Atma-Sphere Class D… Amazing

@fredag 100W with 8Ω, 200 with 4Ω. Current limit protection will kick in with lower impedance loads- as long as not to full power the amp will double its output power, but will reach its protection limit at about 250 Watts.

We have several reports now that leaving the amps on helps them sound better.

@hilde45 This has apparently been answered by others, but customer feedback suggests the break-in time is real and is about 3 days or so. The amp does not seem to have a 'warmup' time though.

Any plans for a higher power mono amps in the future?

Yes.

3.  Amp designers would state their objective… straight wire with gain….low distortion with 2nd harmonic dominant….low distortion with 3rd harmonic dominant …etc….. I think Ralph has said this describes his amp  

A 'straight wire with gain' is better described by our tube amps since they don't double power as the load impedance is halved, which is in keeping with how a 'straight wire with gain' would behave :)

Our goal was as low distortion as possible. What distortion we get tends to the the 2nd and 3rd.

follow up question regarding a higher powered class d amp...would this be of much benefit to people running speakers with 90db+ sensitivity and listening below 85db?

Unless the room is enormous, probably not. In the average room that would work quite well. In most rooms at 85dB and speakers like that the amp would be loafing.

So, one thing that would make it sound more open sounding is to remove the steel plate over the transformer and removed the steel bolt.

I’ll let you in on a secret. Use Non-magnetic stainless bolts and then you don’t have to go thru all that. A regular steel bolt heats up because its a magnetic resistance to the transformer and typically runs hotter than the transformer itself! Change it out to stainless and the transformer draws less power and runs cooler- and may well put out a slightly higher voltage. But FWIW due to the high amount of feedback, the class D has a lot of power supply noise (and voltage sag) rejection. Class D amps running less feedback will benefit more on this account.

The Cardas binding posts are good but even better is to get rid of the posts, put slightly longer wire on the output and clamp that wire directly to your speaker cables using nylon bolts, washers and nuts sitting right outside the binding post holes

Anyone doing this is really asking for trouble! To start with any manufacturer would void the warranty in a heartbeat. It also doesn’t go down well unless you live in a mancave wearing cargo shorts and no hope of a date :) ... if you get my drift

Of course, you would want to put a Furutech super IEC inlet on the amp and get rid of the stock inlet. Get a fuse holder from Acme and wire it off the Furutech inlet and get one of those new yeller fuses from Quantum Science......

This is a really Bad Idea. In our case there’s an AC filter on the backside of the IEC connector (the amp in the video is from the first production prototype run). If this isn’t present, there could be noise issues. In our amp, the rectifiers in the power supply are the main source of noise and we did a lot to snub them, but the noise filter was added to really make it silent. In this way it puts out less noise on the AC line than most tube amps. You could change out the fuse but I think you’ll find that due to the large amount of feedback, the minuscule voltage drop across the fuse will have no effect whatsoever.

How about changing the op amp on the input to a discrete one? Discrete op amps are generally more real and open sounding.

Here’s another thing I’ll let you in on. What is important in opamps is something called Gain Bandwidth Product. If you have enough to support the feedback in the circuit, there will be no ’sound’ of the opamps other than noise (and in our case the noise floor of the amp is mostly from the opamps). Back in the 1960s and well into the 1980s opamps did have a ’sound’ which is why if you have to change out an opamp in a vintage guitar effects pedal, its important to find an exact replacement or the pedal won’t sound right (guitar players are quite picky about that sort of thing!). Our input buffer has a gain of 2, which is to say there’s a lot of feedback and with modern opamps, as long as you don’t ask more than about 20dB or so, they will be as neutral as you can get.

A friend of mine designed one of the higher performance discreet opamps out there (back when it was still worth it) and he’ll be one of the first to tell you what I did above. IOW doing something like this will result in no benefit whatsoever. To design a class D amp you can’t just wing it by plopping parts in; you have to do the math and understand the engineering behind it.

 

 

 

 

 

The thousands of audiophiles all over the world that listen to wires, fuses, jacks, etc......are LIVING PROOF that these things are real. What proof (listening tests) do you have that what they all say is wrong? Please share.

Ok. In our zero feedback tube amps we make them with custom wire, custom resistors and the like because we hear a difference (and not surprisingly, the difference is also measurable if you know what to measure). In any amplifier with really high feedback and the Gain Bandwidth product to support that feedback at all audio frequencies its a different matter! This is because the feedback of the amplifier allows it to reject a lot of the influences of the materials themselves. So in our tube amps you can hear an exotic fuse due to the voltage drop; in our class D amps you can’t, even though they are more revealing.

BTW, Nord, Apollon, VTV, etc. sell amps that have discrete input buffer option....so you can roll op amps......all op amp options sound different from each other.....this buffer is on the input of a class D Purifi module......it uses very little gain.....just like Ralph......you can also put adapter boards in these discrete op amp input boards with regular op amps on them.....yes, they sound different again....and not as good as the discrete ones.....go read their sites..

We’re very aware of how different the experience of various class D amplifiers actually is! One difficulty any class D amp has is how to drive the input comparitor which has a low input impedance and must not have any offsets of any kind at its input (and is otherwise operated well within its common mode rejection range).

So the input buffer design has an enormous effect on the performance of the amplifier. This really suggests to me that there are a lot of buffer designs that really aren’t up to snuff- and are using opamps in a way that causes the opamps to have a ’sound’. If they were designed correctly, you wouldn’t be seeing all this noise on the ’net on the ’sound’ of these amps. In short I am saying that it appears that some of the designers that made these buffers don’t know what they are doing and thus get highly variable results. If they used the opamps correctly they would get two things: less effect from the opamps themselves and a more neutral (and thus more musical) amplifier.

Getting back to Ralph’s amps, can they be stacked since they run very cool?

