What are the best GaN Amplifiers available today?

@koh_i_noor We believe its the filter caps in the power supply forming up to their operating voltage. We thought that since our amp runs a lot of feedback and is thus able to reject a lot of power supply issues that break-in would not happen. But very consistent customer reports indicate otherwise.

So, you are an expert on damping? You spent hundreds of hours with various damping materials and used them everywhere and you know which damping materials are linear and which are not and where to use it and where not to use it?

🙄

We've had optional damping packages available for our products for decades. We have lots of feedback and experience with a variety of damping systems (footers, stands and platforms) as well as extensional and constrained layer damping materials. I've consulted with engineers at firms that make damping materials and damping equipment, some of whom are also customers. Any of them, looking at your post to which I responded earlier, would simply say you don't know what you're talking about.

Some gear is far more sensitive to vibration than others; I've yet to run into a situation where such equipment received too much damping. OTOH I've seen poor or incompetent attempts at damping that has exacerbated problems in the equipment. That is why the myth of 'overdamping' is around. The effect of good damping is easy to hear and  measure.

DO NOT OVERDAMP!

It is impossible to overdamp. If adding too much damping seems to mess things up its because the damping device or material really isn't that good at its job so its adding a coloration of its own.

FWIW I've found our class D amp to be quite resistant to effects from vibration. Damping seems to have no effect whatsoever. If you think about it this is not at all surprising. Damping isn't going to affect a device that is off nor one that is fully on (saturated). It might affect a device in its 'linear' region, but GANFETs and the devices driving them spend surprisingly little time in that region- they are very fast.

Using Gallium Nitride FETs. Is this the same as GaN?

@mjcmt Yes.

How do these various GaN amplifiers perform at low volume (ideally, when paired with speakers known for low volume performance, e.g. horns, BBC monitors, etc).

I can't think of a reason why they wouldn't work just fine. For starters class D amps tend to be low noise. I run mine on speakers that are 98dB and they work fine at low levels.

I found out later that Class D amps and linear power supplies don’t really go hand in hand, for reasons I forget why.

@blisshifi 

There isn't a reason. The kind of power supply makes no difference if its properly designed. We use a linear supply in our amp.

However, the majority are still rather sterile and lacking in dynamics.

I am wondering what is meant by 'dynamics'. IME, most of the time when audiophiles talk about 'dynamics' they are really talking about higher ordered harmonic distortion. Dynamic contrast should come from the recording, not anything else.

In the 1980's all the best measuring equipment was supposed to sound the best and now we know that was not true, maybe in 20 years all the current best measuring equipment will not be considered the best sounding equipment.

​​​​​​@invalid FWIW Dept.: you're talking about measurements made 35-40 years ago or so. Just as any other field of technology, the measurement world has vastly improved. IME sometime in the 1990s, we crossed a threshold where we could finally measure a lot of stuff we were hearing that we could not measure prior. But many audiophiles are still making decisions based on how things were back before that!Imagine trying to surf the web on an Apple 2 of an IBM PC...

The problem we have now is almost as bad: most people including those that make the measurements, fail to understand their significance WRT how the ear/brain system perceives sound.

I’ve determined that the problem with forum anecdotes with regard to amps is that many are lacking a good point of reference. Someone will chime in saying that an amp like the Starkrimson powered their Magnepans or ESLs well, yet this person probably wouldn’t say that had they compared the amp to a big Krell on the same day. That or they will be comparing these amps to a mid-fi offering from a competitor.

Suffice to say I suspect GaN FETs with regard to amps are mostly another flash in the pan technology that hasn’t really moved the needle. There may be exceptions of course, but I am highly doubtful at present.

@helomech Reference is a big deal IMO/IME. I use LPs and CDs that I recorded and mastered myself. It helps having been at the recording session. 

@mapman GANFETs have been around since 2009. They became much more commonplace about 2016; at that point most of the major US players had their own version of the devices and they became a lot less expensive. 

@yyzsantabarbara +1

There are tons and tons of factors in a power cord that effect sound quality.  Resistance is one of hundreds....Bandwidth?  You need wide bandwidth to carry 60 cycles?    Actually, you do....and you don't......long story.  What about all the other factors?  You don't go there, eh?

Actually I have. Bandwidth is required to pass current in the short period of time at the peak of the AC waveform when the power supply filter caps can charge. If the bandwidth is lacking in the cable the caps will not charge properly. I have explained this elsewhere.

I never said you told people to buy 4 gauge power cords.......it is just what you imply.

Ralph would have us all buy 4 guage power cords

😂 When someone is willing to contradict themselves in this manner, its not about fact. Its about making the other person wrong.

It is time to stop with the trolling.

I am listening to a long playlist of Sinead O'Connor music as I type this, and it is goosebumps. I am sure it is 99% her voice but the gear does add to the beauty of hear voice.

