^^ +1 @erik_squires yes, that response on George's part surprised me as well.
Showing 24 responses by atmasphere
Looks like there’s still no takers willing to put it on the line, and post up the 8ohm, 4ohm, 2ohm wattage measurements just before clipping, to see if they are "almost" doubling the wattage for each halving of load, so to prove if their amp/s are going to drive these types of speakers close to their best, not just "satisfactorily"!George, you like to promote GaNFETs as a thing in class D amps. But from this post is seems that you are unaware of something that GaNFETs bring to the table: very low 'on' resistance, often less than 100 milliohms. The devices we are using have an 'on' resistance of 60 milliohms.
That's lower than most linear output transistors! So a class D amp has no worries behaving as a voltage source provided the power supply can provide the current required into lower impedances. IOW less technical terms this simply means that doubling power into 2 ohms is no worries as far as GanFETs are concerned (and for a lot of other switching transistors as well). Our prototypes can double power into 2 ohms and we're not even trying to make it do that. It comes with the territory.
Whether its worth it or not is an entirely different subject as all amplifiers regardless of technology have higher distortion into 2 ohms which is audible as increased brightness and harshness. You can see this in the specs (if 2 ohms distortion is even specified, but you can also see it in the 4 ohm spec as opposed to 8). Further complicating matters is the critical nature of the speaker cable, which must be kept very short and also must have a fairly heavy gauge. At that impedance its very easy for the cable to contribute to the overall source impedance seen by the loudspeaker!
In a nutshell, 2 ohm loudspeakers are impractical and the ability to drive a load like that is over-rated. Sure, we like to play with muscle power and boast about it, but in the end it doesn't matter if you can weld with your amplifier; its intention is to sound like real music (not a hifi, unless its intention is simply to make money) and no amp is going to do that driving such a low impedance.
Another way of putting this is if you want to get the most out of your amplifier dollar investment, that dollar is best served by a loudspeaker that is a higher impedance, assuming that high quality sound is your goal. Any loudspeaker manufacturer can cause their speaker to sound smoother and more detailed by raising the impedance because when the amp makes less distortion, it sounds smoother and more detailed... Funny thing is, everyone knows this, its not some sort of hidden inner sanctum secret. But we still see these silly low impedance arguments when the simple fact is that the low impedance doesn't benefit anyone. You'd think that in high end audio where sounding real as opposed to sounding like a hifi is well understood, but here we are in the 21st century and this nonsense is still going on...
Nothings been put to rest, let’s see the independant test or even his own, that show the rms wattage doubling just before clipping (same distortion) with all three 8, 4, and 2ohm loads, then I’ll eat my words, and congratulate you on doing something that other Class-D’s can’t. I’ve asked the same from Merrill twice now and got no answer.Its put to rest everywhere but (by your admission above) in your own mind. Although we just got news this week that the USPTO is granting our class D patent with all 20 claims supported, that isn't the same as saying we're ready with a production product that is going to independent review. The power supplies that we are using for testing won't support the current of 2 ohms at full power even though the circuit boards we designed will. But the amp behaves as a pretty good voltage source so as long as you stay within the limits of the power supplies we're using, doubling power into 2 ohms is no worries.
I am well aware you didn't make that claim about GaNFETs but if you apply United States 5th grade reading comprehension skills (which these days isn't saying much) you will see that I suggested that this was something you might not have been aware of and by your admission above that surmise was correct. Again, what I said is that the output impedance of an GaNFET-based class D amp is so low that it will easily double its power into 2 ohms from 4 ohms- provided that the power supplies and heatsinks will allow it. These latter qualifications BTW are the same ones that limit any traditional solid state amps as well, so none of this is any surprise to anyone anywhere, except perhaps you.
All we have here are a few manufacturers spruiking their new or yet to be released Class-D amps without any shame, saying they "can handle" the 2ohm load, making out they can drive this speaker to it’s best in the bass. Without presenting any measured proof of their claims, that they can almost double their clipping wattage from 8 to 4 to 2ohms.With GaNFETs having 'on' resistances of only a few milliohms, its obvious that a class D amp using such devices will easily double power into 2 ohms (not that such is important for proper audio reproduction) as long as heatsinks and the associated power supplies can support the current. This has been stated before, its really not an issue of debate to one learned in the art.