@lula As long as they are not being pushed hard, yes. We run them stacked here in the shop all the time.

 

Unfortunately they are a bit too expensive for me, will there be a cheaper stereo version ?

I sure hope so! I want maybe 50W/Channel stereo integrated.

The trick here is understanding that the measurements, if the correct ones are actually made, affect what you hear in a very direct way. There used to be a disconnect in this regard because we simply didn't have the technology to show things like the distortion spectra with any detail. But that's been around for a while and if you understand how the ear/brain system perceives sound you can predict what an amp will sound like by looking at the measurements. Most audiophiles are old enough in this market to remember when that wasn't true. But you're not going to get anywhere if you live in the past 😉

 

Testing has indicated correlations between certain distortion spectra and "averaged" or group perceptions but there has been no perfect correlation established between measurements and human response on an individual basis because at the end of the day, you just can’t account for taste.

@kuribo

There are the rules of human hearing perception (such as Fletcher Munson or the masking principle) and there is taste. They are two very different things and the two get conflated quite often! All humans use the same hearing perceptual rules- otherwise audio as an art would be impossible. What people do with that is different, which is why there is disco, rap, classical, tone controls and the like. Human physiology responds the same way to distortion, so it is something that is predictable and reliable within certain limits (there is individual variance on how the 3rd is perceived, depending on its phase for example).

I wasn’t aware that class D amplification was doable without the utilization of negative feedback

@charles1dad It is. Our first prototypes were all zero feedback and demonstrated to us that the idea of class D was worth pursuit. You also don’t need opamps- again we drove early prototypes directly using our preamps, which have no trouble driving lower impedances like 2000 Ohms. These prototypes didn’t use any opamps.

You can’t seem to get beyond the simple truth that your experienced subjective reality and beliefs are nothing but opinions and are no more valid than anyone else’s. The guru complex.

+1 AGS, Audiophile Guru Syndrome, is an ugly thing :)

His high feedback Op-amps amplifier versus a near polar /opposite approach is something that I find compelling with regard to respective sonic presentation. All that concerns me with audio products is how do they sound? That’s why we buy them.

The reason for both approaches is the same. IMO/IME the amp must not have any increase in distortion as the test frequency is increased. Most amps using feedback suffer this problem- and they sound brighter and harsher than they should partially on that account. In a tube amp, getting it to run enough feedback to get around this problem is impossible regardless of the topology. This is because they lack the Gain Bandwidth Product to support the feedback at higher frequencies- as you increase frequency, the amp requires a lot of GBP; if its not there the feedback will fall off, causing distortion to rise. We avoided that in our OTLs by running them zero feedback.

This is a problem in solid state too- and is part of why traditional solid state amps are known to sound bright and harsh (especially at higher volume). Class D offers a way around this because you can get the loop gain you need to really support a high GBP value- and thus also support high feedback at all audio frequencies.

The reason feedback can be problematic is that the feedback node always has a non-linearity associated with it. This might be the base of a transistor, or the cathode of a tube; whatever it is means the feedback signal is distorted by that non-linearity and so when mixed with the incoming signal doesn’t quite do what its supposed to do (one effect of this is additional distortion is created...). But if you run enough feedback you start to get around this problem. That needs to be at a minimum 30dB and must be 30dB at all audio frequencies. We’re at about 37dB.

Probably more information than you were expecting, and actually IMO this explanation is really the nutshell version so those of you technically-minded I am aware this explanation is incomplete.

If the desire is for more emotion ,soul, humanity, breathe of life type of sound, a high quality SET or OTL is very difficult to match. It just always depends on what you want. Other topologies will excel in other specific sonic parameters/areas.

I think one of the issues you run into when making this comparison is that the zero feedback tube amps (whether SET or OTL) will have a frequency response variation depending on the load impedance (the speaker). I’ve found that anyone using such equipment, including myself, has made accommodations for that issue if they have spent any time trying to make their system sound neutral and musical at the same time. In my case this is mostly to do with the level settings for the drivers, found on the rear of my speakers. I use pink noise to set them up correctly. Whatever those accommodations are though, you have to back them out of your system if you really want to do a proper comparison, which is probably tricky. This is just a personal observation, but once I corrected my system for the voltage response of the class D I found it every bit as involving.

I would like to hear the designer’s take on how his class d design differs from those of Hypex and Purifi and at 3 to 4 times the price, what they offer over the Hypex and Purifi products that makes them worth the considerable difference. I would like to know why he decided to design his own class d modules, especially in light of the fact that it is no trivial matter and a completely different task than designing a tube amp....I would like to know why the designer choose GaNfets over standard fets and what advantages he believes they offer at his operating frequency over standard fets....

We chose to design our own module because it sends the wrong message to the marketplace by using someone else’s- it suggests that maybe you don’t don’t have the engineering talent in-house. Plus we can make it the way we want to. FWIW our modules seem to be lower noise than Bruno’s.

The reason we’ve not used SMPSs yet is we found that if you really want the amp to perform properly, especially at high volume, the SMPS really needs to be designed specifically for the application. Most of our prototypes ran SMPSs and we ran into this limitation quite frequently.

A good portion of our cost is the chassis, which is custom-built and designed to look decent, not man-cave and also be durable in shipment. It seems its a bit over-built! For the last 40 years we’ve gotten dinged on cosmetics a lot; you put in the cosmetics and then you get dinged on price...

GaNFETs are a little faster, but the main reason for using them is to create a lower noise layout due to less strays and lower drive requirements which are for the most part an order of magnitude lower than MOSFETs. In this regard the noise our amps put out on the AC line is lower than many tube amps and most of that comes from the power rectifiers rather than anything to do with the module.

 

 

 

 

 

Would you mind reiterating your past comment of how Class D (yours and perhaps others) have finally made the leap to outperform tubes (yours and others)?

