Please excuse this off-topic posting which doesn't involve US politics at all. I was just wondering what the efficacy of class D amps is when applied to the design of hi-fidelity audio amplifiers? I can't help thinking their functional limit would be public address systems where a fair amount of distortion can be tolerated, not to say expected even. But is it feasible to design a class D audio amplifier that the average, discriminating audiophile (not the complete nutters obviously) would find totally acceptable?
Why not? Distortion is arbitrary, given any possible range of designs. Not fast enough? Crank it up to 10MHz (maybe using those EPC GaN FETs) and wrap NFB around it! Can't afford the speed? Run a grueling pre-compensation calibration, feeding input through a ADC-conversion table-DAC. (Or just generate the PWM digitally.) Does your crappy amplifier exhibit time-dependent distortion as well? Measure and compensate that away, too!
Personally, I was going to build a class D tube amp... just to f**k with 'em. ;-)
Why would there be any intrinsic downside to class D from an audio quality perspective? In a completely hypothetical class D amplifier made from "perfect" devices, I don't think there should be any harmonic distortion at all. The incoming test sine wave PWMs a square wave, is applied two perfect switches, and then the hypothetical switching frequency is sufficiently high enough such that a perfect analog low pass filter of sufficient order can kill the switching frequency to an arbitrarily low level.
As far as I can tell this mathematical ideal is a completely LTI system, there are no active devices being used in areas where there are exponential or square-law transfer functions from the input port to the output port, etc...quite unlike a standard transistor Class AB power amp from the 1970s.
I don't think it's anything inherent in the topology, it's simply how much you're willing to trade off money, design time, and power consumption to approach the ideal. Sure, it's feasible - what do you want to throw at the problem?
Yep. There's nothing "wrong" with the topology, it's just the usual three-legged stool of engineering: good-sounding, cheap, high efficiency, pick two...; )
Several decades ago, Sony built and sold a hi-fi class-D audio amplifier wh ich was favourably reviewed in Hi-Fi News and Record Review.
As Tim Williams points out, a sufficiently rapid-switching output stage can push the switching noise up to frequencies where it can be easily filtered out.
You have to remember that Sigma-Delta A/D and D/A converters are using pure switching to convert between the analog and the digital regimes, and they perform well enough to keep even the golden-eared boys remarkably happy.
They do cycle the switches relatively fast and often to push the switching noise up to high frequencies where it is easier to filter - Watkinson's "Th e Art of the Digital Audio" ISBN- 0240-51270-7 was a full bottle on the sub ject, back in 1988 - I've got the 1989 revised reprint.
You used to be involved in audio - how come you weren't aware of that?
Just put the tubes in anyway, but don't connect them, drill a hole in the chassis and stick some warm orange colored LEDs shining up through the base. It really helps on the power consumption.
It's a highly subjective field, but smpa applications are already fairly widespread - particularly in portable battery-powered applications - or wherever efficiency, heat and cost are an issue. My pocket mp3 player even has pwm output, in order to run >12hrs continually, without recharging.
The main distortion sources in audio signal path fidelity are still the electro-mechanical transducers - microphone/pickup and speakers. For simpler lower-powered signals, well-matched commodity consumer hardware has never been better.
PA systems have their own issues in fidelity, unrelated to the methods of power amplification.
Personally, I think the program material content and availability will always be more important than it's method of transmission.
Music "these days" is just a 110 dB THUMP THUMP THUMP with fourteen detuned sawtooth waves going BZZ BZZ BZZ stacked on top so whatever amp is just fine.
AFAIK there's several "boom-box" quality chips that do just that -- I'm too lazy to go digging, but just checking with what TI has available might cough up some candidates.
I got kicked off of AK or discussing politics LOL. But I got other forums, fukum. This has been done to death on such forums. (fora ?) Technically the re are a couple of things in consideration. First of all is that since you are switching on and off completely, that removes any nonlinearity of the g ain of the outputs completely. If the PWM is done right, that should be an advantage.
