Showing 50 responses by audio2design
Stereophile tester JA does not say what George thinks he says and JA is not an amplifier designer. Neither is George hence why he keeps making wrong statements about Class D. Read some other threads. George can never justify his statements and can never communicate why they are true.
You have the floor George. Tell us exactly the effect of EPDR on Class D. Go for it.
Most of George's post is wrong. EPDR has no meaning really for ClassD. As well, doubling power with half impedance has little meaning. All that matters is distortion with low impedance loads. Most amps that double in power start at comparatively low power at 8 ohms but are still "large" amplifiers. They have relatively low rail voltages. ClassD don't have the same design restrictions as linear amps. An amp that is 400 watts at 8/4/2 ohms has as much capability as an amp that is 100/200/400 watts at 8/4/2 ohms. What matters is what is the limiting factor.
Ask yourself why all the Ncore data sheets have the 2 ohm power ratings in the data sheet right near the top. Some people never move forward and have to bring up a one time comparison of a $40,000 dollar amplifier against $10,000 amplifier that no one can verify. Well not really $10K. They were clones of the Bel Canto, so unknown pedigree, and unknown performance. Now we are talking $40,000 something that has likely never seen a measurement that confirmed performance.
One day George will discover that EPDR is not the same as current delivery. That day is not today.
Yes, please read about EPDR, then you can tell George he does not understand it just like every other person who understands amplifiers has told him.
Equivalent Peak Dissipation Resistance - i.e. EQUIVALENT resistance to equal peak thermal dissipation. Not peak current supply. Not minimum impedance. The EQUIVALENT (not actual) resistance that would equal the peak THERMAL dissipation.
This only has meaning for traditional amplifiers that operate in the linear region, i.e. Class A, AB. It is where the (current output * (rail voltage - output voltage)) is at a maximum. With a linear amp, let's say the current output is 10A, and the voltage difference = 40V at some point in time. That would be 400 watts dissipated across the output devices. That is the thermal dissipation in a linear amplifier.
In a Class-D amplifier, the average (as they are switching) voltage in the output devices, while they are conducting, will be a small fraction of a volt. Let's say it is 0.1V at 10A. The dissipation in the output devices is now 1W. Notice that the rail voltage does not enter the equation? Now practically there is a small contribution from the rail voltage, but there is not a direct correlation like in a Class A/ AB amplifier because obviously they don't work the same way.
Now one thing a Class-D amplifer may due is hard limit current which they will "measure" or at least check for a peak on every single switching cycle. This provides short circuit protection and protection that the output inductors do not enter saturation. As that current will be fixed, this is why in a Class-D amplifier the output power probably does not keep doubling with halving of the output impedance. The power will probably increase a lot between 8 and 4 ohms, because in this case the major limitation in output power is due to the power supply rail voltage. It will likely not double into 2 ohms because you will run into the current limit before the voltage limit.
What does this mean? As long as the amp does not have stability issues at 2ohms, which most newer ones will not, and you don't drive the amplifier into clipping, then the class-D amp will be just fine into low impedance loads no matter what some people who make claims, but don't understand the underlying technology make.
But, of course, this isn’t the case. Typical class D amps are 90% or more efficient into 8 ohms (at max output power), but efficiency drops by approximately 40% into 4 ohms, and 40% again at 2 ohms.
I am not sure you wrote this correctly. I think you mean losses go up 40%, or stated as efficiency, the reduction in efficiency goes up by 40%, i.e. 10% reduction becomes 14%?
Conduction losses are I^2 * R. At the same wattage, half the resistance, current goes up sqrt(2) = 41.4%, so losses must go up 100% due to conduction losses, but realistically total losses are going to be a quiescent component (in this case about 11 watts), plus a linear component and a squared component to come up with a very good model.
For reference as example, the NCORE 500 OEM module is about 95.2% efficient at 400W/8 ohms, about 92.6% efficiency at 400W/4 ohms, but drops down to about 87.5% at 400W/2 ohms. Losses are 20, 32, and 67 watts at 8/4/2 ohms, 400W output.
