All this trashing about over whether or not the GaN Peachtree amps are digital or analog is just a red herring. What does it really matter? What matters is their performance- are they competitive with other products in the market place? Does their use of GaN fets yield some sort of performance advantage over the competition? The answer is clear- NO! There are several other class d amps available with far better performance. Are the inherent advantages of GaN fets being exploited in their circuit? NO! Is their circuit state of the art with some cleverly designed IP? NO! Is it just more marketing audio snake oil? YES! There are far better class d amps available from Hypex, Purifi, and if you must have the flavor of the day GaN, Orchard Audio. All have stellar performance, state of the art design, and no bs marketing claims. 16 pages of wasted time on an inferior product.
That linked article describes a method of generating an anlog sawtooth. DACs put out analog signals.
The important aspect is if the payload is digitised or not... how one generates the sawtooth signal does not make it a digital små. Almost all amps has an on and off switch - don't make them digital 😀
//
//
I totally agree. Getting all excited about the technical difference between digital and switching doesn't really help consumers. If you want to market a an audio switching amp or switching power supply as a "digital" it doesn't change consumer expectations or lead them astray. They still expect it to supply power, perhaps lighter, smaller, cooler running with better noise rejection than a non-switching design. In the minds of most. If it doesn't perform as well as expected, calling it digital wasn't the fault. Maybe it's important in an engineering or scientific setting to avoid some kind of confusion on theoretical terms.
What does it mean to digitise the payload? I assume you are referring to the payload as specific shapes carried by an anlog waveform intended to create timed switching behavior when fed into an appropriate device? But not just timed switching behavior as you've already said that alone is insufficient. What's the missing ingredient? Does it have to be specifically a PCM signal?
The payload in a audio system is the music signal. Tehn there are a bunch of control and support functions / signalling to be catered for. Like bias to an amp etc...
//
//
As an individual component on paper specs maybe not. But sonically maybe yes, but used in the amps talked about here without listening to them it's comparing apples with oranges
In the 2 new Peachtree amps they aren't using the same topology far from it, just for starters one being analog input the other direct digital input from digital sources a whole stage is eliminated, also one is used with very high amount of feedback and the other without any global feedback.
Audio is done finally with the ear, the last time I heard it
Cheers George
Last edited:
Lets do some statement for statement review of the product page tech claims.... (as per link in post #1) here is from their web-page - text coloured red by me:
First and foremost is the GaN-FET amplifier module.
No GaN-FET "module" - just discrete transistors. "Module" sounds professional, mystic and impressive for the common customer...
It has several inherent advantages in a power amplifier that even the best MOS-FET designs simply cannot achieve.
GAN is faster under certain situations yes but not really necessary for audio applications.
A GaN-FET power stage provides a precise high-power reproduction of the Class-D PWM signal with extremely high linearity.
Look - they say its a Class-D amp themselfs - and it is of course ;-) - no problems att all with MosFET. NB - the referent to the "PWM signal"
This linearity eliminates the need for ANY feedback, ultimately allowing for the best possible audio quality providing clean, clear middle and high frequencies and a tight, solid reproduction of low frequencies.
The output signal that tracks the incoming signal the best will provide the highest sound quality - this amp will not stand a chance compared to e.g. latest Hypex or Purifi.
GaN-FETs track the complex audio waveforms so much more accurately than MOS-FETs resulting in significantly more transparent and natural sound. The difference is something even a casual listener can hear and appreciate.
This is outright wrong. The output transistors in a Class-D amp dont "track the complex audio waveform", it switches from high to low with a varying time. The variations holds the music signal. Above it was "PWM signal" reproduction - here "audio waveforms" - cant they make up their minds what the GAN really does in their application? The accuracy of "tracking" is measured as frequency response and distortion. Other amps will be superior to this one as it lacks NFB.
The GaN 1 is also designed so that it does NOT even require a digital-to-analog-converter (DAC). The digital audio signal at the input directs the amplifier outputs to drive the speakers.