I find the class D to be more transparent in my system compared to my tube amps; less coloration also due to lower distortion. At the same time I don't find them harsh, a classic sin of solid state amps in general. I can play them all day and not get tired of them.

Now let's say you are a tube amp producer. You face a variety of problems. Right now a major one is availability. If you're not going to get in trouble with the law (due to sanctions against Russia), you're probably using Chinese or JJ tubes if you're using new tubes. Another one is the industry is tending towards 4 Ohms as the default speaker impedance rather than 8; I see this as a bad thing since all amps make greater distortion into lower impedances, and with many its enough to be heard (audiophiles use 'fat' 'muddy' and other similar expressions for this). But 4 Ohms is a thing regardless of what I think.

Its now possible to build a solid state amp that is as relaxed as a good tube amp, not lacking detail or depth (and maybe more). Since brightness and harshness of solid state is what has kept tube amp producers in business for the last 60 years, and also because tube amplifier power is more expensive, how long will tubes be a viable option in audio? IMO while it will still be a while (owing largely to highly variable results designers got out of class D over the last 20 years having poisoned the well at first) tubes will continue for a while, but even chip-based class D amps that might only be $75.00 are now giving serious tube amps a run for the money!

If I were an amplifier manufacturer that had not got the class D thing figured out, right now I would be worried. The market will be shrinking for tube power products- even in the guitar world. I know a lot of cheap class D guitar amps are so much junk but they aren't all that way, and most guitarists these days rely on their effect pedals for their 'sound'. The guitar market is a lot larger than the high end audio market for tube use- they drive a lot of what is available for high end as a result. If you lose the guitar market, that will be very nearly the end of tubes.

 

 

While these rules of thumb are helpful perhaps to people such as yourself designing amplifiers aimed at as broad a segment of the market as possible, they are not "laws" or universally applicable "truths" like the law of conservation of energy or the laws of thermodynamics. As a result, they can not be relied upon on to be an accurate predictor on an individual basis. Thus, the guy shelling out $5K for an amp would be best advised to trust his own perceptions and tastes and to listen to the amp in his system, in his room, to form his opinions rather than rely on the perceptions, tastes, and opinions of others.

Fletcher Munson is a bit more variable, but the masking principle isn't. Its what made MP3s possible. Also, how humans sense sound pressure does not change from individual to individual; generally, the higher ordered harmonics are used. This is really easy to demonstrate using simple test equipment.  Imagine a world where every individual used entirely different hearing perceptual rules! It would be a good basis for a scifi novel 😉

The patent involves the use of a Circlotron to reduce deadtime. If you are using GaNFETs, the inductive kick of the output filter coil is what really turns the device off (assuming the gate is already off); to allow that to happen a certain amount of deadtime has to exist. As it is there is still less deadtime in a GaNFET design as opposed to a MOSFET design.

Too many don't understand the difference between their subjective opinions and objective fact.

Being objective is a worthy struggle despite it being impossible.

FWIW we've done lots of comparisons and we have a lot of feedback from the field at this point- from a variety of customers that know our prior work. The feedback is surprisingly consistent, in the face of not knowing about any other feedback we've received.

 

 

@wolf_garcia The smaller guitar amps will hang around for a while because they are low enough power that its practical to overdrive them. But the bigger stuff like 100 Watt Marshals are likely threatened, if someone can put a class D amp out there that's worth listening to. The weight is a big variable here. 

 

 

And then we come to digital amps as a category. They are insanely cheap to manufacture as has been pointed out elsewhere in this thread. To my mind, a digital amp is a switching power supply built to drive loudspeakers. If there is a simpler way to harness power from the wall directly to the pre-recorded sound signal it has not yet been invented. Ralph has acknowledged that the box-a plain vanilla box at that-is the major parts cost of this product. When the best switching power supplies sound better than the best linear power supplies I will likewise believe that the best digital amps sound better than the best conventional Class A/Class A/B ss and tubed amps. 

@fsonicsmith Let's clear something up. Class D isn't digital. It's a coincidence that the D and digital use the same letter. Its called class D because at the time (late 1950s) class A, B and C were already taken. Its an analog process.

It behaves a bit differently from a switching power supply. For one thing the switching frequency is a lot higher- often by an order of magnitude. Another thing to understand is that switching power supplies are often used in class D amps but class D amps can run off of conventional power supplies too. But even in SMPSs the noise floor is much lower now than it was 30 years ago. Like anything else people sort this stuff out over time. FWIW our class D is so quiet that most tube amps inject more noise on the AC line.  

@kuribo You do know we have published specs right?

So yes, Class D is more accurately referred to as a "switching amplifier" rather than a "digital amplifier" but things get blurred when binary quantization is involved as binary quantization is at the heart of digital too. 

@fsonicsmith The on and off states of a digital word have meaning- IOW the arrangement of the bits create a word which represents a voltage. In a class D amplifier the on and off states have no such meaning. When the term 'digital' is used, its assumed to not be analog. Despite being a switching technology, a class D amplifier's functionality is entirely analog. 

There's no intention to 'cut you down' or anything like that! Whenever I see anyone use the term 'digital amplifier' while referring to a class D amp, I feel its important to clarify since this is such a common misconception (or misuse of terms). If you do a search, you'll find I've done this on this site dozens of times over the last several years; please don't take it personally.

@kuribo The cost has to do with how the equipment is marketed. We use a dealer network and have worldwide distribution. That requires that we have it priced accordingly. Dealers can be quite valuable because they can provide support!! FWIW if we were to use a Purifi or Hypex module, the amp would actually be more expensive than it is now due to the markup we would have to pay on the module. The cost of the chassis, built extra rugged so it will survive abuse in UPS while being built in limited quantities, is a big reason the amp is more expensive. The funny thing here for me is for decades we were always docked on cosmetics. I was always resisting installing a 1/2" thick front panel on our stuff since it would have raised the cost so much, but that is what our competition has been doing all this time. And we still don't have the 1/2" panels!

