It was actually very valuable insider's information from experienced designer who speaks of taste of food he ate personally.
Example?
What was improved, and what as the result was you getting worse?
Yes, it is directed at you. When you put some "better" component and got worse result, that means that the component is worse. Period.
It can have better look, or better reviews, or some better parameter, but something is definitely wrong with it. Like, not enough power to drive your speakers that were easily driven by an amp with ten time higher distortions, but on ten times higher output power. Or something else. But please stop repeating BS as if sound gets worse because better component exposes some other component that sounded better when something else was wrong.
My point is, if you are going to criticize by sweeping comments all participants of the thread at once, please find some arguments that are not so wrong and silly.
Well, if you are somehow taking offense to my point of comparing apples to apples when making valid correlations, and view my comments as criticizing ALL participants of the thread at once, and rudely attack me for my "BS", then maybe you have issues that I don't care to get involved with. I'd rather listen to better music. Later dude.
Talk to you later, when you are ready to answer my straight question, what was improved in sound quality, and what as the result was exposed in sound quality, that caused overall worse sound quality. Until then I know examples only how each improvement adds to the overall improvement and enjoyment. I heard such comments as yours, but they were made by sellers of crap that were suggesting to buyers to buy even more of other crap, after they complained that the component they sold sounds worse.
Last edited:
Class-D, Now his comments are starting to make sense........
In his defense (not that he needs one) he doesn't so much prefer class D if I've read him right. It's merely that, as a fortuitous consequence of trying to find the simplest topology for class D, when employed by Philips, he came upon a very elegant approach (dubbed UcD) which encloses the L-C output filter within the loop and does so making it an integral part of the self-oscillating system. And it's done in such a way as to significantly enhance the high-frequency performance, compared to just about all other class D topologies. But I don't believe that he would argue that class D is the ultimate way to go, by any means. It's merely efficient, and when that's a requirement you do the best you can.
In fact I once read that Bryston was asked by someone why they didn't pursue their own switchmode amps, and quoted Bruno's comments in response. That may be available somewhere on their site. As I recall he said in essence that one could do better with nonswitching designs from the standpoint of audio quality.
Brad
No problemo, to each his own is what I say.
Bear in mind that I am talking about an amplifier made to play music, not reproduce steaty state sine waves into a lab resistor. This means that its COTINUOUS power will never be over 25W/8 Ohms, to accomodate those transients which will come along.
Given that I have rather efficient loudspeakers, 92 dB/2.83V/1m, and that my room is small (app. 170 ft2), by the time I get to 5 Watts steady state, my window panes are swinging along with the music. By 10 Watts, I have SWAT people raiding me, and the neighbours are evacuating fast (I live in a condo). The truth of the matter is, a decent 50W per side amp will do just fine, and I have a collection of power amps. My trusty and beloved Marantz 170 DC from 1978, practically rebuilt and beefed up, is said to deliver 85 W/8 Ohms, and it just never goes over 2 Watts according to its power meters; allowing for their not too good a dynamic response (a fad in those days), even if that is actuallay 5 Watts, I am still way below its actual capability.
Attaching my Karan Acoustics KA-i180 integrated amp, rated at 180/250W into 8/4 Ohms is really overdoing it, but, as we both know, big sound is usually delivered by big amps, just as usually only big seakers will deliver a big speaker sound. Well, I got both.
So, as I see it, you really need BIG power in transients only, and what I have in mind is a REASONABLE REAL WORLD solution, not an example of a perfect amp.
To you, I would suggest you hop over here and I'll introduce you a friend of mine, Milan Karan, and his top power amp model. It weighs in at around 260 lbs and delivers 1,200W into 8 Ohms, and a hell of a lot more into lower impedances. A gift at around $50k. He has to ship it in a wooden crate.
Plus great food, great girls, an added bonus.
I will accept the food and the girls, but his amp would have to make 1200 watts @.8 ohm ( like
laboratory resistor) to tickle my fancy.... 
Exactly my thoughts when commenting ...
Why is this, from the addional wiring ..?
Yes unfortuantely i have had those amplifiers in the past and while robust (bryston for sure) there sonics , was well ...
I heard Halcro uses so much feedback, they had to make the amps taller than normal to accomodate .... No..?
I remember a lot of amps being unstable on them back in the 70's, many smoke bombs. From experience amps that are stable on such a load tend to have very good sonics and drive anything attitude. My MC3500 was one of the few amps that was tested on such a load back then and passed, not even as much as a bump ....
Back then we used to call it the fire in the hole test ......
i recalled ... Yes !!!
Hey now ... look at the memories .........
Thanks Brad,
My comment was not a put down for class-d, but was more an continuation of my previous comment many pages back about class-d types suggesting feedback is great , to the tune of using 120db or as much as possible...
