I recently had my second class-d design running, and I am using a current feedback 'trick' that you might find interessting.
The idea is that an amount of current feedback is desirable at and above the resonant frequency of the output filter. This way, the filter will not peak (high-q) at light load, introducing a fast phase transistion making compensation of the feedback loop difficult.
The trick is to synthezise the current in the output inductor using the voltage _across_ the inductor and integrating this voltage.
The voltage can be obtained by measuring the voltage on each terminal of the inductor, or you can add a second winding to the inductor. I chose the last approach due to the isolation, and the ease of implementation.
I don't need an integrator, I just run the signal through a RC-filter.
This signal is then used as feedback in an inner loop. The outer loop is then much more stable when the output load is changng.
The current feedback in my design is active from about 1kHz and upwards. This is more than enough to give me the performance I seek.
This teqnique is especially useful in push-pull outputs, where the output current can not he measured referenced to ground.
Regards
The idea is that an amount of current feedback is desirable at and above the resonant frequency of the output filter. This way, the filter will not peak (high-q) at light load, introducing a fast phase transistion making compensation of the feedback loop difficult.
The trick is to synthezise the current in the output inductor using the voltage _across_ the inductor and integrating this voltage.
The voltage can be obtained by measuring the voltage on each terminal of the inductor, or you can add a second winding to the inductor. I chose the last approach due to the isolation, and the ease of implementation.
I don't need an integrator, I just run the signal through a RC-filter.
This signal is then used as feedback in an inner loop. The outer loop is then much more stable when the output load is changng.
The current feedback in my design is active from about 1kHz and upwards. This is more than enough to give me the performance I seek.
This teqnique is especially useful in push-pull outputs, where the output current can not he measured referenced to ground.
Regards
Hi,
this technique of current sensing is used in some of newer multiphase PWM modulators for microprocessor power supplies. (I think Intersil).
It has a minor drawback. LR time constant of the inductor must be the same as the integrator RC time constant, otherwise your inductor current replica will be phase shifted. As inductor has winding made of copper, it has relatively large temperature dependance. (If i remember correctly, copper resistance doubles at 90 degrees temperature rise).
I think that puting a sense resistor in source of each lower mosfet transistor in each of the legs of the full bridge switching stage and then using differential amplifier to sense current across resistors is preferable solution in full bridge amplifiers.
But for split supply half bridge designs it is a simple and elegant solution.
Best regards,
Jaka Racman
this technique of current sensing is used in some of newer multiphase PWM modulators for microprocessor power supplies. (I think Intersil).
It has a minor drawback. LR time constant of the inductor must be the same as the integrator RC time constant, otherwise your inductor current replica will be phase shifted. As inductor has winding made of copper, it has relatively large temperature dependance. (If i remember correctly, copper resistance doubles at 90 degrees temperature rise).
I think that puting a sense resistor in source of each lower mosfet transistor in each of the legs of the full bridge switching stage and then using differential amplifier to sense current across resistors is preferable solution in full bridge amplifiers.
But for split supply half bridge designs it is a simple and elegant solution.
Best regards,
Jaka Racman
I am using a classical discrete triangle/pwm circuit. This is primarily because the design calls for syncronization of the switching frequency using a PLL. The oscillator also has supply voltage feed-forward to keep the modulator gain constant at changing supply voltage. haven't tested that yet.We'll se how that goes.....
I am sorry but I can't give you the complete schematics. There's not much new in there anyways. This amplifier is very application specific, and very much pro-only, so the only ones that would benefit from it, is probably our competitors.
It will require extensive modification to be used in a home-environement.
I will do a version based on this one for domestic use. That might become public domain.We'll have to see how it turns out.
Regards
I am sorry but I can't give you the complete schematics. There's not much new in there anyways. This amplifier is very application specific, and very much pro-only, so the only ones that would benefit from it, is probably our competitors.
It will require extensive modification to be used in a home-environement.
I will do a version based on this one for domestic use. That might become public domain.We'll have to see how it turns out.
Regards
Your approach could be useful in current-driving of loudspeakers when ever someone wants to do that with class D amplifiers.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.