If one inouts a digital signal and can hear music in the other end - be sure these is a DA or no sound will come out. I think they mean a standard DAC chip - which the amps has not. But a D/A converter function it has. What does "directs the amplifier outputs" really mean? Jumbo mumbo instead of saying that a digital to analog conversion is taking place before the analog class-D switching output stage.
Although DACs have continued to improve over the years, there is no DAC better than NO DAC! This concept is not a new one as similar devices often referred to as "Power DACs" made quite a splash in our industry years ago, but this time around, using the concept with GaN-FETs raises the bar to an entirely new level.
This confuses where the DAC is and where the output transistors are located in the signal flow. These two different functions are separate and independent - so this is just the sales dept. talking rubbish. "No DAC" is the concept of not using a standard of the shelf DAC chip - its disney mena that there are no DAC. Unfortunate naming but there you go... NoDac is not often refer to as Power DAC - wrong. No bars was raised for sure. GAN and NoDac has absolutely nothing to do with each other.
//
What makes the GaN 1 so special?
Is it really - lets see...First and foremost is the GaN-FET amplifier module.
No GaN-FET "module" - just discrete transistors. "Module" sounds professional, mystic and impressive for the common customer...
It has several inherent advantages in a power amplifier that even the best MOS-FET designs simply cannot achieve.
GAN is faster under certain situations yes but not really necessary for audio applications.
A GaN-FET power stage provides a precise high-power reproduction of the Class-D PWM signal with extremely high linearity.
Look - they say its a Class-D amp themselfs - and it is of course ;-) - no problems att all with MosFET. NB - the referent to the "PWM signal"
This linearity eliminates the need for ANY feedback, ultimately allowing for the best possible audio quality providing clean, clear middle and high frequencies and a tight, solid reproduction of low frequencies.
The output signal that tracks the incoming signal the best will provide the highest sound quality - this amp will not stand a chance compared to e.g. latest Hypex or Purifi.
GaN-FETs track the complex audio waveforms so much more accurately than MOS-FETs resulting in significantly more transparent and natural sound. The difference is something even a casual listener can hear and appreciate.
This is outright wrong. The output transistors in a Class-D amp dont "track the complex audio waveform", it switches from high to low with a varying time. The variations holds the music signal. Above it was "PWM signal" reproduction - here "audio waveforms" - cant they make up their minds what the GAN really does in their application? The accuracy of "tracking" is measured as frequency response and distortion. Other amps will be superior to this one as it lacks NFB.
The GaN 1 is also designed so that it does NOT even require a digital-to-analog-converter (DAC). The digital audio signal at the input directs the amplifier outputs to drive the speakers.
If one inouts a digital signal and can hear music in the other end - be sure these is a DA or no sound will come out. I think they mean a standard DAC chip - which the amps has not. But a D/A converter function it has. What does "directs the amplifier outputs" really mean? Jumbo mumbo instead of saying that a digital to analog conversion is taking place before the analog class-D switching output stage.
Although DACs have continued to improve over the years, there is no DAC better than NO DAC! This concept is not a new one as similar devices often referred to as "Power DACs" made quite a splash in our industry years ago, but this time around, using the concept with GaN-FETs raises the bar to an entirely new level.
This confuses where the DAC is and where the output transistors are located in the signal flow. These two different functions are separate and independent - so this is just the sales dept. talking rubbish. "No DAC" is the concept of not using a standard of the shelf DAC chip - its disney mena that there are no DAC. Unfortunate naming but there you go... NoDac is not often refer to as Power DAC - wrong. No bars was raised for sure. GAN and NoDac has absolutely nothing to do with each other.
//
I am somewhat confused about all this discussion about triange wave generation. Self-oscillating class-D, which accounts for a good chunk of the state of the art amplifiers in this class do not use this.
