Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Benefits of Class-G and Class-H Boost in Audio Amplifiers [pdf] (ti.com)
48 points by type0 on June 24, 2022 | hide | past | favorite | 31 comments


If anyone wants to easily get into Class D amps, Parts Express has private label Dayton Audio amps [1] with a variety of features and wattage levels. They use Wondom/Sure as their supplier [2].

They are pretty feature rich: they support I2S, Bluetooth, and Sigma Studio for DSP effects (using the Analog Devices ADAU170 IC).

I myself am pricing out a kit so I can build a ammo-can speaker box. I'll probably go with Parts Express as it's a 1.5hr drive to visit the brick and mortar store, and I can get the speaker drivers as well (plus other components).

[1]: https://www.parts-express.com/electronic-parts/audio-amplifi... [2]: https://store.sure-electronics.com/


I wasn't happy with the cheap adau1701 boards. The integrated ADC isn't good, and they only have limited filter capability.

I'm returned mine and got a hifiberry beocreate 4 [1]: DSP + four amps. The adau1451 is much more powerful, and the ADC is pretty decent. As a crossover I use FIR filters, so the four drivers of my small stereo speakers are hooked up to one dedicated channel each. But like with classical crossovers, you need a measurement mic to verify the correct behaviour; unless someone already did that for you.

The adau1452 in the freedsp aurora[2] is even nicer (longer FIR filters), I use that for my big stereo speakers.

What I really recommend/want is skipping the ADC altogether and getting something that connects the DSP (preferably adau145x) digitally to the amps. Both these devices fall short in the regard, as they feed the digital signal into ADCs. But they're both still really good.

Hint if you consider the beocreate: don't use it with a rpi4, that one runs too hot; a rpi3 is perfectly fine for their OS. Once the DSP program is installed you can remove the rpi and use the optical input; or you can use the rpi as the source.

[1] https://www.hifiberry.com/shop/boards/beocreate-4-channel-am... [2] https://github.com/freeDSP/freeDSP-aurora


Thanks for the tip on the Beocreate 4! I’ve been looking for something with a digital input, DSP, and 4 amplified channels. This seems like a sweet spot for me because you can do an active crossover for a stereo pair of two way speakers in one unit. This is a perfect fit.


I did ask some questions about the custom DSP program in 2020, not sure if that still applies. You might still find this useful, especially toumas explanations: https://github.com/bang-olufsen/create/issues/116

By now I just use their OS on a spare RPi. And of course I didn't get around to making Arch Linux AUR packages for their tools ^^"


Very cool! I'll check these out!


I love parts express. They have the stuff to make yourself very high end speakers and a lot of very high quality amplifier and other electronics modules.


I'm not an expert on audio, but what's the difference between [1], and the reference design of [2]?

I know only a bit about electronics, but I would have figured the market would have driven those two things to be roughly the same price if they were equivalent.

[1]: https://www.parts-express.com/ICEpower-200ASC-Class-D-Audio-... [2]: https://www.ti.com/lit/ds/symlink/tpa3221.pdf?ts=16560824848...


From what I know ICEpower chips are intended for (and used in) "high end" applications while the TI ones are intended for general purpose consumer devices. At least that's how they're marketed and an amplifier manufacturer will often state specifically that they use an ICEpower chip in their amp. Whether or not one can hear the difference is another story.


Honestly, no difference besides a red PCB. Most likely the brand recognition and possible(?) warranty for US sales. Price probably includes the US support (if any) if you need help.


ICEPower amps tend to objectively outperform cheaper Class D amps in terms of distortion while also putting out more power.

However, as other commenters have noted, this may or may not translate into audible benefits depending on one's use case.


I bet. But for hobbiest applications, which is better? Wondom/Sure provides a wealth of application info. Most people aren't building something to power Hi-fi systems, but more to just be loud or portable. Low THD is good but it's a matter of finishing George. As someone over 40, I doubt i could tell the difference from .1% THD and 1-5% honestly.


You're right to be skeptical about anything in the audio industry. The decades-long tide of absolute snake oil is massive and it is toxic. Additionally, you're absolutely correct about diminishing returns being an extremely real thing in audio land.

    But for hobbiest applications, which is better?
For the purposes of this reply, I'm defining "correct" sound as "low in distortion, with good dispersion characteristics -- ie the sound remains the same over a wide listening angle"

There has been some pretty decent randomized research from Toole, Olive, etc. that both trained and untrained listeners have a preference for correct sound. It is something that nearly everybody enjoys more, even if they don't know why. In the sense that most people who like butter would enjoy high-quality, grass-fed butter more than they would enjoy the cheapest possible stuff, even if they didn't give a whit about where the butter came from.

For sound... as with any engineering challenge, it's always going to be a tug of war between desirable characteristics such as price, size, power consumption, sound quality, how loud it gets, etc.

Anyway.

If you're building some kind of portable boombox - I highly doubt the extra performance provided by ICEpower would be worth the premium.

