Is GaN worth it? ICEpower article

[Matias]

Very interesting new article from ICEpower comparing GaN transistors to regular MOSFET ones, their pros and cons.

Is GaN worth it? - ICEpower

icepoweraudio.com

 

[welwynnick]

GaN transistors were discussed here as well:

Atma-Sphere Class D GaNFET Amplifiers

Atma-Sphere showed new class D GaNFET monoblock amplifiers at this weekend's Florida Audio Expo. I did not attend as I did not want to tempt myself into spending more money. I am very happy with my system now and already know which components I will be changing out. I borrowed these images from...

www.audiosciencereview.com

 

[boXem]

Too bad the reliability is not addressed.

 

[MaxwellsEq]

Very interesting new article from ICEpower comparing GaN transistors to regular MOSFET ones, their pros and cons.

Is GaN worth it? - ICEpower

icepoweraudio.com

Interesting. Thanks.

 

[tmtomh]

So it appears the performance gains are minimal/nonexistent, and the main benefit is in lower power consumption, especially at idle. So no practical benefit unless or until GaN switches come down in price to be competitive with MOSFETs I suppose. Yes?

 

[ZolaIII]

No it's not! FD-SOI would be but it never really happened.

 

[kemmler3D]

the main benefit is in lower power consumption, especially at idle.

This would make GaN based amplification really appealing for battery-powered devices, I guess? Today this is solved by using efficient drivers, class D, and large batteries. If GaN can become cheap, it will reduce the size of the battery required, meaning we can look forward to lighter, cheaper, louder speakers being carried around in public playing Top 40 and whatever. Yay?

 

[jooc]

This would make GaN based amplification really appealing for battery-powered devices, I guess? Today this is solved by using efficient drivers, class D, and large batteries. If GaN can become cheap, it will reduce the size of the battery required, meaning we can look forward to lighter, cheaper, louder speakers being carried around in public playing Top 40 and whatever. Yay?

Devices that talk to us is going to be one growing application. There will be a multitude of devices linked to the internet and various AI engines, and GaN would allow mellifluous, cleanly-amplified voice interface with minimal power requirements.

 

[kemmler3D]

There will be a multitude of devices linked to the internet and various AI engines, and GaN would allow mellifluous, cleanly-amplified voice interface with minimal power requirements.

There are already lots of such devices (Amazon Alexa, others) and IMO the limiting factor on nice vocal sound quality is not power, but finding the $ and space budget for a very high-quality full-range driver (or multiple drivers.) in a small device.

 

[Sir Sanders Zingmore]

Devices that talk to us is going to be one growing application. There will be a multitude of devices linked to the internet and various AI engines, and GaN would allow mellifluous, cleanly-amplified voice interface with minimal power requirements.

I for one, welcome our squawkie robotic-voiced AI overlords

 

[DVDdoug]

Today this is solved by using efficient drivers, class D, and large batteries. If GaN can become cheap, it will reduce the size of the battery required,

Class D amplifiers are already highly efficient so there's only so-much improvement to be had. You can only reduce the battery if you can further-reduce efficiency. If you can go from 90 to 95% efficiency that's a 5% improvement in battery life (or a 5% smaller battery) and a 5% reduction in heat.

The main advantage of efficiency is that you can get more power out of an amplifier without burning-up the components inside. Most of the highest-power amps these days are class D. (Or you can make it smaller & lighter with smaller heatsinks and no fan, etc.).

Most amplifiers are already free of audible defects.

If GaN (or anything else) can give me more for my money (more power in the case of a power amp) THEN I might be interested!

Generally, I don't care what's inside my amp. I know nothing about the inside of my AVR but I assume the amplifiers are class D, made with MOSFETs or MOSFET integrated circuits.



P.S.
Most of the efficiency/inefficiency of class A/B or class A amplifiers is related to the basic class A/B or class A design and whether you use a transistor, MOSFET, or GaN isn't going to make much difference. (MOSFETs seem to be able to handle more power than regular bipolar transistors.)

In a class D (switching) amp, MOSFETs tend to switch-on more-completely than regular transistors (lower on-resistance) so less power is wasted, they heat-up less, and you can build a more powerful amplifier.

Apparently GaN can switch faster, which means it's partially-on for a shorter time and too means it can run cooler (or at a higher switching frequency). So you could gain efficiency but the question is, is that important if you are already at around 90%? Maybe... Mybe not...

P.P.S.
Manufacturer's ALWAYS tout whatever design choices they make as some kind of feature or advantage...

"FET op-amp circuitry!"
"MOSFETS!"
"Discrete bipolar electronics!"
"Vacuum tube technology!"
"Point-to-point hand wiring"

 

[Sokel]

So it appears the performance gains are minimal/nonexistent, and the main benefit is in lower power consumption, especially at idle. So no practical benefit unless or until GaN switches come down in price to be competitive with MOSFETs I suppose. Yes?

Not so minimal if you consider that the high power ones (as all high power ones) are not so innocent in idle.
My 1200as2+300a2 combo sits on 40w idle (1200as2 alone is 31 watt).

And that 6.67Khz is sweet,both in Mosfet and GaN.

 

[Vhond]

Hifirose's newest integrated (RA280) is also using 'next generation GAN FET's'
HifiRose RA280. Marketing or not??

 

[Zapper]

Class D amplifiers are already highly efficient so there's only so-much improvement to be had. You can only reduce the battery if you can further-reduce efficiency. If you can go from 90 to 95% efficiency that's a 5% improvement in battery life (or a 5% smaller battery) and a 5% reduction in heat.