This is a comment that I often see made by people that purport SETs. We might have a difference of opinion here; mine is that the equipment should not add or subtract WRT to the beauty of the recording. In this way, the more neutral you can get the equipment to be, the more it can bring out nuance in the recording without editorial. 

Ralph would have us all buy 4 guage power cords and use whatever for line level (Mogami is fine, with him....a friend of mine started with Mogami and thought is was good...then he tried two others that BLEW away the Mogami).  He has said over and over that balanced cables make no difference in sound.  Even his customers hear otherwise.  You have to listen to know....you have to listen to know.....and listen with an open mind.....an open mind.

This is a misrepresentation; a strawman and is false by definition. I did not made the statements implied in this post nor have I received the feedback he implies.

I've explained that voltage drop is part of why power cords sound different elsewhere on this site. I've also mentioned that there is another variable, which has to do with bandwidth.

WRT balanced lines, I've pointed out many times that if the equipment using it does not support the balanced standard (very common in the 'high end audio' world), differences in cables will be easily heard since the shield of the cable and the dielectric (insulation) have become part of the 'sound' of that cable. The Golden Age of Stereo recordings (1958-1960) demonstrates that the balanced line system works quite well, unrivaled by single-ended connections.

Its one thing to have an open mind; quite another to ignore fact in favor of made up stories.

Those with Ralph's amp can do a quick test.....just remove the covers and remove all the hardware on the toroidal transformer.....lift it off the chassis and place a quarter inch thick piece of wood underneath.....put the transformer back down and listen.  This CANNOT be measured.....but it can be heard....by YOU!

Actually this one is easily measured (you don't even need test equipment, you can do it with your hand for Pete's sake) and it is audible for a simple reason.

All toroidal transformers are supposed to have a very compact radiated magnetic field, but in practice they are a bit sloppy. Transformer manufacturers supply a mounting bolt with the transformer, but that bolt is always been a regular steel bolt and is magnetic. If you use it to install the transformer, it will be a magnetic short to the transformer. So you'll find that it heats up more than the transformer itself (you can feel this with your hand... you can see how laughable the idea is that this can't be measured 🤣). So its easy to show the difference in temperature and output voltage of the transformer when the magnetic steel bolt is replaced by a nonmagnetic stainless part.

We've been doing that for at least 30 years.

Fun fact: the bolts holding EI core (conventional) transformers together are also magnetic. They are usually insulated from the metal of the transformer and this does mitigate a lot of the magnetic load they place on the transformer core. But not all of it. You can reduce the temperature of a conventional power transformer by using non-magnetic stainless bolts (don't forget to keep the insulators!) and you can increase output power in tube amps slightly by changing out these bolts in output transformers for the same reason.

You can see here that someone claiming this can't be measured hasn't even bothered!

The issue with measurement is often figuring out what to measure, and then sorting how. For example, if you want to measure the effect a fuse or power cord has on an amplifier, you don't measure that fuse or power cord- you measure the effect on the amplifier.

Similarly if you think that a certain tweak has improved the bass in your system you demonstrate this by doing a frequency sweep in the room, install the tweak and measure it again. A customer of ours used this technique to show that a certain filter capacitor he had installed in a power supply was not only improving the bass but also reducing distortion. He was able to show a correlation between the perceived improvement in bass and clarity of the system and his part replacement in this fashion; that the perceived benefit was more than just expectation bias.

I can go on with other examples but you get the point. If you hear a difference, and especially if others report the same difference, then if you think about it you can probably figure out why that's happening and measure it. But you do have to think...

I remember very clearly the first time I heard a power cord make a difference- I was at CES in 1990 and heard a set of Magnaplanar MG3s that had been modified with a wood frame made by George Cardas. It was an impressive frame and did wonders for the speakers. I asked him if I could play our MA-1s on them after hours, he agreed and we made it happen. The amps seems to have a prodigious ploddy bass on the speakers; George suggested a pair of his power cords, which we installed. The bass was instantly improved! Switching back and forth the difference in the power cords was not subtle. I bought the pair of power cords on the spot.

It really bothered me that I could hear this but there was no clear cut explanation why, other than the mysterious claims that so many cable producers make. Turns out it was something simple: voltage drop across the power cord, easily measured. I sorted this out by measuring the output power, output impedance and distortion of the amp, which led me to measuring the input AC voltage at the IEC connector. Voila!

Power cords are one of those tweaky things that a lot of people over on ASR will deny has any effect on a system's performance. Power cords are subject to Ohm's Law like anything else. When I see that kind of denial, I like to ask if they caused their hand to move and make the measurements they laud so often, rather than just blindly making a pronouncement. If there's no measurement they are no better than the subjectivists they denigrate.

Generalized statements are painful. While Ralph may prioritize noise elimination, is it true that he believes it is the only thing that matters?

