DIYMobileAudio.com Car Stereo Forum banner

GaN FET Based Amplifiers

38K views 88 replies 32 participants last post by  lpreston 
#1 ·
There has been some recent comments on Facebook regarding a "revolutionary" new amplifier from Soundigital out of Brazil. The key to the new amp is apparently the use of GaN FETs. One of the only other applications I have seen on the audio side of things is a couple of systems from Technics. So, any of the DIYMA amp gurus have any feedback?

The specific Soundigital amp in question, the SD300.2D...
Soundigital Amplifiers - Lançamento SD300.2 GaN Power

http://soundigitalusa.com/wp-content/uploads/2016/02/final_final.pdf






More info on GaN (Gallium Nitride) FETs...
Class D Audio

EPC9106 EPC | Programmers, Development Systems | DigiKey

The Panasonic/Technics use of the tech...
Panasonic breathes new life into Technics
 
See less See more
4
#3 ·
I was hoping for a bit more feedback than this, but yes, the EPC 9106 development board is $1338. If the Soundigital SD300.2D uses the same (or similar) board, it would explain the $1500-$2000 retail pricing that was mentioned on FB.

Among some other benefits stated by the EPC, one of the things that stuck out to me was the switching speed of 400+ kHz. This is on par with what JL is doing with their HD amps and greater than what they have in their XD amps. I have no idea how that compares to other Class D amps, but since we see switching speeds this high in the HDs I don't believe this alone will make GaN FETs a "game changer."

I was really hoping Ricky, Victor, deveds50 and some of the other amp pros around here might be willing to chime in.
 
#4 ·
Forgive me, but doesn't increased switching speed mean lower efficiency? I always thought you wanted just enough speed, not too much for audio.

Eric
 
#5 ·
I have no idea, but I would guess no. Just based on the claims of 90+% efficiency.
 
#7 ·
I can tell you that I will be using them in the near future I knew about these back in Dec and got to see the amp at SBN. So Until I have mine I can't say much, but will definitely be interesting..

I am currently running 2 - soundigital 800.4 and switching out to 2 400.4D and one 800.4, so far, I'm very happy with the power in the small package.. lot less power consumption and having the ability to have my hatch back..
 
#8 ·
I look forward to seeing your feedback.
 
#9 ·
I want to know what the middle of the paragraph in the first image is referring to. Something about QRR? QPR? I can't really read it. It says it's related to distortion, and class D amps have low amounts of it, but these new transistors have zero amounts of it. What is that all about?
 
#10 ·
I know absolutely nothing about this (hence my initial post), but I can use google and "Ctrl+C" and "Ctrl-V" like nobodies business!

EPC eGaN FETs and ICs for Class D Audio Applications said:
The quality of sound reproduced by the audio amplifier, measured by THD (Total Harmonic Distortion), DF (damping factor), and IMD (intermodulation distortion) is influenced by the characteristics of the power transistors used.

eGaN FETs’ near ideal switching performance due to lower propagation delays and faster slew rates (due to their lower gate capacitance) and zero QRR enable very short dead times to provide lower open loop distortion, lowering the THD and overall losses. This reduces feedback, driving down T-IMD and DF to provide a step jump in the sonic quality of Class-D audio amplifiers and lowering overall losses.
http://epc-co.com/epc/Portals/0/epc/documents/briefs/AB003%20eGaN%20FETs%20for%20Class-D%20Audio.pdf

This paper from Infineon might also be worth taking a look at...
http://www.infineon.com/dgdl/Infineon+-+Application+Note+-+Power+MOSFETs+-+OptiMOS+200V+250V.pdf?fileId=db3a304344ae06150144b1d2f8250165
 
#11 ·
Yeah I saw that FB post and saw Steve's praise of them among a few others, though of course folks balked at the looks of it, which I couldn't care less. I'll be interested in what folks report on the GaN amp's abilities. Appears they'll be doing a 4-channel as well.
 
#12 ·
I saw the praise as well, but I really have to wonder if these things are really all that special? Or is this just hype?