For the NC500, the current limit is 26A. That puts maximum possible RMS power at 2ohms = (26/sqrt(2))^2 * 2 = (26*26/2)*2 = 676W, not too far from the rated power of 550W as a ratio, and even closer if you add losses above to the 550W.
I smell another fuser
Sorry, not a "fuser" but I do understand AC circuits. Not so sure you do based on your posts.
May want to check out some tests at audiosciencereview.com With the exception of the Benchmark, nothing out there matches the best Class-D on many critical specifications.
Nope, but you do have trouble understanding how Class-D amplifiers work, and amplifiers in general as it relates to loads. I can and have talked in great details the real limitations of Class-D amps as it applies to low impedance loads. I Challenge You to do the same.
and that it’s switching frequency is too low for it’s output filter to remove completely without introducing large phase shift down into the audio band.
And wrong again. Been fixed for over 5 years now. Sorry that your information is dated and you don't understand enough about amplifier design to know how this works.
can your attenuator drive Class D amps
you are a **** why bring that up if you know the answer already. If they have >33kohm input impedance buffers it can
Why do you keep posting wrong information?
But no need, I go direct into my Class-D’s from my dac
I don't know your DAC, but I hope it has very low output impedance. The NC500 without buffers needs a low impedance source or you will degrade its performance.
I don't know what his agenda is, but given he spreads grossly false information and refuses, absolutely refuses to educate himself such that he understands the statements he makes, I don't get it at all.
Come on George several messages above I went into great detail about what EPDR is and how that impacts a Class-D amplifier. I went into great detail about hard current limiting and explained why the watts don't double at 2ohms. I think I made it simple enough that many people can understand.
Tell me, at this point, who do you think people are going to believe, someone who goes into great detail to explain the process, relates it to actual amplifiers, and uses actual numbers to back up their arguments, or someone who can't do any more than link to an article on Stereophile that they don't even understand?
I will address one other thing on Class-D amplifiers, thermal shutdown. Class-D amplifiers will often shut down early in amplifier tests when the current load is high and continuous. Why is that? This is actually a design choice. The market, at least initially for Class-D, is compact, light weight, efficient amplifiers. Sure, there is a portion of the market who is quite happy with an 80lb monoblock Class-D, but for the most part, that is not the market, even in audiophile land. People want a powerful, efficient, and still compact amplifier. That puts limitations on the size of the heatsink. Most Class-D amplifiers don't use any more than the small aluminum plate heatsink they are built with. As well, for good circuit operation, the switching devices are small and surface mount to keep parasitics low. Unlike a linear amp, you can't mount those transistors at the end of inches of wire or PCB trace if you want good performance. Now sure, you could design a system, easily, but more expensive, where you could effectively couple to a large heatsink and keep the circuit compact, but that would greatly increase implementation difficulty and cost. I did a prototype once using a heat pipe assembly designed for a PC.
Now I think we all know that the ratio between the peak wattage and the average wattage is quite low. 20:1 or even higher is typical. So even though you may be hitting 500W peaks, 10-20 watts may be your average. Most Class-D amplifiers, knowing their target market, take this into account. Class-D amplifiers are already efficient, so creating a design that allows very high peaks, does not require a design with high quiescent power draw and dissipation. Since it is not needed, it is not included. That works great for music, but does not work great when someone is trying to run continuous tones for maximum wattage testing.
The ’switching noise’ of any class D amplifier is a sine wave at the switching frequency. If the filter is properly designed, the sine wave is quite small and low power- it has to be so because the amplifier can’t interfere with other services, like AM broadcast (1.5MHz is at the top of the AM band). But an amp that is switching at 600KHz will have the same amount of noise and 600KHz is in the AM broadcast band too- the same rules apply.
An amplifier is an unintentional radiator. FCC (and others) will only test conducted emissions at <=30MHz. Radiated emissions is tested at >=30MHz.
The noise would be a square(ish) wave as the transistors are hard switching. It would become a sine due to the output filter.
This is very much not the case. Most are based on modules, and those modules are not designed with thermal paths / heat sinks that will allow full power continuously or more specifically high current continuously. That's fine since music doesn't work that way. The devices normally used are not the absolute lowest RDSon, but more so tuned for good switching performance. When they heat up a lot of course, that RDSon goes up.
Oh George, I don't think you know the meaning of the world troll. I am sure I upset a lot of people here. It comes with the territory. They, like you, spew how you want the world to work or believe it works, I have the knowledge and experience from research to practical to know quite well how all this "audio stuff" works, right from recording tech all the way to small details in reproduction. That is the benefit of actually doing research (real research, like in a lab), and developing products instead of whatever you do.
Your position on Class-D is laughable. Atmasphere has explained in detail, at a simple level you can understand, why so much of what you say is wrong. I have explained, in terms simple enough for you to understand, what so much of what you say is wrong. Yet you refuse to take even 5 minutes to understand, thinking that reading an outdated effectively graph in Stereophile gives you more knowledge than people who have actually designed amplifiers, who understand both the math (mainly control systems theory), the electronics, and the practical implementation details. But no, you are a greater expert than us, even though you can't even begin to explain the simplest details about your claim.
I would go back to making yourself useful trolling fuse threads, and let people who know what they are talking about talk about class-D. There are no doubt people on Audiogon more knowledgeable on some areas of audio than I am, but I expect the people you claim are "upset" don't fall into that category. I am quite happy for Atmasphere or Duke to correct me, I would rather be shown wrong so others can learn, and mislead people. That is my character. What is yours?
Really, do you really want to embarrass yourself yet again?? Calling me a fuser. What a laugh. Everyone who has read my posts here knows you are lying, but you are still upset I have shown you don't know how electricity actually works. Oh well.
As well, it is time you stopped this classless attack on Atmasphere who has been pretty much right on the money with everything he says. Your comments about "product protection" are something you should be embarrassed as an adult to make. Atmasphere is just too nice to say what he probably really wants to.
I don't see anything in Cyrill Hammer's bio that would suggest the extensive knowledge in signal processing or advanced control theory or even ultra high speed switching for analog reconstruction that would make his opinion any more valid than many many others, and perhaps less. That GHz comment really makes me suspect his depth of knowledge on the topic is really quite weak.
You have proven repeatedly you don’t have anywhere near the knowledge to make a comment on someone’s credentials. Your continued lack of understand of even the most basic aspects of amplifier operation, class-D operation, and even basic electricity make your comments on this topic meaningless.
However, someone with a basic knowledge of signal processing, the math of signal processing, noise sources in a switching amplifier, or the basic math defining their performance, or say the related math of DACs would know that a comment about needing switching in the GHz range is really quite laughable. That tells me he really does not know what he is talking about and is just making an off the cuff comment or parroting someone similarly unqualified to make a comment.
It is not healthy to put on a pedestal people who simply reinforce your biases without ascertaining if they truly have relevant knowledge in the topic of discussion. You don’t have any reason to place high value in Cyrill Hammer’s knowledge w.r.t. Class-D. Based on the public information, which we can assume we both have equal access to, I don’t perceive anything in his experience set that would make me place high value in his opinion on this matter, unlike say Bruno Putzey who has shown repeatedly, both a high understanding, and an ability to move the technology forward.
I will state again for the slow kids at the back of the room, Soulutions "talent" is
The comment about GHz switching was not made by their "talent", it was made by a person with technical background, but ultimately the business manager.
in the GHz range
or even GHz.
You really think these don't say the same? He clearly said, "or even GHz". That is a silly statement that shows a clear lack of understanding of the fundamental way in which Class-D works.
The PUTZ is you George and that is being kind. Is that the level of name calling YOU need to use when you have absolutely 0 ability to form a proper argument? That says far more about you than about me.
As Atmasphere has clearly said, and as actual measurements have shown, you don't need multi-MHz and certainly not GHz switching speeds to not have phase shift or switching noise in the audio band. YOUR knowledge is evidently sorely lacking in this area. Common sense would tell you that signa-delta DACs don't need to "switch" in the GHz range in order to not have switching noise in audio band. It should tell you that. What that tells you is a mystery.
Given his comment about 1GHz for Class-D, that statement you just made is obviously not true. I would say it is self evident to someone who understands the technology with any degree of competence (that would exclude George), that Atmasphere and I, on this area, seem to know far more.
I am sure there are areas of circuit design that Cyrill is more competent than I. I am also pretty confident there are areas where I am more competent, and again, that ludicrous GHz comment makes me question who is the high powered technical talent at Soulution. A link to a website proves nothing. Let’s do a little digging ...
Darn, I must be clairvoyant! ... or just experienced enough to know people who really know what they are talking about and those that do not?
Then again, based on his work experience which has been more on the business side, less on the technical, maybe I was giving too much credit?
I will give him props, even if just repeating what his designer said,
Cyrill Hammer: For a “solid-state” amplifier design the speed (e.g., bandwidth) of amplification is one of the most important criteria. This speed or bandwidth has nothing to do with the MHz-range frequencies that can be reproduced by such an amplifier; the bandwidth is required to make the “feedback loops” of solid-state designs work properly.
Hey @atmasphere , sound familiar? ... of course, I take exception to their claims to be pioneers in either recognizing this or designing with this in mind. A friend did an app-note in the 90’s on a current feedback audio power amp topology that accomplished this. Of course op-amps have been doing this for quite a few decades as well.
Stop posting tired graphs, of an older technology that no longer represents state of the art. Are you familiar with the saying correlation does not equal causation? You are using the ignorant premise that because one Class-D amplifier has lots of phase shift that all Class-D must. THAT YOU REFUSE TO LEARN what it means to include the filter in the feedback loop shows your dogmatic ignorance.
What is really funny is you idolize Soulution, even though the person who designed it is not the person quoted ..... well guess what SUNSHINE, Soulution uses a ton of feedback, just like Atmasphere has said repeatedly. A TON. They do it by having a large gain bandwidth. Really this is 2nd year circuit in University for engineers. Oh right, you don't actually have an engineering education. Silly me.
How is that any different from digital audio,The amount of power the filter has to contend with in Class-D amps compared to a dac, now your showing how ridiculous your statements are.
Seems George does not understand how LC filters work.
note the size of the 4 x!! output switching frequency filter coils in this $53K ML Class-D they had to use, and that’s a monoblock.
Without talking with the designer I cannot be sure, but looks like they may be toroidal air-cores or a very low permeability magnetic material to fix a perceived issue (that quite possibly does not exist). This caused it to be large. Designers do crazy things all the time. Class-D is a big step up for many amplifier designers used to somewhat cookie cutter designs. There really is not stopping anyone from using a 2 stage LC filter at a high frequency, well other than there is no sonic value to doing so.
That's about 450Khz, which with most speakers will never make it to the driver. Point?
Instead of just posting a link, you may want to actually read it and understand it. Note the differences between open and closed loop. Note how the closed loop has better performance at low frequencies but the performance degrades quite a bit at high frequencies? This is a consequence of inadequate design, pretty much as atmasphere has repeatedly stated. Thank you for providing yet another excellent example.
This eval is tech evaluation platform. It is not intended to push the limits of technology or topology. If someone is looking to buy modules and DIY, they would be better served with ncore or purifi modules with significantly better performance.
I don't try to hide that I think a lot of "tweaks" are highly questionable, and as opposed to what many believe, I have had a large group of audiophile friends who over the years have tried them, so I have been exposed first hand to them, and see difference in blind/sighted listening. Some, based purely on the claims, are physically impossible. I know stating that irks someone. Almost never has anyone ever come back with truly empirical evidence to support their position, even if they resort to "I can trust my ears". I do find leaping from "I trust me ears" to "it must be true" distasteful, almost as much as "I like it", therefore "it must be accurate", however, those argument and conversations play out much differently from this.
We are not "arguing" over likes, or personal impressions. We are discussing hard, factual engineering topics and concepts. These really are effectively black and white topics, and there is an absolute right or wrong. People who obviously have a deep and accurate understanding of this topic post detailed explanations and why things are the way they are, work the way they are, etc. We don't sugar coat, of exaggerate.
In the face of that, someone else, with clearly an incomplete and inaccurate understanding of many of the related topics to Class-D repeatedly posts technically erroneous information. This is not even debatable. It really is black or white, right or wrong. They are wrong. Absolutely provably wrong, and they have been. They don't recant, or consider they may be wrong, they double down and repeatedly post false information labelling it as factually correct, even though they cannot describe it in their own words or justify their position.
Last, I find the repeated personal attacks on atmasphere totally uncalled for. It is like he has a personal grudge.
You would be hard pressed to find a ferrite core in a mid/tweeter for even a mid range speaker. Air core inductors are fairly cheap at the frequencies needed. For woofers, there are trade-offs w.r.t. resistance / core impacts.
In speakers, the cross-over is in the audio band. In a Class-D amplifier it is not. In a speaker you don't normally have steel nearby. In a Class-D amplifier you may.
No one in their right mind would use a ferrite core coil in the midrange and tweeter xover of a serious speaker. Ferrite messes up the sound. So, why do we have ferrite coils on the output of all class D amps? It is because they are smaller, cost less, have less resistance, run cooler and have better contained fields so there would be less rf outputting the amp.
Less resistance is definitely important leading to high damping. They also have a closed magnetic path which means they are not effected by external steel which could even include component leads ... and they don't turn external steel (which could be anything) into a speaker either.
Given the very low distortion and IM, it is hard to argue that the design choice is poor. All depends on the ferrite material and where you operate it on the curve. Lots of well rated audio products have output transformers. Choose your poison.
Is that the same NC500 you bypassed the buffer clearly violating the recommendations of the app note that the amplifier stage be driven by a low impedance source to maintain performance?
There is also capacitance in that input which means that a high impedance input will cause phase shift and a non flat frequency response.
The difference between me and you is I understand feedback, you do not. One of the amplifiers you previously went gaga about Soulution, uses a ton of feedback. You just don't know what you are talking about.
So back the question, how much of the phase shift in the Technics is from the filters, and how much from intentional correction in software? ... inquiring minds want to know.
Georgehifi whines about feedback in ClassD while idolizing solid state amps with tons of feedback.
Georgehifi idolizes Technics while ignoring they need to use DSP and variable delay to compensate for phase shift in their output filter.
Ignores circuit design requirements then faults performance.
Just dogmatic rambling at this point. The hole is dug too deep to be willing to admit error.
Yes, atmasphere, I come here for much the same reason, and georgenofi, is an almost constant stream of misinformation.
Yes I sure do admire them in Class-D circles for advancing/sorting out (in the SE-R1) one of Class-D’s main problems, switching noise filtering which creates up to 70-degrees phase shift down into the audio frequencies https://ibb.co/MS3xC6M.
Please all note, that georgehifi, I assume out of previous technical ignorance, but now I have no idea, refuses to accept the fact that Technics does not eliminate phase shift with only high frequency switching. They also apply digital filters to shift the phase at higher frequencies. They must do this because they do not have the technology to address it in a different way.
You mean like these amplifiers that you idolize?
Gryphon, “big” Krells, or D’Augostino, maybe a JC1 pair
The other problem is current delivery into very low impedance speakers, like the classic linear high end amps can, that could take some more time, but they can drive the speakers that aren’t such a savage/nasty load like Wilson Alexia etc, so it’s not an overhaul big problem.