What I've learned lately on this thread is to avoid using speakers that present loads to the amp of less than 2 ohms. I believe if speaker designers and manufacturers decide to rule out the use of the majority of amps working well with their speakers, then we can certainly decide to rule out the purchase and use of their speakers.@noble100 +1 on this comment and post. This is exactly the issue- IMO/IME its irresponsible to create such low impedances where serious power is required for the simple reason that distortion will be higher, and increased distortion will make the presentation less musical.
why can't discussions of audio stuff, hifi stuff and music listening stuff, be non confrontational ?really you have to ask this?? Because there's way too much BS and Voodoo in hi-end audio Mr D
Its not so much voodoo (although that is a problem); on this thread it really comes down to the posts of exactly one person.
Good quality sound? into hard speaker loads it’s all about current.This statement is misleading. Its really all about distortion.
Spectron amplifiers can deliver peak currents of 65 amps, with a staggering peak power of 3500 watts per channel for over 500 msec (!), which allows the amplifier to deliver the full transient (burst of music) without current or voltage “clipping”. "this statement is also misleading- here's why:
Current cannot exist without voltage. When the two are together you have Power, and the relationship is thus: 1 Watt = 1 Volt times 1 Amp.
If you don't know the voltage, you can use algebra to sort out the power anyway, using Ohm's Law. That relationship is used here:
Doing the math and giving the amp the benefit of the doubt, lets assume that the speaker load is 1 ohm. You'll see why in a second. Power equals Resistance times Current squared. So if we have 65 amps, the square of that is 4,225 watts into 1 ohm (just multiply that number by the impedance of the load if you want to know what 65 amps into higher impedances is). You can see right away that 65amps has nothing to do with the output of the amp. It likely represents how much current flows when the power supply of the amp is shorted for 10 milliseconds. IOW its really a measure of the capacitance in the power supply. FWIW, our MA-2 power amplifier (Vacuum tube class A triode OTL) has about the same current by this measure.
FWIW this sort of current number being bandied about is a common myth with solid state amps.
WHY!! they keep making low input impedance amps these days is beyond me, go back to 48k (the once recognized industry standard) or even 100k like it used to be with tubes, makes for a whole better range of preamps ,tubes SS or passive able to drive them properly.The reason you see this has to do with the fact that a filter often has to be installed between the audio input and the encoder (comparitor or the like). The filter is there to remove high frequency noise that might otherwise get amplified and add to the noise floor of the amp. But good high speed comparitora have issues of their own and can be unstable with a high impedance inputs. Hence low input impedances if there is no input buffering.
There’s no need for it in many cases, as sources and preamp have enough gain these days. It’s just "another opamp in the signal path" that’s not needed in many cases if the input was high enough instead of 2.2k, who wants that in hi-end if it can be taken out and one less opamp yay!!! gotta be good for the sound, and the business.This statement is incorrect. Of course there is a need for an input buffer in many cases- a lot depends on the encoding scheme of course (for example with PWM the gain is based on the difference between the amplitude of the triangle wave in the circuit and the input signal). Things might not work out in the design that you have enough gain- particularly if feedback is employed. So an opamp or instrumentation amplifier might be used to make up the needed gain. Quite often this isn't a lot of gain, maybe a gain of 2 to 10. You could do it with an input transformer which has certain advantages, but it should be obvious that an input transformer adds a whole set of problems of its own.
Our preamps can drive such low input impedances- any preamp that supports the balanced standard can, but in high end audio such preamps are quite rare.
