step S3 is as follows: calculate the ratio of the instantaneous amplitude of the current frame to the average amplitude of the past N frames, where N is 80 to 150; When the ratio is greater than the threshold T and the interval between two adjacent candidate beats is in the range of 250ms to 1200ms, it is determined to be a candidate beat; After detecting 3 to 5 valid candidate beats, the phase-locked loop is started to track the beat sequence and predict the time of the next beat.

4. The method for decoupling and controlling the multi-dimensional audio features of plasma illumination according to claim 3, characterized in that, The threshold T ranges from 1.3 to 1.

8.

5. The method for decoupling and controlling the multi-dimensional audio features of plasma illumination according to claim 1, characterized in that, The method for extracting the frequency band energy distribution characteristics in step S3 is as follows: use at least three sets of bandpass filters to divide the audio signal into at least three frequency bands, calculate the energy of each frequency band, and normalize to obtain the energy ratio of each frequency band.

6. The method for decoupling and controlling the multi-dimensional audio features of plasma illumination according to claim 5, characterized in that, The at least three frequency bands include: a low frequency band of 60Hz to 250Hz, a mid frequency band of 250Hz to 2500Hz, and a high frequency band of 2500Hz to 12000Hz.

7. The method for decoupling and controlling the multi-dimensional audio features of plasma illumination according to claim 1, characterized in that, In step S41, the mapping between instantaneous amplitude and base discharge power is a linear mapping: P_base=P_min+(P_max-P_min)×A, where A is the normalized instantaneous amplitude, 0≤A≤1, and P_min and P_max are the minimum stable discharge power and the maximum allowable discharge power of the plasma device, respectively.

8. The method for decoupling and controlling the multi-dimensional audio features of plasma illumination according to claim 1, characterized in that, In step S42, the parameters of the discharge power pulse signal are: pulse width 5ms to 50ms, pulse amplitude 1.2 times to 2.5 times the basic discharge power, pulse rise time ≤ 2ms, and pulse fall time ≤ 8ms.

9. The method for decoupling and controlling the multi-dimensional audio characteristics of plasma illumination according to claim 1, characterized in that, In step S43, the mapping relationship between bandwidth energy and discharge drive frequency is as follows: When the low-frequency energy accounts for more than 50%, the driving frequency is set to the first frequency range; When the mid-frequency band energy accounts for more than 50%, the driving frequency is set to the second frequency range; When the high-frequency band energy accounts for more than 50%, the driving frequency is set to the third frequency range; When the energy share of multiple frequency bands does not exceed 50%, the driving frequency is the weighted average of the corresponding frequencies of each frequency band; The specific values ​​of the first, second, and third frequency ranges are determined based on the resonant circuit parameters of the plasma device, and the absolute frequency values ​​may differ for different devices.

Images