So we found a way to limit that cost a bit, but the simple fact is that if you do this stuff in the way that we are (the chassis is formed of 3/16" aluminum for example, to prevent deformation with the weight of the power transformer) its simply more expensive. Now if we were to use a SMPS it would allow the chassis to be cheaper, but our research has shown that if you really want the amp to perform the way it should, the SMPS will be custom built for the job. Otherwise it will current limit or other such nonsense and we certainly experienced that!

So we didn't skimp on the power supply or the chassis. We just built it to proper engineering standards without cutting corners. IMO one of the reasons you see such variable comments about how Hypex and Purifi amps sound is because of how they are executed.

Indeed, those marketing Hypex, Purifi, and several other class d amps using the modules of others can "tune" the sound through the input buffer op amp choices

@kuribo If they can ’tune’ the sound just by changing opamps they are either using really terrible opamps or don’t know what they are doing!

We’ve had the idea of a SMPS meant for the amp for a couple of years. That’s not a trivial design project although its certainly an one than the module itself was! Of course we looked at available SMPSs including those meant for audio, but again we ran into not only performance (which impinges the sound and performance of the module) and protection issues but also cost issues! If we make it ourselves then we solve three problems at the same time. In the meantime the toroidal supply is less expensive, works right and we have the right protection although we need a more expensive chassis to support it. IME though, that chassis helps with the mancave/WAF issue. In my case this means that I don’t have to have a mancave to run my stereo (although I do have to be careful about cargo shorts); it can be in the living room without inciting ire. Ya picks ya poison I suppose.

I should also mention we're concerned about noise. For example Bruno uses a current pump to create lower voltages for opamps and the like. We found that is noisier than using a more expensive supply for the same task. As a result the amp puts less noise on the AC line than most tube amps.

I agree with that whenever SQ can be affected, but making thick chasis just to  protect the amp during the shipping seems to me unreasonable - just pack it well! 

We've always packed our equipment well to survive FedEx/UPS treatment. But experience has shown that even though its packed well, the chassis can be deformed by the mounting bolt for the transformer if the transformer is heavy enough and the chassis thin enough.

These are the claims of customers- they prefer the "sound" of certain op amps over others, not mine!

If that is so either the opamps were really terrible or the designer didn't know what he was doing! With any module you really don't need a lot of gain, so why would decent opamps affect the sound?? Answer: if the buffer is properly designed and if decent opamps are used, they won't.

I’m curious why the Atmasphere Cl D amps are so low powered, compared to many others.   Most D’s seem to claim fairly high power. I guess the switching power supplies are not so constrained by size and weight limitations.  I had a D-Sonic, based on Pascal modules, that produced 800 watts into 8 ohms.  I went back to tubes, though.

@lloydc You just put your finger on the reason why, which is why bother if the amp won't sound like real music 😉 All that power doesn't do any good if the customer is going to go back to tubes (and lower power as a result...).

What are the cases in which measurements don’t matter?

I refer you to Daniel R. von Recklinghausen who was the chief engineer at HH Scott. He debunked, decades ago, the idea that measurements and the subjective experience are not intimately connected. 

@kuribo These two statements, one right after the other, seem contradictory:

I have provided an alternative view: subjective opinions carry no factual information. We must each listen for ourselves to find our own truth.

A lawyer might ask, in a court of law, 'which is it? Were you lying then or are you lying now?' 😁 I'm not suggesting that you're lying; I am suggesting that its impossible to be truly objective.

Subjective opinions can carry quite a bit of factual information. If all those opinions say the same thing and the people producing them are unaware of each other they carry quite a bit of weight. It seems to me that you've still not made the connection of how important measurements are to the subjective experience (and I see that all the time on the subjectivist side as well...). The rules of human hearing are the reason why; as I pointed out earlier all humans use the same perceptual rules. So if we can sort out what's important to the ear, then we can make the measurements to show if we've made progress. The former is the tricky bit!

You are aware no-one is going to beat Bruno's numbers any time soon. For that reason alone I really don't see what you are so concerned about what measurements we get. If that is the only concern just buy a Purifi and be done with it. The problem I see with that I already outlined.

Our goal was to make a class D amp that could allow anyone to enjoy the music and not worry about class D, tubes, class A or any of that stuff. In that regard we feel like we succeeded. A secondary concern is that tube production is waning due to a variety of events having nothing to do with the technology. IMO tube power amps are on borrowed time right now. The rest of it I've already explained.

 

 

 

 

I agree on point 1. Measurements ought to be much more accurate, They are currently so useless and sloppy. Ought to go to to a at least few more decimal places.  We can agree on this.  Or not.

Its a matter of whether the person making the measurements knows what he is doing and whether the equipment needed is available. If you want to know how an amplifier will sound look at these things:

Distortion vs frequency; for best results it will not rise

Distortion spectra at 1 Watt

Distortion spectra at -6dB of full power. This particular measurement is where SETs fall on their collective faces as this is where the higher ordered harmonics show up, causing the amp to sound 'dynamic'.

In the case of the distortion spectra the lower orders must always be of enough amplitude to mask the higher orders regardless of the overall amount of distortion. THD as a measurement usually tells you almost nothing and is a good example of Daniel Von Recklinghausen's famous comment.

I would also point out that the transition from most tube amps to Class D amps is probably more challenging for owners than most. Most speakers that sound good to someone with a tube amp will sound different with most Class D amps.

Most loudspeakers are meant to be driven by a voltage source which is an amplifier that either doubles power as the load impedance is halved or cuts power in half as the impedance is doubled (tubes, if acting as a voltage source, do the latter and this does depend a bit on which tap is used on the output transformer).