Here is the comments from a manufacturer of these Purifi’s Class-D amps when I queried him on the unnecessary low 2.2k input impedance. I think it could be because of the use of bjt input stage instead of fet, I’ve yet to suss that one out.I explained why already. This is a common issue and why you see many Class D amps with some sort of buffer input. Many opamps can drive lower impedances like 600 ohms no worries and since so much feedback (+30dB) is being used, the assumption is that the distortion added by the feedback on the opamp will be quite minimal. But there are some class D amps where the designer opted to simply have a low impedance input for the very simple reason of having less circuitry- a purist approach. But you have to have a source that can drive it. Its nothing to do with FETs or BJTs.Yes you are absolutely correct that you can use a high output DAC directly into this module. "That would be the ultimate solution"
just if they don’t try to double from 8 to 4 to 2ohm, you won’t get the very best out of them, they will be compromised in the bass, if you don’t mind that that go ahead.This statement is false. The ability to double power as load impedance is halved has nothing to do with how well the amp will play bass- even if the bass is where the load impedance is lowest. Quite simply, it doesn't work that way; a lot more has to do with the output impedance of the amp which is independent of its ability to double power.
George, the quote above suggests that you do not understand how a voltage source works.This statement is false. The ability to double power as load impedance is halved has nothing to do with how well the amp will play bass
In case you don't understand what a voltage source is, solid state amps that can double power as impedance is halved are behaving as a voltage source. Having a low enough output impedance is for the most part what makes for a voltage source. A voltage source is defined as a circuit that can make the same voltage (not power) output into whatever load it drives. Now no circuit can really do this perfectly; perfection does not exist. So constant voltage is considered load invariant under the Voltage Paradigm rules. Historically ElectroVolice and MacIntosh led the way to this concept back in the late 1950s.
To be a voltage source, the circuit does not have to double power as impedance is halved. But it very much has to be able to cut power in half as impedance is doubled! This is why tube amplifiers can operate as voltage sources and its also why a solid state amp that can't double power into even 4 ohms can nevertheless still play bass correctly into a 2 ohm load. Whatever power it makes into 2 ohms, to act correctly without distortion it will be half of that into 4 and half again into 8 ohms.
So for those amps that do not have the power supply capacity, or the current capacity to survive higher powers in the output section, or for tube amplifiers which are well-known to not double power as impedance is halved, the designer simply has to make sure that the amplifier has enough negative feedback so as to reduce the output impedance such that the circuit behaves as a voltage source into the expected loads.
If the amplifier innately has a low enough output impedance it may not need the feedback to achieve constant voltage operation. An example of this are the Ayre amplifiers. But any class D amp (GaNFET or not, even sans feedback) has a low enough output impedance as well. So as long as they are operated in their linear region 2 ohms is no problem.
Again, this is defined by output impedance. In the case of a class D amp, the output impedance formula is dominated by the 'on' resistance of the output devices. Even in an older MOSFET based class D amp, the output impedance is usually well below 0.1 ohm, meaning that they will have at least 20:1 damping factor into a 2 ohm load. If the amp runs feedback (and most of them do) the output impedance will be even lower than that.
The real issue of a class D amplifier driving 2 ohms is actually the filter at the output. What I regard as a bit odd here is that while you have acknowledged numerous times in other discussion the issues of a filter at the output of the amp, in this particular discussion you've not brought out this salient fact even once! This leads me to think that you have no idea what you're talking about. (Please note that what I think you know and what you actually know are likely two different things.) BTW, the way to deal with the filter issue is to design the filter to operate at as high a frequency as possible- hence the advantage of GaNFETs as they allow a much higher filter frequency since the switching frequency is so much higher. This allows the filter to still be effective for hifi purposes when presented with a 2 ohm load.
You like to routinely discount my comments out of hand and insult me at every turn, despite the fact that I went to school for this sort of thing and actually design amps (including GaNFET class D amps) for a living. You're a fan on Nelson Pass- why not ask him?
Yes you use this old chestnut everytime as it suits you to bring it out every time your OTL’s are brought into question doing the same thing.This statement is false. Our OTLs don't double power and in fact are not intended to act as a voltage source. They are intended to act as a Power Source.
So it’s simple for all, instead of doing war and peace to confuse everyone as he usually does.While this comment is a bit of a word salad, if I make it out right its mostly correct and doesn't repudiate anything I've said. Just about any class D amp that can make 1000 watts won't have any troubles making as much power as the ML2 will do into 2 ohms (100 watts) so what this comes down to isn't whether the amp can do 2 ohms better, its simply if it sounds better at all.
