I have pretty much finished this amplifier that seems to satisfy my desire for a decently clean and simple class D design that I might be able to get motivated to build, given the chance.
I started with the idea of taking two N-ch MOSFETs and trying to design the simplest yet low distortion discreet circuit that I could come up with.
The output stage seems pretty well optimized, but I lack desire now to continue further optimizing the rest of the circuit. I would most likely tinker with a p2p version and if it could work much at all that way, I would be pretty confident about it.
The circuit values might be pretty close to optimum in the current mirror and the input and feedback circuits. You'd probably want to select the phase lead components as aggressively as possible in such a way to cause high frequency roll-off a little above the upper audio frequency limit. That would provide maximum load damping.
I show it for +/-40v operation. With some modifications, it ought to work on as much as +/- 100v with at least some performance loss. That's all for now. I will try to answer questions if I can.
I have left in some test options in the input and feedback circuit. I find that inverting input mode is cleaner while noninverting mode gives better DC offset.
General Thought: Why class D?
I think that since amplifiers want to try to become unstable and oscillate, why not just go with the flow and just promote it?
I started with the idea of taking two N-ch MOSFETs and trying to design the simplest yet low distortion discreet circuit that I could come up with.
The output stage seems pretty well optimized, but I lack desire now to continue further optimizing the rest of the circuit. I would most likely tinker with a p2p version and if it could work much at all that way, I would be pretty confident about it.
The circuit values might be pretty close to optimum in the current mirror and the input and feedback circuits. You'd probably want to select the phase lead components as aggressively as possible in such a way to cause high frequency roll-off a little above the upper audio frequency limit. That would provide maximum load damping.
I show it for +/-40v operation. With some modifications, it ought to work on as much as +/- 100v with at least some performance loss. That's all for now. I will try to answer questions if I can.
I have left in some test options in the input and feedback circuit. I find that inverting input mode is cleaner while noninverting mode gives better DC offset.
General Thought: Why class D?
I think that since amplifiers want to try to become unstable and oscillate, why not just go with the flow and just promote it?
Attachments
luka said:
Thanks. I might be able to build it after I build and test out my circuit for driving 60Hz toroidal transformers for another project.
On the "N-Champ" project, I am weighing the option of bridging this circuit to negate power supply pumping. I might be able to bridge it without duplicating many parts. A bridge would also permit using lower voltage. Lower voltage MOSFETs can have pretty good specifications and a lower voltage circuit is easier to work with.
Not bridging means fewer components, but bigger filter capacitors to absorb pumping. The power supply can be rated for less peak power output and be smaller.
Hi Eectrone
Have you made a prototype and tested the circuit?
In my 2x150W car amp I use just 2x 4700uF supply capacitance yet pumping is not a problem. You have to evaluate it. Also, you have to try the output stage and everything to ensure that it works. Class D simulation is only reliable for control loop analysis.
BTW: I recommend differential input and feedback. You will find that the single ended arrangement picks up EMI resulting in circuit disturbance, particularly when the output is near the rails and pulses become narrow...
Have you made a prototype and tested the circuit?
In my 2x150W car amp I use just 2x 4700uF supply capacitance yet pumping is not a problem. You have to evaluate it. Also, you have to try the output stage and everything to ensure that it works. Class D simulation is only reliable for control loop analysis.
BTW: I recommend differential input and feedback. You will find that the single ended arrangement picks up EMI resulting in circuit disturbance, particularly when the output is near the rails and pulses become narrow...
Eva said:
Hi Eva,
I did something not completely different a long time ago and thought that I might have an opportunity to reenter the process of tinkering with real parts again in the future.
Your thoughts are very good. I am traumatized by a long-ago filter capacitor explosion which spewed some very nasty stuff everywhere.
You give me impetus to consider the bridge with differential input and feedback. I was just considering how to implement that topology yesterday.
Those sticking PN junctions at narrow pulse widths do their mischief!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.