I agree that potential efficiency gains are small from the perspective of power usage. But the reduction in power dissipation is relatively large. In your example, going from 90% to 95% efficiency halves the dissipation in the device. This allows much more compact packaging.

Does that matter in audio? Maybe for a portable high power system such as premium BT speakers. Maybe high end car audio. It will be unattractive for very compact applications like earbuds where the amp is on the SoC. For home audio, GaN might have some performance benefits and some definite marketing benefits.

 

[Killingbeans]

Too bad the reliability is not addressed.

Mostly discrete GaN FETs? Do the chips with integrated drivers have the same reliability issues?

 

[boXem]

Mostly discrete GaN FETs? Do the chips with integrated drivers have the same reliability issues?

I was told by an ex. TI guy that they (TI) experimented with GaN for audio and that the reliability was not exactly stellar. No more specifics, sorry.

 

[PHD]

GaN class D is only worth it if operated at several MHz fsw range.

 

[wwenze]

Battery life is decided more by quiescent power than transient power, especially when most users don't use transient power more than a fraction of a watt.

So GaN helps because even when outputting no signal voltage the transistors are always still switching. (This is also why your 200Vcc 30A monster class D eats way more quiescent power than that 2W radio thingy)

More savings are probably to be had by using inductors and capacitors of smaller size due to the higher frequency.

 

[DonH56]

I only skimmed the article and barely read the other comments here so probably duplicating, sorry. My perspective is as a design engineer who has used GaN for RF and seen them used in power supplies (which switch at around 200 kHz to 1-2 MHz). Device physics was not my career focus, though I did design a few as part of various research projects, but GaN I have mostly observed from a distance. Take all this as somewhat informed opinion sprinkled with a bit of practical experience and theoretical analysis.

IMO, the power benefits of GaN should be mostly at idle, where less charge storage can lead to reduced idle losses, and at very high power where you might gain a hair in efficiency. I would expect the biggest power benefits at idle for audio circuits. This is supported by various articles touting GaN for low-power devices, but they are also a focus for high-power applications such as EV battery circuits where every little bit of efficiency matters.

GaN should allow higher switching speeds, moving the switching noise further beyond the audio band and making it easier to filter. Audible benefits debatable, of course. Higher speeds allow smaller components but increase the impact of component and board layout parasitics, such as inductor/capacitor series resistance (ESR) and parasitic board layout capacitance and trace resistance/inductance.

Lower charge storage and lower ON resistance can lead to improved distortion. Certain device/material characteristics also contribute to lower distortion. Whether or not this is audible I do not know; my opinion is that audio amplifiers for the most part already exhibit very low distortion well below audibility especially compared to speakers and the source material.

GaN reliability is a big topic. It has been around for a long time and commercially viable for 5~10 years or so -- I think. GaN is less thermally conductive than Si so thermal management may be a bigger issue, somewhat offset by their lower loss (so less heat to dissipate). From a materials viewpoint GaN is harder to manufacture and impurities more difficult to prevent, but some of that is manufacturing immaturity relative to Si. The result is that, as recently as a few months ago (last article on the topic I recall), GaN reliability still lagged Si but was improving. Reliability will improve as manufacturing improves. I have not read much about GaN reliability so do not know much about it other than general handwaving and memory of the few articles I have read.

FWIWFM - Don

 

[kemmler3D]

I only skimmed the article and barely read the other comments here so probably duplicating, sorry. My perspective is as a design engineer who has used GaN for RF and seen them used in power supplies (which switch at around 200 kHz to 1-2 MHz). Device physics was not my career focus, though I did design a few as part of various research projects, but GaN I have mostly observed from a distance. Take all this as somewhat informed opinion sprinkled with a bit of practical experience and theoretical analysis.

IMO, the power benefits of GaN should be mostly at idle, where less charge storage can lead to reduced idle losses, and at very high power where you might gain a hair in efficiency. I would expect the biggest power benefits at idle for audio circuits. This is supported by various articles touting GaN for low-power devices, but they are also a focus for high-power applications such as EV battery circuits where every little bit of efficiency matters.

GaN should allow higher switching speeds, moving the switching noise further beyond the audio band and making it easier to filter. Audible benefits debatable, of course. Higher speeds allow smaller components but increase the impact of component and board layout parasitics, such as inductor/capacitor series resistance (ESR) and parasitic board layout capacitance and trace resistance/inductance.

Lower charge storage and lower ON resistance can lead to improved distortion. Certain device/material characteristics also contribute to lower distortion. Whether or not this is audible I do not know; my opinion is that audio amplifiers for the most part already exhibit very low distortion well below audibility especially compared to speakers and the source material.

GaN reliability is a big topic. It has been around for a long time and commercially viable for 5~10 years or so -- I think. GaN is less thermally conductive than Si so thermal management may be a bigger issue, somewhat offset by their lower loss (so less heat to dissipate). From a materials viewpoint GaN is harder to manufacture and impurities more difficult to prevent, but some of that is manufacturing immaturity relative to Si. The result is that, as recently as a few months ago (last article on the topic I recall), GaN reliability still lagged Si but was improving. Reliability will improve as manufacturing improves. I have not read much about GaN reliability so do not know much about it other than general handwaving and memory of the few articles I have read.

FWIWFM - Don

Non-engineer question: Do the 'idle power benefits' apply only when no signal is present, or would there be some partial benefit when the signal is at low amplitude? As we know, the RMS of music is pretty low compared to peak values most of the time.

Portable speakers don't often sit idle for very long (auto-off with no signal after ~5 mins is common) but if the efficiency also applies to some extent when it's not 100% idle I could see it being helpful.