This has been an interesting conversation for me. I've heard things said here and repeated other places. For instance, PWM is inherently an analog signal because if you play it through a speaker without filtering you'll still hear the music, and that's apparently a digital no-no. To be truly digital it appears people want the signal abstracted in such a way that it cannot just be listened to directly. Now consider a PCM audio file that goes through DSP to get converted into a PWM file. I guess this would be considered a 1 bit audio file? In any case, it seems to me that if you stream this digital signal as a 1 bit signal at the right speed, you did the digital to analog conversion whether you intended to or not! So this gives me the impression that by the definitions I'm learning on this thread, streaming digital audio signals cannot be 1 bit! They have to have a longer word length, otherwise the signal can just be plugged into an analog amp and you'll hear something recognizable. Unless there's some way to do 1 bit encoding that's not PWM or PDM. Am I right here?
Actually self-oscillating amplifiers still work with a triangle wave.
It's only being created by the self-oscillation itself, not by a separate modulator.
This can also be measured btw.
Which has some advantages, especially in/near clipping behavior (as well as a few disadvantages).
But the creation of a PWM signal is inherently made by comparing a triangle wave to a sine wave.
Even if you would make a modulator from a clocked flip-flop you will find this triangle wave back.
Just before it goes into the integrator and comparator.
Pre-filter Class-D amplifiers work super similar as a delta-sigma modulator.
Last edited:
Different perhaps but "better" is purely subjective. An amplifier has one function: to make a small input signal larger. Those that do it with the least added distortion are the "best" amplifiers. If you prefer the sound of added distortion, that's your perspective, just don't try to sell this outdated design bathed in a snake oil marketing campaign as any sort of competition for state of the art amplification.
I don't agree with that, there are many other reasons to claim an amplifier "the best" besides just distortion.
That is an assumption.
There is a certain point like visual resolution (televisions) that our hearing is incapable anymore to notice any differences.
At that point even "the best" in sense of just raw distortion goes out the window, since there is no "best" anymore compared to another amplifier with less distortion performance.
At that point there is no difference.
What else is an amplifier tasked to do besides take a small signal and make it larger? What are these many other "reasons"? I am not talking about subjective performance, I am addressing objective criteria- you know, the things that can be measured and compared. Whether or not the distortion is audible matters not- clearly an amp with 0.000004 distortion is performing its function objectively better than one with 0.004 distortion, audible or not. An amplifier with distortion in the noise floor of the measuring equipment doesn't have a "sound", thus if one likes the "sound" of an amplifier, it would seem what they actually like is the distortion profile.
Every stage that's not needed is just a distortion/coloration addition.
Just answered your own statment there.
These Peachtree amps the GaN-1 and the Carina-GaN have 1 x less stage inside them (analog input buffer) that other Class-D have. And also need 2 x less stages coming from the digital source to them (d/a conversion, and analog output buffer) they just need the digital output from that source.. So there are 3 x stages eliminated when using them, not to mention the zero global feedback.
Cheers George
Last edited:
That's the ultimate question. What's the final result coming out of the amp? It seems better in theory to reduce the number of conversion steps, but if it means a more troublesome type of conversion step has to be employed then it might not be a net gain. I had a couple of the old Panasonic receivers that used the TI chips, I believe. I was super excited about the idea of "pure digital", which I fully understood as a switching amplifier with switching power supply and a digital PCM input that got converted to PWM. I wasn't so concerned with strict terminology as I've encountered here. I noticed they sounded thin and bright, weak in the bass, probably due to lack of feedback. I didn't hear any next level magic.
Ok so these GaN fets are not all they are made out to be...Here is another simple question for all of you math surgeons: after suffering through Mr Putzeys vid, I checked out the Purify amp specs and found that the gain is only about 13db is that possible or I misread? The Hypex has buffered/unbuffered or selectable gain up to ~26 db. What kind of source would you need to make 420W/8ohm with the purify with such low gain? I was thinking the purifi was a cheaper option but then I noticed they don't have an SMPS and the product doesn't seem as mature as the Nilai. A couple extra bucks on the Nilai maybe well spent.
- Status
- Not open for further replies.