The closest thing out there in terms of similar power and class D is probably the Mosconi D2 150.2. From what I have seen those D2 amps are well received. I would imagine that the limited edition Soundigital GaN at $1500-$2000 would be a pretty significant premium over the D2 (not sure what retail is). Is there enough difference in this new amp tech to make it worth it? I'll maintain a healthy dose of skepticism until I see some independent reviews/tests/commentary.
 
#15 ·
I read some of the information about GaN on Texas Instrument's site. I agree, these seem really interesting, but I'm skeptical as to whether or not they'll offer a practical improvement, as opposed to a theoretical one. Current class D amps are kicking ass fullrange, they offer great FR and low distortion. These chips may be better on paper, but I wonder if in the real world their advantages are going to matter.
 
#16 ·
Not an amp pro, but I know just enough about electronics to be dangerous.

GaN FETs have a lower on-resistance and switch off faster and harder than silicon. For Class D amps, this has a bunch of benefits.

First, power losses are reduced (i.e. efficiency goes up). So you need less heat sink. And because the GaN FETs conduct electricity so much better than Si, the devices themselves are smaller. This means smaller amps that run cooler at the same or higher power levels.

GaN FETs switch much faster than Si. So you can increase switching frequency without losing too much efficiency. This in turn allows smaller inductors in the power supply, EMI, and output filters. It also allows the feedback loop and output filter to reduce their effect on the audio band, for better transient response and slew rate.

GaN is harder to manufacture than Si, but it uses a lot of the same techniques and equipment. Because the conductivity per unit area is so much better than Si, devices are smaller, and you can put more of them on a wafer. This should bring costs down pretty quickly as the chip factories gain experience. EPC is already offering (or claiming) GaN devices cheaper than the equivalent Si FETs.

GaN doesn't require a new design approach in the same way that MOSFETs did over bipolar transistors, or bipolars over tubes. But GaN makes radically better power switching MOSFETs than silicon. The only real drawbacks are unfamiliarity and device prices, and the device costs are coming down pretty quickly.

I think in 15 years we'll be saying "Remember silicon MOSFETS?".
 
#20 ·
:rolleyes:

what about GaAs or SiGe?
what about organic developments?
Si mosfets are here to stay.
And BTW unfamiliarity is a not a problem as supporting circultry is more similar than you it seems think.
 
#17 ·
Thanks for the response chucko. A lot of what you're stating is what I was picking up looking at EPC's website and videos.

I guess the question for today is, do they warrant the significantly higher price tag over traditional MOSFET Class D amps? Either through a recognizable audible difference, increased efficiency, smaller footprint (via reduced heatink), etc.

It would be interesting to see how the Soundigital amp fared in a Richard Clark or similar bling listening test.
 
#18 ·
I guess the question for today is, do they warrant the significantly higher price tag over traditional MOSFET Class D amps? Either through a recognizable audible difference, increased efficiency, smaller footprint (via reduced heatink), etc.
My opinion? Not today. The audible differences would be slight; maybe some golden-eared folks might find them worth the money.

The EPC amp is a demo piece. It's intended for electrical engineers, not for audioheads. A mass-produced board would be cheaper, even using current parts.

I think as GaN sees wider adoption in other areas (e.g. power supplies, where efficiency and power density in space and weight are big deals), mass production will bring device prices down, and we'll start to see some radically powerful and tiny amps at a modest premium or even a discount to current silicon Class D.
 
#22 ·
Nope,slap decent case on it sell it on diyma for reasonable money and sell lisense to big names so they can charge you 2k.for those people to whom Victory Sonics is not good enough brand to buy.
Story of my life.

Sent from my SAMSUNG-SM-G925A using Tapatalk
 
#23 ·
Nope,slap decent case on it sell it on diyma for reasonable money and sell lisense to big names so they can charge you 2k.for those people to whom Victory Sonics is not good enough brand to buy.
Story of my life.

(I couldn't bring myself to post One Direction.)
 
#24 ·
Hello Guys !