A small number of speakers, but significant in high end audio, are meant to be driven by amps that behave more as a power source rather than voltage source (SETs are an example). The most likely embodiment of this is a tube amp that runs little or no feedback. Some of these speakers have level adjustments (meant to allow the speaker to be adjustable to the voltage response of the amplifier) on the back, like my Classic Audio Loudspeakers. If these adjustments are present they will have to be set up correctly but the speaker will accommodate both types of amps. Another example of this is the Sound Lab electrostatic.

But that will never tell you with any real certainty if Random Joe will like your hot sauce. I don't buy based on market research- I try for myself. And I know many people that aren't going to pay $5000 for an amp because a group of people on an audio forum says its s'wonderful! They will want to hear it for themselves in their system, in their space. I also know many people who wouldn't even consider an amp at $5000 without seeing the measurements first.

@kuribo 

Random Joe's taste in food is something quite different from his ability to taste- to detect flavor (and withstand a hotter sauce 😁). If you conflate the senses vs taste in the audio realm then the art of audio becomes impossible! Its not just Fletcher Munson- I chose that one because its so well-known that there is a switch for it (the 'loudness' switch) on many products. But there are other rules- here's one (easily proven by anyone with simple test equipment): the higher ordered harmonics are used by the ear to sense sound pressure. That one is ignored by the industry most of the time (which is IMO why there is so much stuff that is harsh and bright)... The masking principle is what made MP3s possible. While there are minor variations from individual to individual, these rules are shared by all people with hearing and they are far more alike than different! If you doubt that then all audiologists are just so many charlatans 🤣

The fact that you have to try it for yourself is understandable. As audiophiles quite literally because of limitations in measurement and how spec sheets are designed going back 70 years, despite whatever anyone has said about a product be it audiophile, reviewer, dealer or manufacturer, if his lips are moving he's lying. We've all experienced this so often in the audio world that we don't even think about it- we just have to get the product home to find out no matter what the spec sheet says.

We've had the ability for the spec sheets to really show what the equipment is going to sound like for some time. But it appears the industry would go into conniptions if they actually printed the evidence of why their gear sucks so much! (FWIW Bruno's efforts cannot be quantified since he doesn't sell a finished product. You only get to hear what someone has done with it and that isn't always good... so much for the amazing specs...)

So we still have to audition it at home and make up our own minds. Pretty well defeats the purpose of showing all those measurements, even for someone like yourself! By going with 'measurements only' you are open to confirmation bias. If you really want to be objective you have to drop the measurements thing. Sounds oxymoronic but this is the world we live in today...

We still have no way to accurately predict what an outcome will be for a randomly chosen individual listening to reproduced music- will he prefer a tube amp or class d? Again, the fact that there are successful class d amps and tube amps is proof that there is no consensus subjective choice.
 

In time, the tube amps will be gone. Not because of a lack of tubes but because they've been eclipsed and people will wonder why they go through the hassle when subjectively better sound is available at less cost. That hasn't happened yet simply because class D has taken some terrible missteps in the last 20 years.

1. Distortion versus frequency: flat or rising?

2. Frequency response vs load: independent or dependent?

3. FFT vs frequency: level of even, odd, 2nd, 3rd, later harmonics?

4. Distortion + Noise versus power for different loads: increasing or decreasing? Level?

5. Power stage efficiency vs power at different loads: comparison between GaN and regular mosfets

With people making claims of this amp’s superiority, I see no issue with asking for some proof. When the manufacturer states that there is a lot about an amp’s subjective performance that can be gleaned from the measurements, I say please show us.

I have stated that some of the above are important parameters to look at. I've also stated that our amp has some of those properties. That the distortion does not rise with frequency isn't something that needs to be shown; its a simple statement of fact (you could have surmised the truth of this by looking at my comments about feedback, Gain Bandwidth Product and of course distortion vs frequency). You either accept it or you don't- much like you might if you saw a graph on our website- which you might believe or you might not. The bottom line here is trust. You don't trust other's opinions since you don't trust their ears to work the same as yours despite the fact that all ears work the same despite your remonstrations. It is apparent though that you are embrace the subjective experience as anyone else here.

If you'd like a tip to improve credibility here are two:

First, refrain from attacking others even when they attack you.

Second, you would do yourself well to drop one of two stories about yourself, either that measurements are the only thing to look at, or that you have to try it for yourself. You can't take both positions at once since others see a contradiction even though you may not see it.

 

 

 

How can you begin to know what someone will prefer? How can you begin to assume that you know what "better" means to someone else?

I answered that question several times on this thread. You deny its reality and I can't help that. So let's put it a different way- It really helps to keep up on research FWIW...

And yet it’s your product and you are the one making claims about your product and not providing measurements, saying "You either accept it or you don’t- much like you might if you saw a graph on our website- which you might believe or you might not." That sounds to me like "trust me". Asking and expecting me to trust some stranger is even worse.

And it isn’t, as I have said, that I don’t trust their hearing, it’s that I don’t trust their perceptions/taste.

What is a claim I've made? Just so we're clear on this. FWIW I've not asked anyone to 'trust me'. I was pointing out with that comment that such is an issue- its one thing for a manufacturer to put up information on a website. It's quite another when someone else does it. As a result I don't see the point; we already know what the amp does. The question now is really to our customers- do they think that the amps do what we said they would? My experience of the last half century is that if they don't think so they won't be customers.

And yet it’s your product and you are the one making claims about your product and not providing measurements, saying "You either accept it or you don’t- much like you might if you saw a graph on our website- which you might believe or you might not." That sounds to me like "trust me". Asking and expecting me to trust some stranger is even worse.

And it isn’t, as I have said, that I don’t trust their hearing, it’s that I don’t trust their perceptions/taste.

