Back in my early audio days I remember getting all the reference material on audio amplifier design that the college library had. I saw information on the "Tigersaurus" and later-published information on class G and then later, class H. The Tigersaurus was a very brute-force way to increase the SOA of the output stage and was based on the earlier Tiger design. It was later when I was given a copy of Electronics World that I was intrigued by the c. 1966 design of a class D amplifier.
I'm not sure at this time what to call the idea of modulating the hot side of a SMPS to produce a varying signal on the secondary side. Maybe that way offers no real advantage from regular class D, where the power supply voltage on the secondary side remains rather stable.
I'm not sure at this time what to call the idea of modulating the hot side of a SMPS to produce a varying signal on the secondary side. Maybe that way offers no real advantage from regular class D, where the power supply voltage on the secondary side remains rather stable.
Varying the supply voltage is known as "dynamic supply voltage" and can be used with analog, digital, and hybrid amplifiers. It is normally used to decrease supply voltage at low volume settings for better efficiency. It is not currently used for modulation of the supply voltage in time with the signal as it would be hard to design a power supply fast enough for that. However, it is possible to implement with digital by adding a FIFO buffer to add a delay.Electrone said:
There's an implementation for an analog amplifier that switches the supply voltage.
I've played with the idea a bit, first by driving a speaker in class A mode with the output of a modulated SMPS, and later by doing simulations of a power supply that provides tracking power to a subwoofer amplifier. The response of the power supply was limited, but it should work for a subwoofer amplifier.
Eva said:
I have to admit that I am mostly just doing this project as a an electronics diversion, not for listening. I am considering power supply pumping, size, cost, and complexity, also, although the unregulated power supply does tend to be simpler. Where this project might actually lead could be convenient for me but look difficult to others.
star882 said:
Less heatsinking could reduce size.
If you achieve low enough capacitance at the switching node, you can reach fully resonant switching when the class D stage is idle. There is no need to change supply voltage. This only requires a lowish output inductance to get enough resonant current to charge capacitances in time. This means no idle switching losses in the transistors, only in the cores of the inductors and in gate drive. Have you ever seen a class D amplifier producing +/-120V music on 4 ohms without any heatsink? It's one of my latest achievements 
Some TI chipsets adjust the supply voltage to the volume setting. Apart from efficiency improvements, it increases the effective resolution of the modulator. If 24 bits of resolution (or equivalent for a hybrid) seems hard enough, maintaining that at low volume settings without reducing the supply voltage is even harder, since it would have to be designed to an even higher resolution in order to get enough steps at low volumes.Eva said:If you achieve low enough capacitance at the switching node, you can reach fully resonant switching when the class D stage is idle. There is no need to change supply voltage. This only requires a lowish output inductance to get enough resonant current to charge capacitances in time. This means no idle switching losses in the transistors, only in the cores of the inductors and in gate drive. Have you ever seen a class D amplifier producing +/-120V music on 4 ohms without any heatsink? It's one of my latest achievements
I think you should design power electronics for electric vehicles. 3600W is considered tiny for the average EV, except electric and hybrid bicycles of course. I don't think it'll take much to apply your design to a hybrid bicycle inverter. Nearly 5HP without a heatsink would attract attention everywhere!
Electrone said:Congratulations, Eva. I'm too much of a romantic idealist to struggle long with the imperfect world of things like PN junctions remaining stuck on.
One of my pending (for a long time) projects is a SMPS with emitter switched bipolar transistors
but this is a pain for class D, modulators can't cope well with the (long and variable) turn off delay.Eva said:
One of my pending (for a long time) projects is a SMPS with emitter switched bipolar transistors
I can't recall ever seeing a MOSFET controlled by its source now that you've reminded me of your project.
I remember one configuration (cascode?) that had two MOSFETs in series. The lower MOSFET had the gate driven as normal. The upper MOSFET had the gate connected to a bypassed supply of about 15v or so. It is supposed to reduce the problem of stray capacitance (from drain to gate) slowing down turn-off since the upper MOSFET would effectively have a low impedance driving voltage and a low AC impedance from gate to ground.Electrone said:
I can't recall ever seeing a MOSFET controlled by its source now that you've reminded me of your project.
The lower switching element would work best with a MOSFET since it would deal with low voltages and low voltage, low resistance MOSFETs are commonly available. The upper switch could possibly be a bipolar, however.
Eva said:
One of my pending (for a long time) projects is a SMPS with emitter switched bipolar transistors
Is it really still worth using those cascodes considering how fast modern IGBT:s are?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.