Hi, i want to realize a fully digital closed loop pwm modulator for my class d amplifier but i've no idea on how to start with my design.
Googling i've found sigma delta modulator and self-oscillating class d amplifier implemented in analog way (input+feedback -> integrator -> comparator -> flip-flop).
At the moment the mosfets of my class d amplifier (or pwm inverter) are driven by an open loop classic pwm modulator (sawtooth wave compared with digital sinusoid witch samples are generated by internal NCO).
There is an A2D converter tha acquires a partition of output voltage and current.
PWM pulses are generated by an FPGA and output signal is processed by the same FPGA.
My idea is to change pwm method in order to sense the output and adjust pwm in proper way as the output impedance change.
Is there any tutorial/book/application note from which start my studies?
Could you help me?
Thanks
Googling i've found sigma delta modulator and self-oscillating class d amplifier implemented in analog way (input+feedback -> integrator -> comparator -> flip-flop).
At the moment the mosfets of my class d amplifier (or pwm inverter) are driven by an open loop classic pwm modulator (sawtooth wave compared with digital sinusoid witch samples are generated by internal NCO).
There is an A2D converter tha acquires a partition of output voltage and current.
PWM pulses are generated by an FPGA and output signal is processed by the same FPGA.
My idea is to change pwm method in order to sense the output and adjust pwm in proper way as the output impedance change.
Is there any tutorial/book/application note from which start my studies?
Could you help me?
Thanks
Most likely you will find some inspiration when searching for fully digital control of SMPS rather than in literature of class D amplifiers.
Class D amps are pretty demanding in terms of sonic aspects, which calls in the digital world of real time feedback systems and PWM generation for expensive high end stuff in order to achieve just unspectacular performance.
But for most SMPS requirements also low cost digital solutions are sufficient. Consequently the fully digital approach becomes more and more popular in SMPS, but not yet popular in class D amps for audio.
After understanding some of the fully digital SMPS controls, you can spec it up to 4-quadrant operation and to your sonic requirements and might end up with a combination of a powerful DSP+FPGA.
P.S:
While your current solution is still an open loop system you could reuse and extend it for learning and performing general experiments about feedback systems of class D amps.
The basics and pitfalls you understood there will help you during digital implementation.
P.P.S:
Depending on your backround and daily available time you can expect a project time between 6 months to 3 years until your digital approach meets an average performance like a 30 dollar module from ebay.
Class D amps are pretty demanding in terms of sonic aspects, which calls in the digital world of real time feedback systems and PWM generation for expensive high end stuff in order to achieve just unspectacular performance.
But for most SMPS requirements also low cost digital solutions are sufficient. Consequently the fully digital approach becomes more and more popular in SMPS, but not yet popular in class D amps for audio.
After understanding some of the fully digital SMPS controls, you can spec it up to 4-quadrant operation and to your sonic requirements and might end up with a combination of a powerful DSP+FPGA.
P.S:
While your current solution is still an open loop system you could reuse and extend it for learning and performing general experiments about feedback systems of class D amps.
The basics and pitfalls you understood there will help you during digital implementation.
P.P.S:
Depending on your backround and daily available time you can expect a project time between 6 months to 3 years until your digital approach meets an average performance like a 30 dollar module from ebay.
My first attempt at a class d discrete amp sounded OK.
It had no feedback so power supply noise/hum was a problem.
I then moved on to the IRS2092 which is a different ball game, fussy about pcb layout, fussy about decoupling etc etc
It had no feedback so power supply noise/hum was a problem.
I then moved on to the IRS2092 which is a different ball game, fussy about pcb layout, fussy about decoupling etc etc
The requirements concerning control loop dynamic are quite relaxed with smps compared to a class-d audio-amp.
Digital control is becoming more popular with power supplies nowadays.
Concerning class-D I cannot see any good reason to go for full digital loop here, but "digital" seems to be buzzword with magic inside sometimes.😉
Digital control is becoming more popular with power supplies nowadays.
Concerning class-D I cannot see any good reason to go for full digital loop here, but "digital" seems to be buzzword with magic inside sometimes.😉
Last edited:
- Status
- Not open for further replies.