If you're listening to something at home in a decent listening environment... then maybe it would be worth it. Even a quiet room is going to have a noise floor of like 40+ dB.

However, at the end of the day, the speakers themselves make the biggest difference.


finishing george = diminishing returns.. stupid mobile.


Meanwhile I'm wishing to have a local speaker setup that can hit 80 dB SNR... if you count all non-linearity-caused components as "noise". For comparison, your .1% THD is 30 dB.

I explicitly don't care how the non-linearity is corrected, just that it does.


That ICEpower PCB amp looks like an affordable little unit.

Big capacitor is 330uf 400V.

You should hear what one of those caps "sounds" like when included in its own kind of vacuum tube circuit.


Good times in the 2000s with the Tripath amplifier chips. Vinnie Rossi took them out of mass-market Technics surround sound boxes and created some really nice amplifiers.

Then he rolled his own, using batteries to drive the amplifiers, independent of the AC mains electric supply. A 2007 deep dive at Six Moons warning very heavy audiophile reality

https://6moons.com/audioreviews/redwine8/

I have a base model Signature amplifier and I really like it. Rossi followed up with improved models that replaced the sealed lead-acid batteries with LiPo, and then ditched the batteries altogether with a super-capacitor design.

I can't find the old company, maybe it is no longer in business. But looks like Rossi has a web site still up at https://www.vinnierossi.com


This seems like only half way to the ideal solution...

The ideal solution would be to use a DC-DC buck boost power supply to directly supply the speaker. Get rid of the intermediate power rail and ESR losses by effectively converting the power twice.

Obviously you will need to do some work on the control algorithms, because rather than your power supply aiming for a nice stable output voltage, you're now aiming for a voltage waveform exactly matching the target signal. You also no longer care directly about ripple voltage, but do care about linearity and THD.


Class H sounds a lot like envelope tracking in RF.


I wonder why these haven’t become more well known among Audio folks, more power at higher efficiency is a good thing, if it comes with suitably appropriate low levels of undesirable distortion


Almost all of the top-rated audio amplifiers use class-D switching already; the class G and H changes are only needed in battery-driven devices.

E.g.: https://www.audiosciencereview.com/forum/index.php?threads/t...


A lot of the stuff at the high end of the chart is using not only the same architecture, but the same modules from the same supplier. Interesting that the class-AB amp that has been at the top of their ratings for years has finally been beaten by a switching amp.


Is the Benchmark AHB2 (the previous champ) really class AB? I think it's more like a switching (class D) amp.

Those AHB2's are tiny and run cool while putting out a few hundred watts of power - I think that rules out class AB.


They’re small and cool because they have the “class-h” active rail system described in this app note. Someone else said that’s only for battery-powered mobile devices but that’s not true.


I thought it was the opposite; class-D is more likely in battery powered because it is higher efficiency than G or H, and G or H are used in high powered amplifiers that have an AB output stage but are thermally limited?


It would be nice to do something about efficiency in the left half of the charts too.

I understand that my Bluetooth speaker only lasts a few hours when cranked up to full volume. But when I have it on as quiet background music I'd really like it to last a week.

Audio is perceptually logarithmic, so quiet sounds use very little energy to produce. That means that with carefully designed stuff we should be able to get really long battery lives.


Interesting. Are there any products that use this? Development boards?

I hope to build a musical instrument someday with an integrated speaker.


This has been around for years. Self's webpage [1] notes that class G amps date back to 1976. They've certainly been used in commercial applications and I'm certain you could buy PCB kits at some point in history.

Apparently, there were a bunch of class G amps that appeared after 5.1 setups got popular, just because of how much less heat they produce per channel.

If you want an efficient, small amplifier these days, class D is also an option. They're cheap, small, and powerful. Much more efficient than class G. If you are just trying to stick a speaker and amp in a musical instrument, I would start by getting an amp that is built into an IC with something that can work with a relatively simple power supply.

[1]: http://www.douglas-self.com/ampins/classg/g.htm


Among the things that have happened since 1976 are super high performance MOSFETs, that play directly into the hands of Class-D amplifiers. But actually making Class-D amps robust involves a steep learning curve and an expensive oscilloscope. This is the reason why, for example, musical instrument amplifiers tend to use pre-made Class-D amp modules such as those made by IcePower.

The amplifier that I use for playing the bass is Class-D, as are all of the audio amplifiers feeding speakers in my house.


Hmm. IcePower looks pricey and it's probably not for portable, battery-powered audio?


They're not for battery power, since a major part of their appeal is integration of the power supply and amplifier on a single board. Some of the other modules sold by Parts Express are designed for lower voltage DC power. Then it becomes a matter of how much power you need, for how long, and what that translates into in terms of battery capacity. Speaker efficiency also plays an obvious role.


When I've done my own designs, the output filter of the class D is rather sensitive to which speaker is used; it seems like class D is great when you are building an integrated product (HTIB (do those even exist anymore?). powered speaker, instrument amp &c.), but H would be better for separates.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: