Gallium Nitride GaN Class D Amplifiers
In my recent research for a possible upgrade to my current amp (Benchmark (AHB2) I was reading about the new higher end design for Class D. I'm very interested in learning more about these new GaN(Gallium Nitride) designs. Three companies are offering some very well reviewed products and they are not going crazy with Watts per channel:
Orchard Audio offers a 250 watt Amp
AGD a 100 watt
Atmos-Phere also a 100 watt
What's interesting is while Orchard is a new company AGD and Atmos-Phere have been around a while producing high end Tube amps. In almost every review it is noted how these newer designs sound like Class A or Tubes with all the benefits of Solid State. One reviewer couldn't go back to his tube amps after extended listening to the Orchard. No wonder AGD and Atmos-Phere are getting into this technology. It's very exciting as these amps are highly efficient turning over 90% of the power they draw into sound compared to about 78% with A/B designs and I believe even significantly lower for Class A. They run cool and usually weigh between 10-28lbs. I plan to do more investigation. Small size and light weight with decent power is very attractive. There are also mono block offerings for more power if needed. These are not cheap Class D products. They are well designed and constructed.
Orchard Audio's base model is around $2700and their Dual mono version with larger power supplies is $5500.
Both AGD and Atoms-Phere are $5000
Mono blocks for each are about twice the cost.
Has anyone had any experience, demos, etc ., with these types of GaN Class D amps?
@marco1 , @tvad What SQ characteristics do distinguish from the auditioned Class D Gan amps that cause you to still prefer tubes. I suspect I would agree with you and pick tubes but for my rooms allergy to heat. Have you ever experienced a class AB amp that got you closer to your tube preference ("soul"?) than a Class D Gan? If so, which one(s)? |
#1 on the list was 3D soundstage. Depth was more convincing using a 300B SET tube amp. #2 was texture. The tube amp conveyed a bit more. Regardless of what chip amp I’ve auditioned, they all sounded a bit sterile and artificial compared to the 300B tube amp. The AGD Audion GAN mosfet amp was the least so. I have never experienced a Class AB solid state amp in my system that came close to a tube amp. I have heard Class A that approached a tube amp - Pass Labs XA-30 and XA-60.
|
A while back I asked John Stronczer at Bel Canto about the GAN amplifier modules. His take was the folks at Hypex have a lot of experience with, and have done a lot of research with the NCore modules, and if there was an advantage to the GAN’s they would be in the forefront of their implementation. I’m not in tune with the industry like John is, but for what it’s worth, I do know Bel Canto class D amplifiers sound extremely good and natural. I’ve owned several over the past 10 years. Presently, I’m listening to a Rogue Audio Pharaoh 2 that also uses Hypex NCore class D modules, and which also sounds quite good in my system. I think it’s as much about the overall implementation as it is about the amplifier module in the circuit. Just my two cents worth. YMMV. |
I'm sorry, but I have to disagree with your statement that "They have low parasitics, but that only applies at high frequencies, not audio." The statement is completely inaccurate and misleading. The Class-D output stage runs at high frequency (our GaN modules at 800Khz), and the output MOSFET, whether Si-based or GaN-based, must have extremely low switching losses to operate efficiently. Any "parasitics" impact the efficiency and purity of the switching waveforms, causing slope variations, overshoots or undershoots during the transition. All of these factors affect the quality of sound reproduced. To better understand the fundamental principle of operation, it would help to look at the principle of operation of a switching amplifier (CLASS-D). Many articles are available online that can help you understand this better. |
@aw-agd If you read my post you will see that I’m assuming linear operation. I made the same mistake you did: I didn’t read carefully. Also, it turns out you actually agree with my statement you said you disagree with, since 800k is high frequency. And finally, I’m a PhD EE in power electronics who designs converters using GaN, so I’m already quite familiar, but thanks for your concern. |