Its not by job to educate although apparently I do that with all the time I spend writing... Your position seems to arise out of you don't accept that the sense's perception and taste are two entirely different things! The former is autonomic and the latter conscious. So far as your writing is concerned, you conflate the two, which is a common mistake.

Not a lot of amps with distortion 120db down that would be said to have a "design issue".

Exactly!- but I said 100dB down IIRC and I do consider that an issue. Most of the numbers guys I run into think maybe -85dB to -90dB is acceptable. I know better, but good luck trying to convince them of that! So I don't bother.

Your reply sidesteps my statement, that is, that we can not accurately predict what an outcome will be for a randomly chosen individual listening to reproduced music- will he prefer a tube amp or class d? 
 

Actually it addressed it directly. In due time 'he' will prefer a class D amp as it will sound better, making the tube amp not worthwhile for all its extra hassle. That is what I said earlier. BTW this phenomena is already occurring. 

You have painted broad strokes and made claims but not delivered on the proof. You are a manufacturer in a hobbyist forum with a self-interest in a product. I believe it reasonable to hold you to a higher standard. When someone is charging 4 times what a market leading product costs, never mind that it has half the output power, I think it unreasonable to just say "trust me".

This statement is false. Its not me you need to trust. Your trust issue is that you aren't aware that others can hear the same things you do- you don't trust their hearing since you think it can't be the same as yours. In the past you've suggested not in so many words that perception and taste are the same thing. At some point you will be disabused when you drop that story.

First of all, many of these amps, with different opamps, have been tested and while all the results I have seen show that the performance of the amp with these various third party buffer/op amps isn't quite at the level of the stock Purifi buffer and opamp, the measurements do not show any differences which would normally be considered audible in any way. In other words, if there were some issues with the design, one would expect significant differences.

IMO this is a classic example of how the industry in general does not recognize some fundamental properties of the ear brain system. Its as if research of the last 40 years is being ignored. The ear is a lot more sensitive than you think it is and so these differences are more significant than you think. Here's what that's about: the ear uses higher ordered harmonics to sense sound pressure. The ear has about a 130dB range which is pretty crazy. This means that distortion product 100dB down is audible if its unmasked. The ear assigns tonality to all forms of distortion (for example, higher orders are assigned 'harsh and bright'). BTW this latter fact of human hearing has been know for well over 90 years and should not be controversial to anyone but the uneducated. 

So what might be considered 'insignificant' if often quite audible. Audiophiles have been dealing with this problem for well over 70 years now; its why we have to take stuff home to hear it for ourselves. This is the inconvenient truth of audio.

 I use measurements to screen out products that don't meet my engineering, design/construction, and objective performance standards. From there, those which make the cut get further consideration and audition. I thus use both objective and subjective criteria. I see no contradiction in using both criteria in the purchase decision.

Well you're not alone in that. Everyone I know of does the exact same thing. The contradiction is in the way you've stated your position in the past. Thus I've revised my assessment; you're mistrust isn't others hearing so much as what meets their 'engineering, design/construction, and objective performance standards'.

 

You may have already covered these issues earlier in this thread, but could you please explain a little more about the following two issues with regard to your Class D amps?

1. I noticed that the switching frequency for your amp is 500 kHz while several other GaNFET amps use a higher switching frequency. Did you decide that 500 kHz represented a "sweet spot" for switching frequency in your design? Does the choice of higher rates have unwanted side effects that can adversely affect sound quality?

2. What factors entered into your choice of a toroidal power supply rather than a SMPS? What number and size of capacitors did you use in the power supply and how did you decide on the amount of storage needed?

The allure of a higher switching speed is increased loop gain which allows for more feedback. The downside is that you need to have deadtime and that value is a constant that does not change with switching frequency. Distortion thus increases with the switching frequency, so I guess in a way there is a sweet spot. The other issue is the faster you switch, the crazier the stray inductances become which can result in parasitics at some pretty high frequencies!

We used a toroid simply because a SMPS really should be custom-built for its application. There are a good number of advantages to SMPSs such as voltage regulation and oddly, lower noise (wasn’t the case 30 years ago!). But you have to be careful about current limiting issues which can really be a thing if the amp is subjected to lower impedance loudspeakers and in particular ones with crazy phase angles. So its easier to simply use a toroid.

And Ralph, you have responded to virtually everyone else’s questions and statements including those of the bizarre troll, and yet you won’t respond to mine asking-earnestly- whether your class D amp has true balanced topology and if not, the sonic compromises if any?

Sorry- that was an oversight on my part. If you don’t know, we’ve been pushing balanced operation longer than anyone else in high end audio; our MP-1 was/is the first balanced line preamp ever made for home use. Its not something we backed away from in the class D!

Who is Ralph to declare that his subjective opinion of the sound of his own product is absolute? Ralph will no doubt diplomatically respond that he is not just referencing his own Class D amp. If so, than tell us Ralph which other Class D amps are you declaring to be subjectively better than the best tube amps other than your own?

:) as Capitan Kirk once said ’Who do I have to be?’

Seriously though we think our class D sounds better than our tube amps although we’ve not done serious comparison with our bigger tube amps. We think our tube amps sound better than most other tube amps so its a simple logic statement at that point. But I really wasn’t commenting on our amp in particular; I was commenting on the fact that class D offers a technique (rather than specific amps, although there are class D amps out there that get very nice comments such as the AGD and Orchard) wherein this is possible, since it is very easy to develop the Gain Bandwidth Product that will allow the designer to support an enormous amount of feedback at all audio frequencies. Prior to the recent evolution of class D this really was extremely difficult until the mid 1990s with solid state and impossible with tubes.

That is why we never ran feedback with our OTLs, since keeping distortion vs frequency linear across the audio band is pretty important if you want the amp to not sound harsh. That’s easy in a tube amp if you don’t run feedback! Our OTLs have as few frequency poles as you can get in a tube amp and even with them we ran into issues with their phase margin. IOW, very difficult to prevent oscillation even with a carefully designed feedback loop if running large amounts of feedback. Conventional amounts were no problem but had all the downsides that have given feedback a bad rap in high end audio.