Look at complimentary BJT’s with massive power supplies that I’ve been suggesting for these loads, then look at your Class-D Mosfets usually not complementary as the P channel stinks, then the 8th grade maths don’t mean ****..This statement is false. One of the advantages of a class D circuit is that since the output devices are only switched on and off, matching the devices is unimportant and the need for a complementary device eliminated. McIntosh
MC2KW is class AB and has 2,000 W in 8, 4 or 2 ohms. In his book it is a bad amp with not enough current because it does not double down too...Yes- as long as it can behave as a voltage source, doubling power is unimportant. Our amps don't even act as voltage sources and they've been getting awards in the high end press for decades now. Its a common audiophile myth that doubling power is the most important thing about an amplifier when other aspects such as distortion or bandwidth are often far more important. A myopic viewpoint is often limited...
George’s arguments are irrational.To which George replied:
So you my friend are the irrational one not me.George's argument was both incorrect and irrational. This is different from saying that a person is irrational- the principle to use of course is 'attack the argument not the person'. Calling someone else irrational as we see here is a forum violation. In addition, it employed a logical fallacy known as a 'Strawman', since George was trying to say that he had been called irrational when that didn't happen. Logical fallacies are by definition false.
M.S Electrical Engineering, University of Southern California (2017)Here is more of the text George didn't include which precedes the above text by Jacob VonWagoner:
BJT's can give you a lot higher gain. Just take a bunch of components and compare them, and you'll find the BJT's give you better gain characteristics and therefore require fewer gain stages.The words 'gain', 'linearity' and 'bias voltage' tell you that this has nothing to do with class D amplifiers which do not operate their output devices in the linear region. Instead class D amps go from Off to On and spend as little time in the middle (linear region) as possible. This is the reason they are so efficient.
Obviously while all this stuff is correct with regards to conventional amplifiers it has nothing to do whatsoever with class D amps. George's remonstrations notwithstanding.
You’ve now got to be joking!!, out of courtesy I didn’t mention which OTL’s wouldn’t drive these 89db JBL’s 1400’s, but now you leave me no option, they were a pair of your M60’s!!!!! and now your saying they’ll do a nice job on Wilson Alexia!!!MA-1s. Not joking- Wilsons have always been fairly efficient despite their impedance. They seemed to sound just fine to me. They played loud enough too but it was a smaller room IME. We also noted that particular customer was going through power tubes faster, which is no surprise since a lower impedance will cause more of the power generated by the output section to be dissipated in the output section- and that's hard on tubes. We finally got him to install a pair of ZEROs and then the tubes stopped failing, plus he had more power. The Wilsons traditionally have been fairly efficient so none of this should come as any surprise. The national sales manager at Wilson (John Giolas) had our amps for several years.
@georgehifiStereophiles JA’s summation, "practical use with speakers whose impedance doesn’t drop below 3 ohms."
You like to put up specs; now you have to also live with what they say. And what they are saying is:
The Mytek can make about 300 watts into 2 ohms and therefore is able to double power from 75 watts into 8 ohms to 150 watts into 4 and finally 300 watts into 2 ohms. If we take @viber6 at his word, his amp plays nearly 5dB louder than yours into any load your amp can drive and sounds better at the same time. This assumes that the bass region is where the 2 ohm load occurs- where most of the musical energy lies.
But if the 2 ohm load occurs at 8 or 10KHz the story is quite different since so little power is used at that frequency. Then the Mytek would be able to play nearly **10dB louder** than your amp and sound better all the while if we take @viber6 at his word.
Quite simply an amplifier does not have to double power into 2 ohms from 4 ohms in order to behave as a voltage source. The only time this might be an issue depends on the impedance curve of the speaker and at the full power of the amp. If less than full power things are very different!
Of course you can take my word for anything I say I hear. Also, because I am not in the business I tell things like it is without concern for politics.I didn't doubt you :)
This thread should have been finished after the 5th post. I think your response nailed it perfectly. Class-D and speakers < 2ohms just don't work, with maybe the exception of some of the latest offeringsWith class D amps the older amps were limited by their output filters- not their current. With newer amps that can switch faster, the output filter can be set to a higher frequency and so 2 ohms isn't the concern it used to be.