From what I've read there are two types. One, at idle switches fully but at a perfect 50/50 duty cycle. Another one switches on the outputs only a lit tle bit at idle which is actually more efficient. That type only turns on t he outputs in a way to create bias of a sort.
And then Crown now has a patented (they claim) class D output circuit that is supposedly even more efficient because it dumps the inductive kick from the output filters back into the power supply. I understand the concept but how much you actually gain is a good question. Plus it has a higher compon ent count but Crown is not a low end company. The extra components are only a coil ad cap in the output filter, they need two instead of one per chann el.
Which brings us to the output filter. It all depends on the switching speed . That determines the output filter. Now, if you could go 10 MHz fine, but just try switching at that speed with precise PWM control. And then you get into RF and EMI problems.
But then, a regular CD puts out at 44.1 KHz and they sound pretty damn good . We have to put this in perspective.
** Hi-fi amplifiers operating in class D have been around for several decades, gradually taking over from class B and other schemes in both domestic and professional applications.
The arrival of MOSFETs was a game changer with switching speeds that allowed PWM frequencies of 250kHz or more with very low losses. THD figures are somewhat higher than the best linear designs but 0.1% across the audio band is easily enough achieved.
The advantages are reduced heatsink size, no need for fan cooing and a somewhat smaller PSU. If the amp has a SMPS as well, then there is also a major reduction in weight.
Eliminating RF noise from the output and immediate environment is an issue and it is rare to see a class D amplifier with AM tuner in the same box. Not many 5.1 channel receivers sold use class D.
"Done right" everything is perfect. However, there are the nonlinearities of the filter components to deal with, in addition to the modulator.
There are more modulation schemes than that. One can use tri-state drivers to reduce power, as well.
The energy gets stored back in the power supply caps so it doesn't have to be restored from the supply. I've seen some other pretty interesting schemes, too, but they're under NDA.
You do with any high-fidelity class-D amplifier. That's the purpose of the output filter. BTW, TI has a 2MHz class-D amplifier (most I've seen are in the 300-500kHz range).
owed PWM frequencies of 250kHz or more with >very low losses."
Regular BPTs could do that for quite some time but it took some current to drive them. With MOSFETs all you have to do is effectively charge and disch arge a capacitor. With a BJT to switch it that fast and get it saturated to ok some power which negated some of the efficiency. And you still had the c apacitance problem. You also had to go negative to make sure it turned off fast enough. So yeah, MOSFETs were a nice development.
omewhat smaller PSU. If the amp has a SMPS as >well, then there is also a m ajor reduction in weight. "
Now you're talking Icepower. Familiar with them ? They got fifty watt modul es, hundred watt modules and who knows what else. No heatsinks at all. The only heatsink is the copper on the board. Icepower modules are made by B&O which is considered a high end audio company.
ue and it is rare to see a class D >amplifier with AM tuner in the same box ."
I bet. The AM receiver would just about have to be in a Faraday cage. i rem ember, in this country, that if you had a color TV you had to keep AM radio s away from them if you wanted to hear anything useful. The sound and color subcarriers combined to interfere with the IF frequency of 455 KHz. If you switch at 250 KHz, the second harmonic is pretty close. (in more ways than one)
The bottom line with class D amps is how they do the PWM and the feedback. Ideally the PWM would be done with a comparator fed by a sawtooth wave from an oscillator. As long as the inductor at the output does not saturate the n it is pretty linear. (or predictable) the main other factor is the feedba ck and the open loop gain.
Audiophiles, and I mean ones who know what they're doing have come out with having alot of gain and cutting it down with negative feedback does not qu ite do it. Somehow it sounds different. I would love to have a sit down and a few beers with Bob Carver to talk about this, he probably knows more tha n anyone else on the planet about this. But if he is even still alive he is probably in Depends and a wheelchair. He was an engineer when I was ten ye ars old.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.