Class D offers a way around that problem. We started with zero feedback prototypes but found out quickly that a zero feedback class D amp has to have a very stable oscillator for the triangle wave, else you get a high noise floor as the oscillator drifts in frequency. With self-oscillating designs this source of noise is eliminated.

How can you then, in good conscience, even offer your tube amps for sale?

Funny thing about that, people still want them and don’t always believe things that I say, as we’ve seen on this thread. You get used to that over time :) But I can see a day when our tube amps become custom order only or gone altogether.

This man has solved the problems leading to 20 years of "terrible missteps" of all that came before him?

No. Class D has been an evolving tech for quite some time. Clearly others have gone before us and it is they that solved the lion’s share of difficulties, Bruno Putzeys in particular. In a way I think we were lucky in that in our design it worked out that the primary distortion sources tended to produce lower ordered harmonics, so although the overall distortion is much lower, the actual distortion signature otherwise would make you think you’re looking at a tube amp.

how can you possibly tell the thousands of us with tube amps and preamps that if we only gave your Class D amp a good audition we would come to the realization that your Class D amp is subjectively superior?

If you look back you’ll see that I answered that question several times on this thread for kuribo. He seemed unable to accept the answer, as best I can make out because he thinks the autonomic senses and taste are the same thing.

One thing that is a bit of a hurdle, now that you bring it up, is the higher output impedance of zero feedback tube amps in particular. This causes some ’warmth’ the class D won’t have because the tube amp will likely have a slight FR error in the bass associated with the impedance rise with most speakers have in this region. The ear interprets that extra bass energy as a bit of warmth and it causes the perception to tilt slightly away from the highs.

Once you compensate for that (on my speakers I used pink noise to set them up with each amp comparison since the driver levels are independently adjustable), then you see what is really afoot.

You have side-stepped your bold declaration that tube amps, the entire category bar none, are on their way out.

@fsonicsmith That was not my intent. To my understanding I've not side-stepped this issue at all:

IMO power tubes are on life support- even in the guitar industry class D has been making significant inroads in the last couple of years. Its clear that any manufacturer of amplifiers now has to contend with class D technology in a way that they did not a decade ago; in particular manufacturers of tube amplifiers will find their market shrinking dramatically over the next ten years. The reasons for owning tube amps (the 'sound') is being heavily eroded by advances in class D technology; any manufacturer that ignores this does so at their own peril.

Would you please expound upon your statement: "keeping distortion vs frequency linear across the audio band"? I am afraid I don't quite follow.

@atulmajithia In order for an amplifier to exhibit the same smooth character as the music itself, in order to not sound bright, distortion must not rise with frequency. This is one of several characteristics needed to allow the amp to be musical and not bright. If a tube amplifier is zero feedback and has sufficient bandwidth (not including the output transformer if one is present) then this really isn't a problem. Of course the designer needs to pay attention to other issues but distortion vs frequency is pretty important.

The ear is very sensitive to higher ordered harmonics since it uses them to sense sound pressure. It also assigns a tonality to all forms of distortion and higher ordered harmonics get the value of 'harsh and bright'. In addition distortion products occurring in the Fletcher Munson region (3-7KHz) causes distortion in that range to be more easily heard.

TIM is a product of an amplifier having high feedback but also has part of the amplifier outside of the feedback circuit so unable to compensate for certain types of distortion. An example might be the base of a transistor in a differential pair, wherein the input signal is applied to that transistor while the feedback is applied to the base of the transistor's mate. This is a common circuit in many solid state amps over the years. Class D amps  of the self-oscillating variety (like ours) don't have that kind of input circuit and so avoid this problem.

Amplifier design and amplifier performance has been dictated by the limitations of loudspeakers. There is no perfect amplifier because there is no perfect loudspeaker. 

[snip]

Think of an amp as a complimentary solution to a problem elsewhere in the chain, not the source of the problem and impediment to sonic bliss.

@fsonicsmith

Your first statement above isn't correct. Amplifier designers do not limit their designs on account of a limitation in loudspeakers- everyone I know in the business, whether tube or solid state, is trying to wring the most performance out of their product that they can. Amps aren't perfect for their own reasons, for example no-one is going to ever get rid of distortion no matter the speaker.

IME/IMO when you try to work synergies in your system you wind up with more distortion. For example bright amplifiers are often so due to unmasked higher ordered harmonics; they might get paired with dull speakers in an attempt to be tonally neutral. Since the brightness comes from distortion rather than a frequency response error the result is often less than stellar- there will be some musical cuts that will bring out the weakness in this approach sooner or later! Better that each bit in the system stand on its merits rather than its weaknesses.

I had a photo of the inside of the amp that is the subject of this thread and commented that the parts could not justify the price.

FWIW our pricing tends to be less expensive than our competition because we price to a formula rather than what the market will bear. You really have to just actually price everything out. Don't forget to include labor (and all that comes with that); all the work including the board assemblies is done by hand. You can see from the splash page on our website what that is about. A class D amp is a very different thing from a tube amplifier but once you understand the differences (for example, surface mount components) you can see that craftsmanship is still a thing.

What preamps work well with these class D amps?

The gain of the class D and the input impedance (100K balanced or single-ended) are both easy for any preamp made.

Maybe we should ask @atmasphere ? And avoid descending into extension cords, which have nothing to do the power cords on these amps...

Power cords affect audio equipment due to voltage drops occurring at 60Hz but also at much higher frequencies.

The more feedback you have in an amplifier the greater its ability to reject anything that isn't the signal, such as noise,

As a result, the class D is relatively unaffected by power cords at low power levels. But as the power is increased the power cord would be more important. It will not affect the amp nearly so much as power cords affect our tube amps!