@blisshifi Of course not! @ricevs apparently has a lot of misconceptions and is putting words in my mouth, making claims that I never made. My position about distortion has been clear for a long time. I've been building class A triode zero feedback OTLs for over 45 years and in amps like that everything you do makes a difference because the amps lack feedback. This is the same as in SETs, of which most also lack feedback.

Just because something was knowledge 90 years ago, like 2+2=4, does not mean its not true today. What is different now, which simply had to be accepted back then, is we know why our ears respond to distortion the way they do. In his comments, Ric conveys that he has not obtained this as part of his knowledge base. He's got a lot of company in that regard- many solid state amplifier designers don't really care what sort of harmonic spectra their amps make.

But some do and not surprisingly their amps get more regard in the audiophile community where people 'follow their ears'. For one to push exactly that sort of philosophy, which I've no problem with at all, since that very thing has kept me in business, that the reason why is being described as poppycock is really a bit astonishing and ironic. You'd think this would be of paramount importance to one who has built his business model 'following one's ears'!

Obviously not all amps are the same. Those with very high feedback amounts will respond less to things like fuseholders, IEC connectors and the like because they have the ability to reject that which is not like the signal. But amplifiers with little or no feedback are very prone to these influences.

Regardless of how much luck one may have had doing mods, class D amplifiers are quite sensitive to layout problems, stray capacitance and parasitic inductance that other amplifier technologies simply are not. It may be that you don't hear any audible artifact from a modified amplifier, but without testing it you really don't know what's actually going on. That's a fact and no amount of remonstration on Ric's part will change that. Integrity requires that sort of testing for noise be conducted, if one is being paid to make changes.

The simple fact is that audio would not exist as a hobby if engineering did not exist. Nothing that we do in the audio world is magic- there is an engineering correlation to everything we hear. But you have to know how the human hearing perceptual rules work to be able to sort out some of the whys.

I find it amusing that someone is trying to denigrate me by lumping me in with the measurement camp. I'm not part of that; what I've found is that correlation I mentioned. Both the subjectivist camp and the measurement camp hate the idea that you can draw a direct line between what we hear in audio circuits and what we can measure in them. Many audiophiles still live their lives according to the myth that we can hear things we can't measure (which was true in the 1980s) as if somehow measurement technology had not marched into the present the same way that every other tech has improved over the last 35 years! Imagine trying to surf the web on an Apple 2 😁

Folks, just so you know, if you are going to be successful with a class D product, one thing it has to do is meet EU Directives for radiation, so it can obtain the CE mark for use in Europe. In the US, you have to meet FCC part 15.

But its more than that. If the class D circuit makes noise, which will be in the form of RF energy due to parasitics in the design, this noise can leak into the AC line as well as being radiated to other components in the system. Both digital and analog gear can be quite sensitive to this sort of noise. It might blot out certain FM stations on your tuner and generate noise on the AM dial. GANFETs in particular are insanely fast and can switch at some amazing speeds. Just before release of our class D we were concerned about the devices running warmer than we thought they ought to- we found that there was a tiny bit of noise coming from the comparator chip, causing the outputs to switch at 60MHz!

When you change out parts in the output filter or elsewhere in the design, the change can result in parasitic noise due to extra inductance that is introduced. An example is the lead inside a capacitor, which can and is a different length depending on the part used.

For this reason the product needs to be tested for RF and AC line radiation. Failing to do this means that the probability of noise introduced through modification is very high!

Any class D designer will tell you this.

They will also tell you that any such modifications voids the warranty.

@ricevs 's comment about distortion is simply false and this has been known for 90 years in the audio world 😀 (ref.: see Radiotron Designer's Handbook. 3rd edition). Perhaps he could get out there and show us all how its done by designing and building his own design from the ground up.

Maybe since you’re now on this thread, could I ask you to share how you voiced your GaN monos, especially in comparison to your OTLs? Did you attempt to preserve the same level of openness and bandwidth to your other amps (albeit making it easier to drive speakers with)? How would you compare the sonics of your GaN amps to your OTLs?

@blisshifi We did not 'voice' either our OTLs or the class D. Most of the work on the amps during R&D was to eliminate noise. We lucked out in that once we had that bit sorted, the things that cause distortion in our circuit tended to cause lower ordered harmonics, much like a tube amp, but at a lower level. I prefer the class Ds to our OTLs, once they have broken in. They are more neutral but easy to listen to; when comparing side by side the big difference most people comment on is the class D seems a bit easier to hear into the rear of the sound stage; they are more 'focused' if that makes sense. 

Which AGD amps did you compare to the Atma - were they the Audion? Would you characterize them as rolled off on top, or just sweet?

@blisshifi If you look at the comments comparing the Atma-Sphere amps to the AGD, you'll see that they contradict each other- some say one is sweeter or darker and the other the other 'way around. FWIW the Atma-Sphere has bandwidth well past 30KHz, so of someone says it sounds 'darker' or the like, its not because of frequency response- its due to how the amps in comparison make distortion.