SounDigital GaN amplifier uses a IR/Infineon solution, it is a cascode configuration (GaN + Mosfet) to make a perfect device. There's no information on the market about IR GaN, few companies received access to use this GaN.

The advantage on GaN is the fast swiching, low RDS on, low Qrr, etc. It means that we can run in a higher frequency, (1Mhz on SD GaN), then the output low pass filter is smaller and the class D can reach a higher top end. The dead time is lower, improving the distortion and also the Qrr is zero. The THD at 1Khz at a half of the power is 0,002% and the S/N ratio is better than 100dB, there's no attenuation at 20Khz. Impressive numbers for a class D amp. The sound is very natural and free of ground noise.

The cost for low production is high yet, it is a numbered 200 units of a limited series. All new technology starts expensive and fall the price in mass production, we hope to sell this product cheaper in the future.

We have developed the GaN just to show the SounDigital technology in produce the first GaN based car amplifier ever. New comertial products are in development and I hope you can listen a SounDigital amplifier to prove that we are working hard to do the best.

Thank You !
 
#25 ·
Thank you for visiting this thread and giving us the opportunity to ask you questions about your product. I have three:

Can you explain what Qrr is and how a lower value is beneficial?

What is this "dead time" you refer to, and why is a smaller value better?

Also, I heard there is no gain knob on the amplifier you put into production. What was the reason for this? Do you plan to include a gain knob on future products?
 
#31 ·
I volunteer. A head to head with a PDX V9 and F4. Just need one. Ears have heard pure class A. Ears have heard better A/B. Hit me up Soundigital. :)
 
#29 ·
I might be making a big deal about nothing. The specs on the website state input sensitivity is 0.5-6.0v. This conflicts with reports I've read that it has no adjustable input sensitivity. It would be nice to know one way or another. Or see a picture of the ends of the amplifier showing the controls available (or lack of).
 
#30 ·
I see... Hopefully SounDigital will uncover that mystery.
variable sensitivity must have some kind of adjustment, maybe automatic.
Many power amplifiers designed today has fixed sensitivity rating.
most car audio amplifiers has built in preamp stage.
 
#34 ·
Thank you for answering my questions. I have two more.

On your website the input sensitivity is listed as 0.5 - 6.0 vPP
Does that mean your amplifier will produce full power with a 2v input signal?

Also, you claim that not including a gain adjustment will reduce the noise of the amplifier. Did you build and measure a design with a gain adjustment to compare to the design without one? What was the result of that test? I am really interested to see how much of an improvement is possible.
 
#35 ·
The information about 0,5 - 6Vpp is wrong, the correct input sensitivity is 1,5 RMS fixed.

Any opamp, even the more quiet produce some noise, you can find the input noise on datasheet, the parameter is shown in (nV√Hz). When you add a opamp to set the gain and make some filter you are adding noise, more opamps, more noise. Also, a trimpot or a switch are antennas to capture noise. We prefer a more pure sound instead some functions on this product and we reach what we want. The background noise is perfect.

Thank you !
 
#39 ·
He told me it was on his midrange.. He's running a Beyma midrange and I need to see how low it gts but, but, I'm intrigued now..

And no, I'm not saying this because I'm associated with Beyma or with the Importer of Soundigital Amps.. I'm saying this as my honest opinion, I'm already using Soundigital amps in my car and they sound good, but this amp takes it up a level..
 
#40 ·
Well any and all newcomers with a quality product or new innovation to the US market I'm always a fan.


Sent from my iPhone using Tapatalk
 
#45 ·
Nope. Completely different amp. Try to keep up. Start with the OP.
 
  • Like
Reactions: leepersc
#51 ·
I have the 800.4 from that line clear top w/ fans and then 2 - 400.4 from the evo line. I'm planning on having 2 of the gan amps for meca finals. Working in qualifying so i can redo my install.

Break down: 400.4 are bridged for the mids and tweeters (200w per speaker), 800.4 runs the 8" mid bass and 10" subs, both getting 200w a piece.

The evo line is no slouch either.. holding up pretty well.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top