Will the Atmas Class D be playing at Capital Audiofest?  Hoping so. 

@mapman We're hoping to have it there in the Classic Audio Loudspeakers room. I won't be there.

I see Bob hasn’t responded. Do you have a couple photos of your class d amp with the black faceplate that you can post…or link to?

No- there's not that many built with black so far and none in house. It didn't occur to get some shots of the black ones prior to shipment.

Good to know this Ralph. But it is not an option when you have a tube preamp - right?

@milpai That probably depends on the tube preamp. I don't leave any of my equipment on all the time, except my DAC.

The other part of the question....can they be made differential from INPUT to OUTPUT, as many conventional SSs are (ML, Pass,Moon, etc)? Is there a point?

@fredag

Things are a bit different when you’re employing switching technology, but FWIW the comparitor (which is at the heart of any class D amp) we use is a balanced device. The reason to go fully differential (we’ve been doing that with our tube amps longer than anyone else as far as I can make out, since the mid 1980s) is to reduce distortion and noise. The distortion sources in a class D amplifier are very different from what you encounter in regular non-switching circuits. So the need isn’t the same.

@fredag They have a balanced differential input (and RCA inputs are provided too). I run them balanced at home.

Thanks for any further thoughts based on your extended experience with the A-S Class D amps. 

@sdl4 There is a review in this month's (June 2023) HiFi+ magazine. The reviewer had given us a very nice review of our M-60 triode OTLs about 2 years ago. He does make some comments of comparison. In both cases he was using the DeVore O96s.

Are you still planning to build a version with somewhat higher power output?

We started with 100 W into 8 Ohms, 200 into 4 simply because if you really need more power than that your speaker is impractical. The exception seems to be ESLs like the Sound Lab, which are easy to drive in that they don't need an exceptional amount of power, but challenge solid state amps because the bass impedance is so high (30 Ohms).

We would like to make a more powerful amp, but to be noticeably more powerful than what we have already, it would have to be 600-800 Watts at least. This is simply because of the logarithmic character of the ear, which is why deciBels are used to measure sound pressure.

This kind of power won't be possible with the existing module. As always, the challenge will be to keep distortion down and keep it benign.

Yes, but I think it is more evident with tube amps than with s-state?

Yes.

Regardless of the amplifier though,  its distortion will always be higher driving a lower impedance. For this reason, if sound quality is your goal your amplifier investment dollar will be best served by a higher impedance speaker, all other things being equal (which of course they never are...).

However the class D seems to do nicely on lower impedances. Our local dealer likes how they drive the Magnaplanars in his store.

 

@audioquest4life Into 16 Ohms they will make half of their 8 Ohm rated power. As you already know though, that's plenty for the T1s. You might have to adjust the rear panel controls slightly since the class D has a lower output impedance compared to a tube amp. I've been driving my T3s (also 16 Ohms) for the last 2 years and they work great together!

Is there an official answer whether or not it’s ok to leave the atma-sphere class d amps on 24/7?

Its fine. The only concern is when turning a preamp on or off you might get a substantial thump so I would shut the amps off during those times.

I was watching a YouTube video from REL about connecting their subs to class D monoblocks. They recommend connecting their red wire to the amp’s positive speaker post and the black wire to chassis...not the negative speaker post. I’d prefer to hear from you if this is how it should be connected or if you recommend something different?

@veerossi I would hook it up to the red and black. If you run into trouble with buzz or hum, you can try tying to chassis, but I don't think its necessary.

Also in the hum and buzz department: some preamps and sources are not grounded. If using one of them and also the single-ended input of the amp and you have a buzz or hum, place a jumper between pins 1 and 3 of the XLR. You will only need this if the preamp or source is not Grounded.

I have the remote now, but considering downgrading to the manual volume if there is a lot to gain in terms of sound quality.

@veerossi @armstrod If there's a difference its pretty slight. I've not noticed it.

A brief history of gallium nitride (GaN) semiconductors

The fact still remains that Albertos amps use a switching device nobody else can source. Again, one that was designed for Audio, not for a battery charger or DC power supply. 

The only parameter that can make a GaNFET more suitable for audio is a low output capacitance. Other than that there's no way to optimize a GANFET for audio.

GaNFETs are designed for high power switching applications. While that might include a SMPS or battery charger, in such applications the cost of the output switching devices can play a pretty big role in the device used. So a battery charger is far more likely to use a MOSFET instead owing to an order of magnitude lower cost. GaNFETs are more likely to be motion control or other high power application (such as class D operation) where precision high efficiency/low noise switching is important. 

Anecdote department only: we have more than just 3 customers that have compared the AGD to our amps and the result was the other way 'round (one of them commented above). In the end anyone will still have to try it out and see.

@srama I recommend a balanced preamp if you can. That's what we use here in the shop and what I use at home.

@srama No.

Does the quote above suggest Alberto (AGD) is misleading in his saying his GANFET is made for audio?

No- not in my opinion. IMO it suggests he had a batch made that have low output capacitance. Low output capacitance reduces ringing in the output section (which is also caused by circuit board trace inductances and power supply issues- they can be in parallel with the output capacitance of the device which results in ringing and noise) as it switches and thus helps the devices to run cooler. The ringing can be controlled by a snubber filter but the output devices will still run hotter. So a low output capacitance is desirable.

In practice we found no need for a snubber circuit as there was no detectable ringing. IMO GaNFETs allow for lower noise designs. I think this is very important as switching noise and parasitics can mess with digital equipment even it the amp meets EU emissions directives. So you want to be many dB below the maximum allowed by most government entities. We found that many tube amplifiers radiate more noise than our class D.

@aw-agd, Likewise I would appreciate it if you didn't try to gaslight me.