2 ohms requires more current of course. Most output sections can handle the current; we're using devices rated at 30 amps each which is double the current capacity required for our prototype to make full power into 2 ohms. The problem isn't the class D so much as it is whether the heatsinks provide adequate cooling (important even for GaNFET-based output sections into that sort of impedance) and whether the power supplies have adequate current to support the power.
So like any other amp: some do some don't. Simply inquire with the manufacturer. Yes- just that simple... 5 pages later.
By all means please show the do’s in Class-D, hopefully independently tested, impedance v rms wattage all at just before clipping into 8, 4 and 2ohms, the same distortion figures.Ha! That's pretty funny!
**No** amplifier will have the same distortion figures. All amps regardless of technology used have higher distortion into lower impedances. This has been seen over and over again in tests by JA at Stereophile, Bascomb King and others all over the world.
And right here might be the simple salient fact to the pointlessness of the inquiry: If you want the best results from your amplifier investment dollar, that dollar will be best served by a speaker load of higher impedance ***IF HIGH QUALITY SOUND IS YOUR GOAL***. (Sorry about the caps- I was looking for emphasis, not shouting). This is where that magical line in the sand occurs- the difference between the results sounding like real music or just a really good hifi.
Plain and simple low impedance speaker loads are a Bad Idea and it doesn't matter if you can weld with your amplifier- it simply will make more distortion into that lower impedance, and that distortion will be audible as less detail and harsher. Doesn't matter if the amp can double power or not- the higher distortion will be there nonetheless. Again- its easy to see in the specs and independent measurement.
The story of Sisyphus is there as a teaching moment and applies directly to the the task of driving lower impedances.
Sez you, some of the very best are insanely hard to drive, and why class-d, OTL and you have a problem with that.Sez the specs of all amps. Obviously this statement is misleading; George is cherry picking. On the one hand he likes to trot out Stereophile ratings to support his claims but when that same source doesn't support his claims then its ignored.
Some of the very best speakers are also very **easy** to drive. The issue is simple: Do you want the sound of a good hifi or the sound of real music?? If the latter, then you will want a speaker that is both state of the art and easy to drive since distortion is what separates the men from the boys.
While you and Ric Schultz are technically correct that higher impedance speakers get better performance from amps than low impedance speakers, this ignores the overwhelming superiority of low impedance electrostatics for most musical criteria, except in large SPL dynamics where the conventional dynamic speaker excels. If someone designed a higher impedance stat, that would be great, but the disadvantage of somewhat higher distortion (and it is still fairly low) into their very low impedance is vastly outweighed by their superior transient response, clarity, coherence, less coloration, etc. The old dictum is still true--the speaker is the most important element in the chain, with the possible exception of source quality. Get the best speaker you can afford, then get the best amp that will drive that speaker.Electrostats are a bit of an exception. Their low impedance derives from a capacitance and is at the upper frequency extreme where very little energy exists! So you don't need a powerhouse to drive them. Sound Labs are a great example and some of the very best ESLs ever made. We have lots of Sound Lab customers- about 80-85% of all our MA-2s go to Sound Lab owners. IMO, the Sound Lab is one of the top 5 speakers made price no object. Even Martin Logans don't need that much power- we've driven them with our M-60s (and a set of ZEROs) no worries.
But I don't agree about the 'best speaker you can afford' thing! The reason is simple- you may find that you have a preference for an amplifier technology- tubes for example- and if that is the case buying an incompatible speaker will simply be money down the loo. So- if you know what kind of amp you prefer, then get the best **compatible** speaker you can for it.
Wilson, Thiel, ML esl’s, MBL, Sonus Faber JM labsOf these only the Thiel is really difficult to drive. We have customers with Wilson, Martin Logan and JM Labs. And we make tube amps that don't double power or even act as a voltage source. The mbls aren't that hard to drive either; just have a bump (not dip) in the upper midrange driver that can make an amp with no feedback sound bright. But the mbls are easy to drive; most of their 'inefficiency comes from the simple fact that they are omni-directional.