One of the better known Class-D amplifiers has an instantaneous current peak of 26 amps. For those better at math, that is 1350 watts, though that would require a suitable rail voltage. How long it can sustain that is a matter of power supply and thermals. Most Class-D amplifiers are designed for real world performance without added cost that will only be beneficial in a bench power test. You would be very hard pressed, if not impossible to find music where the average power is >1/4 max peak where average is a 50msec window which is exceptionally short. We are talking loud grunge, not what most of us would consider music. More realistic would be 1/10th where the max peak is compared to the maximum average over a sustainable period. So that 1350 watts peak would match up with 150W continuous on the bench. If you have an amp that can do 1000W continuous, and 1000W on peaks, on real music, it will have no benefit over 150W continuous, 1000W peak in real world usage.
p.s. George, before you bring up 1 ohm, I will remind you that EPDR has no bearing on this discussion. Bringing it up only shows more lack of knowledge.
I think by now everyone knows why georgehifi does not address specific points made by atmasphere and I with logical, and numerical arguments of his own, using his own words. I think we know why he avoids addressing things like Technics using DSP to compensate for phase. It would be nice if he didn’t repeatedly in every Class-D thread have to tell us why.
George will never "prove" anything. He is full of conjecture, but not a thought out fact in site. Not one word or technical analysis of his own. Just tired stereophile links for data does not indicate that he thinks it means or he tries to use one example to imply all must be like that.
Hey georgehifi, still waiting for you to tell us why all those amps you idolize have tons of feedback and it is okay. Still waiting for you to tell us why Technics must use digital filtering techniques to compensate for phase-shift in their amplifiers.
I should point out that any amplifier designer who knows his stuff of which you know who is not, could create an analog filter on the input that creates an inverse phase-function to compensate for the amplifier phase shift. This is essentially what Technics does with their GaN amps, but they do it in the digital domain.
queue deflections and insults ....
Well put djones51
The time coherence is w.r.t. blending drivers at the cross-over point, not across the whole output. If you ever looked at electrical phase versus acoustic phase you would see that what you describe is a rarity in speakers if at all. One simply needs to look at the impulse functions for that. However at the crossover point it is important so you don’t have peaks and valleys in the response. Turns out our ears/brain have never been shown to be that sensitive to phase, not surprising when you look at the physical structure, but I digress.
It does appears that the phase shifts almost 70 degrees, but if you understand LC filters and output impedance, you would know this would not be possible as the L would make for a very high impedance, the C a low impedance and you would get a much faster drop in the response curve. To be honest, it is not completely clear whether they are looking at the phase of the output impedance or what. I know when you look at an amp that includes a "special" 1200AS you see a pretty clean square wave at 10KHz which would be impossible if the phase response was that variable over those frequencies.
and the BelCanto which I think has an older 1000AS, has a very good 1KHz square wave indicative of little phase-shift from 1Khz to 10-20KHz.
The 700AS has about 25 degrees at 20Khz, half that at 10KHz, so at cross-over frequencies fairly minimal. Just remember that people put super tweeters on top of their speakers and don’t align them to fractions of a mm .. but I digress.
georgehifi8,388 posts01-26-2021 10:25pmYou you don’t even understand what you asked, they utilized one graph instead of two just to confuse the likes of you
George telling other people "they don't understand". That is ironic.
I don't know, you would have to have some serious vision problems or some serious dogma problems to this this is an unrecognizable square wave .....
For one it was done at very low power, would be much worse at normal power
There you go georgehifi, illustrating that YOU do not understand how electronics work and how Class-D amplifiers work. Signal level will have 0 effect on this. There is something that will though.
And here is the distortion of a very good amplifier:
Makes your "good linear amp" look like a 1970’s discount Realistic Receiver.
And if you can't see that that 10KHz square wave is a quite good square wave, then you are even less "aware" of audio/signals than I thought.