I have tried class-d and it's not for me , but there are others amongst me who swear by them , yes they are deaf ...
Thanks for the link.
Why take one side of the fence? Take both sides, use any hard data you can find and your ears both to an equal extent.
Ideally I think one should listen first, and look at the papers after, if you can.
Also take wild examples into account, like a Skullcandy versus a high-end speaker. Now, is the paper clearly showing you the difference between the Skullcandy and the speaker? If not, there is something clearly not in the papers, it's only sensical, right?
Kindest Regards,
Nicky
Why take one side of the fence? Take both sides, use any hard data you can find and your ears both to an equal extent.
Ideally I think one should listen first, and look at the papers after, if you can.
Also take wild examples into account, like a Skullcandy versus a high-end speaker. Now, is the paper clearly showing you the difference between the Skullcandy and the speaker? If not, there is something clearly not in the papers, it's only sensical, right?
Kindest Regards,
Nicky
I have tried class-d and it's not for me , but there are others amongst me who swear by them , yes they are deaf ...
I tried class D back in 70'th, and it was for me. It was not for neighbors though, because they heard on their radios what I recorded on my tape recorder.
Sound quality was superb. Crisp, smooth, low noise. Wider bandwidth than with conventional HF bias mixed with signal current.
Yeah they sound like toobs .....
You never heard class D tape recording amp, did you?
I am still a newbie in most things audio, but am starting to form some overall opinions, which might be off track, while still trying to learn what is important to measure (and thus to design toward). Please feel free to point out anyplace where I might not be correctly targeted, in what follows.
Power supply was mentioned as number one consideration. Current slew rate and potential limitations due to capacitors were mentioned. Transient distortion was mentioned. These were gratifying to see, because I have been tending toward thinking that these are some places where differences could exist that could discriminate between systems with merely good or great performance and those that are truly exquisite, probably because of differences in the accuracy of their reproduction of transients.
Power supply: The "signal path" that actually makes the sound is through the power supply; specifically, the currents from and to the power supply. The small-signal path and everything up to the output devices is obviously just as important. But that's not what is driving the speakers. And the power supply seems to have taken a back seat because the focus often tends to be on the small-signal "signal path" and the rest of the active "amplifier" circuits.
One could have perfection up to and including the base or gate drive signals of the final output transistors, but ALL transistors are actually passive in their behavior, i.e. they are just valves, not pumps, just voltage- or current-controlled resistances, and if the transient current that is demanded (allowed, actually) is not available to flow through them exactly when and how it is needed, then transient reproduction will be imperfect.
So that would mean that the transient response of the power supply, as seen by (at) each output device, could be very important. Specifically, the decoupling capacitors' ability to accurately supply transient current demands would be very important.
I have been trying to put numbers and equations to that, in another thread, and so far it looks like it would not be difficult at all to end up with insufficient transient slew-rate capabilities, even just due to parasitics, particularly parasitic inductance.
So I suspect that could be a problem in real amplifiers and needs to be measured and needs to be analyzed during the design phase. i.e. It could be an opportunity to make a better-sounding amplifier than those who don't worry about it.
For those who are still interested: After reading part of the latest edition of Henry W. Ott's book, "Electromagnetic Compatibility Engineering", I realized that inductances in parallel will have a lowered total inductance, just as (similar algebraically, at least) parallel resistances have a lower total resistance, but only if there is no MUTUAL inductance. So one potential aid to implementing more-effective decoupling capacitance might be to use multiple parallel capacitors that do not share conductors, up to the point to be decoupled. If you use multilayer circuit boards then this might not be of as much concern. For the rest, it might be very important, i.e. more difficult to implement.
Then I thought, "Why not apply the parallel-inductance-reduction concept to the entire power and ground rail structure?". That would imply using multiple parallel conductors for each power and ground rail. Both the main smoothing caps (or most of them) and the decoupling caps could each have their own copies of the power and ground rails. Theoretically, we could then make the power supply impedance, as seen by each active device, as low as we wanted, just by using more parallel rails and decoupling capacitors (but actually limited, unfortunately, by geometry). If you use multilayer circuit boards, this might translate into distributing multiple power and ground feed points for the power and ground layers.
Even though I thought I already knew the answer, I modeled the concept with LTspice, using three 1000 uF smoothing caps, for the two cases of a) one conductor each for power and ground rails and b) three conductors each for power and ground (with one set for each cap), while also modeling the parasitics of all of the conductors and the capacitors, and the impedance vs frequency plot as seen by the load was significantly better with multiple parallel copies of each rail.
Again, my preliminary calculations appear to show that if one is just placing single decoupling capacitors per rail "as close as you can" to each active device, then even if the caps are sized correctly there is a significant probability that transient-current capabilities would not be sufficient for accurate transient reproduction, at least in the worst case, for some amplifiers.