The more general point is that the sound character is largely determined by the speaker, closely followed by the recording. Most good amps today have superb specs and sound fairly close to each other.@viber6 Actually this isn't true. It is true that speakers have a lot of sound character (as does the room) but amps don't sound the same at all, nor do all amps interface correctly with all speakers.
One example is how solid state generally doesn't work well with full range ESLs. Full range ESLs have about a 9 or 10:1 range of impedance from the bass to 20KHz. For example a Sound Lab is 30 ohms in the bass and between 1.5 to 3 ohms at 20KHz depending on the position of the Brilliance control (if the control were removed it would simply be 3 ohms). (BTW, Sound Lab solves this issue to some degree by having a bit of adjustability to their speakers- bass settings, midrange settings (IIRC) and the Brilliance control.)
Now most solid state amps are designed to work as a voltage source, which is to say they maintain the same voltage regardless of load. This is the idea of doubling power as impedance is halved or halving power as impedance is doubled. So If the solid state amp makes 100 watts into 8 ohms, On a Sound Lab ESL with 30 ohms in the bass it will make only about 26 watts but with the same level signal applied at 20KHz will make 300 watts- over a 10:1 difference! The problem here is that unlike a box speaker with a driver in it, the ESL's impedance curve is based on capacitance rather than a base impedance influenced by resonance. Put more simply, it **has the same efficiency regardless of impedance**. This causes solid state amps to be bright on top and unable to make power in the bass. Martin Logan gets around this by simply having a super low impedance at high frequencies and most solid state amps (tube amps to a greater degree) can't make power into that impedance and so they tone down the brightness. In a nutshell, ESLs are not Voltage driven; they are Power driven. For more on this see http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php
Conversely a B&W 802 is designed for a voltage source amp (solid state) and so tube amps have a very difficult time making bass on this speaker.
But put even more simply, we've been getting spec sheets from amplifier manufacturers for 60 years, but we can't tell what the amp sounds like. So we have to audition the amp on our speakers to see if it satisfies! This simple fact is common knowledge with all audiophiles.
Of course we have to examine what is considered 'superb specs'; the simple fact of low THD doesn't mean the amp will sound uncolored or musical to the human ear- in many cases quite the opposite! This is why tubes are still around 55 years after being declared 'obsolete'. Its complex, due to how the human hearing perceptual rules work; how we perceive sound pressure (the presence of higher ordered harmonics, FWIW), how the ear converts distortion into tonality and the masking principle all play a role.
Because some manufacturers want their amplifier to conform more closely to the human perceptual rules, they are willing also to have 'inferior' specs. Its a simple fact that as our current regime of test and measurement goes, we're not always measuring the right things.
With the stat speaker, impedances can be as high as 30-100 ohms in the upper/lower bass, and 1 ohm or less in HF. This is more than the 10:1 ratio you cite, but still nowhere near as much as the inverse RIAA curve. Still, according to your theory, the stat speaker should sound like a less drastic inverted RIAA curve, which I didn't find in the listening. Also, all dynamic speakers have marked variations in their impedances, and your theory would predict that a SS amp would produce markedly different sound from different speakers in relation to each speaker's impedance curve.
@viber6 this isn't correct. (BTW what I'm talking about here isn't theory, its more like audio history. The Power Paradigm is what was around before MacIntosh and ElectroVoice developed the Voltage rules in the late 1950s.) With most dynamic speakers (there are exceptions in high end audio) the speaker is meant to be 'voltage driven' so variable power output is what you're looking for in the amplifier in order to get flat frequency response. The idea is plug and play, no need to adjust any midrange or treble controls (like you see on older speakers, stuff from the 50s and 60s). Of course like anything else, this approach solves one problem while introducing others.
The brightness of an unequalized RIAA curve would **not** have been compensated by the way the ESL and solid state amps work together. I think you will find though that the 10:1 ratio I described is generally pretty close- that describes Quads, Martin Logans, Accoustats and Sound Labs which are the ESLs with which I have the most experience. If you can find one that is outside of that (for example: 100:1) I'd be very interested in knowing about it!