Coincidentally, maybe, in that light it seems like paralleling a larger number of lower-power output devices should be able to help accomplish the same goals, and is probably already providing some of the same benefits to transient response capabilities and accuracy, when used, even if only accidentally. i.e. Even with typical topologies that are already used it could be partially implementing the multiple-parallel-decoupling-capacitor concept (although multiple parallel caps per output device might still be needed). And it should also lend itself well to implementation of multiple parallel copies of each power and ground rail, if needed.
Cheers,
Tom
Last edited:
Jan,
I have done so previously.
Bruno simply brushes the effects of "distortion of distortion" of as non-significant and excludes them from his math and figures.
He does that as a pure ex cathedra assertion (which incidentally is false) without any proof or even showing a first order effects assessment of the quantity and quality of these effects.
This means that as correct his remaining math is, it is utterly useless as it starts from false premises.
Instead of endless theorising Bruno could have taken an actual amplifier and actually measured it, to show what is going on.
Ciao T
By popular request, I have placed this article online here:
Linear Audio | Online Resources
(2nd article down).
Opinionated perhaps, but Bruno is one who shores up his opinions with facts and figures. If someone would not agree on something with Bruno, he would have to show the flaw in his reasoning or his facts and figures. So far, no callers
I have done so previously.
Bruno simply brushes the effects of "distortion of distortion" of as non-significant and excludes them from his math and figures.
He does that as a pure ex cathedra assertion (which incidentally is false) without any proof or even showing a first order effects assessment of the quantity and quality of these effects.
This means that as correct his remaining math is, it is utterly useless as it starts from false premises.
Instead of endless theorising Bruno could have taken an actual amplifier and actually measured it, to show what is going on.
Ciao T
@gootee
Tom, your point is well taken.
Speaking strictly for myself, some 20+ years ago, I decided that parallelled capacitors make better sense than a single mammoth one. Thus, by default, I will rather use 2 x 10,000 uF rather than 1 x 22,000 uF.
As a matter of fact, I'll use a triplet: 10,000//10,000//4,700 uF, or some such. They will all be placed before the voltage/current regulator, but no matter what I may do, the actual power amp board will contain 100uF//1uF//0.1uF set on each power line.
What I should also mention is that I will also use a power line filter, to clean up the mess coming in through the grid before it even gets to the power transformers. It is not a power line modifier or shaper or whatever, just a straight filter, and since I've been making and selling them since 2002, I won't go into detailed description lest I be accused of advertising. Its merits have been established over time - perfect it is not, but it does help audibly on most equipment. The point is, I pay very particular attention to my power supplies.
Years ago, I accepted 75 micron copper tracing on my epoxy boards as my standard, that's double the industry standard, not to even mention wide traces, etc. The problem of parasitic capacitance is not a new one. In my case, it has led to eliminiation of some compensation networks, which where found to be completely unnecessary bceuase the quality of the artwork (done by a friend who has been in it professionally for the last 30 odd years, having worked in a HF lab and the military - I am good, but he's WAY better) and the materials used is such that it simply doesn't need what it used to. And certainly much less than a standard industry product, although some will probably always be required.
So your points are well taken, and I'm reasonably certain that most people here are well aware of the potential problems. Poor layout and loose artwork will quickly doom even the best of designs.
Tom, your point is well taken.
Speaking strictly for myself, some 20+ years ago, I decided that parallelled capacitors make better sense than a single mammoth one. Thus, by default, I will rather use 2 x 10,000 uF rather than 1 x 22,000 uF.
As a matter of fact, I'll use a triplet: 10,000//10,000//4,700 uF, or some such. They will all be placed before the voltage/current regulator, but no matter what I may do, the actual power amp board will contain 100uF//1uF//0.1uF set on each power line.
What I should also mention is that I will also use a power line filter, to clean up the mess coming in through the grid before it even gets to the power transformers. It is not a power line modifier or shaper or whatever, just a straight filter, and since I've been making and selling them since 2002, I won't go into detailed description lest I be accused of advertising. Its merits have been established over time - perfect it is not, but it does help audibly on most equipment. The point is, I pay very particular attention to my power supplies.
Years ago, I accepted 75 micron copper tracing on my epoxy boards as my standard, that's double the industry standard, not to even mention wide traces, etc. The problem of parasitic capacitance is not a new one. In my case, it has led to eliminiation of some compensation networks, which where found to be completely unnecessary bceuase the quality of the artwork (done by a friend who has been in it professionally for the last 30 odd years, having worked in a HF lab and the military - I am good, but he's WAY better) and the materials used is such that it simply doesn't need what it used to. And certainly much less than a standard industry product, although some will probably always be required.
So your points are well taken, and I'm reasonably certain that most people here are well aware of the potential problems. Poor layout and loose artwork will quickly doom even the best of designs.
- Status
- Not open for further replies.