For accuracy, the beta of the devices should be 100 or more. Temperature increase, you know. We only use high beta select devices.
bear said:
No.
For real-world crossover distortion (good design) with a sine wave excitation, if the amplitude of the sine wave is very small, it's not enough to fully exercise the crossover nonlinearity. The distortion percentage goes to zero as the amplitude of the sine wave goes to zero. As the sine wave amplitude becomes larger (but not too large), it exercises the crossover nonlinearity fully, yet the sine wave amplitude is still small enough that the crossover distortion amplitude is now a significant portion of the fundamental amplitude. As the sine wave amplitude gets larger still, the crossover nonlinearity has already been exercised fully, so the growing fundamental starts to swamp out the crossover nonlinearity, and the percent distortion decreases somewhat. Then as the sine wave amplitude becomes yet larger, the large-signal nonlinearity swamps out the crossover nonlinearity, so the distortion percentage grows again as signal amplitude increases. That's the reason for the distortion vs. power graph looking like it does for the Ayre amp. At very low levels, THD+N increases again, but that's just the effect of noise, not distortion.
john curl said:Dendrites! It's THEM, gentlemen!
As a metallurgist, that was absolutely priceless!!!
Exactly.
Below some minute level (for a low bias amp, or class "B" amp) the thing is effectively a class A amp...
My point was that once the xover distortion occurs (fully exercised?) it is always present, no matter how much higher the level of the primary signal... the percentage of the distortion component decreases as the main signal increases. Of course other distortions may add in as well...
So, if it is too low a level to be heard when the amp is cranking, then why worry about S/N or other matters?
Who is listening to 100-200watt amps at average levels in the 1 watt range?
What % is the distortion at the "1 watt level" anyhow? How many mV does that work out to be?
Now take it another way, are we then saying that if you run in "high bias" (what is that AB1 or 2 using our nomenclature?) of say 15 watts, do we never ever get to hear any of this "bad" distortion? That power level meaning that the signal will automatically mask the distortion? Does that mean that the best the amp has to do is what %THD? What about IM, too, doesn't really matter then?
Wait a second, should TIM not matter at all then? Since it occurs with higher level signals in the main? Or am I incorrect about that?
And, why would an amp that has a very high current supply that doesn't move much at peak outputs sound different at all on creshendo or complex passages if the distortions become masked at high levels compared to the same amp with a "softer" supply? Or is that not what happens?
Just trying to understand it myself...
_-_-bear
PS. I am quite aware of Dr. Geddes preferences in electronic hi-fi gear...
Below some minute level (for a low bias amp, or class "B" amp) the thing is effectively a class A amp...
My point was that once the xover distortion occurs (fully exercised?) it is always present, no matter how much higher the level of the primary signal... the percentage of the distortion component decreases as the main signal increases. Of course other distortions may add in as well...
So, if it is too low a level to be heard when the amp is cranking, then why worry about S/N or other matters?
Who is listening to 100-200watt amps at average levels in the 1 watt range?
What % is the distortion at the "1 watt level" anyhow? How many mV does that work out to be?
Now take it another way, are we then saying that if you run in "high bias" (what is that AB1 or 2 using our nomenclature?) of say 15 watts, do we never ever get to hear any of this "bad" distortion? That power level meaning that the signal will automatically mask the distortion? Does that mean that the best the amp has to do is what %THD? What about IM, too, doesn't really matter then?
Wait a second, should TIM not matter at all then? Since it occurs with higher level signals in the main? Or am I incorrect about that?
And, why would an amp that has a very high current supply that doesn't move much at peak outputs sound different at all on creshendo or complex passages if the distortions become masked at high levels compared to the same amp with a "softer" supply? Or is that not what happens?
Just trying to understand it myself...
_-_-bear
PS. I am quite aware of Dr. Geddes preferences in electronic hi-fi gear...
bear said:
I have no quibble with his work. I just think he is more interested in saying THD does not matter than, as some here, proving how .001% 7th is just what makes an amp "unlistenable".
I think it's ironic that when he first posted some THD pictures of one of his examples of a "bad" cheap amp there were a huge number of mains hum intermods in the plot. The amp clearly had more than one problem.
john curl said:
That interesting. And what about current sharing? If trying to maintain only ~13mV across the emitter ballast resistors, you only need a small temperature differential across the heatsink of so many output devices to cause considerable (idle current) error.
Cheers,
Glen
bear said:
bear,
Sounds like we're pretty much on the same page here. Horrors! 🙂
Well, we don't when it's cranking. But when the music gets quiet, that's when noise matters. I don't think it's too important for a power amp, since the front end of the system sets the noise performance anyway. But for a phono pre we should care very much! That said, I'm not sure what you're getting at.
Everybody that's not a head-banger. See Bob's article in AudioXpress about the 2006 RMAF. That was where I met Bob originally. That article is here. In that demo, the average power on his specially built meter was reading 1-2W, even with surprisingly loud listening. Speaker sensitivity was about 88-89 dB re 2.83VRMS.
I'd agree with that. I very much like John's approach of a high-power amp with many parallel devices, each operating at optimum bias. Then for normal listening it's class A. This isn't the same thing as using fewer devices and overbiasing though.
I dunno 🙂. I don't subscribe to Scott's THD hypothesis, at least without modifying it such that the threshold is very strongly dependent on how many higher harmonics are present.
Maybe yes, maybe no. But if the amp is designed for ultra-low THD-20 at high power, TIM will be wiped out too. The myth that this distortion "hides" from conventional sine wave tests was dispelled long ago. Since TIM ends up being equivalent to sine wave distortion that increases as the square of the amplitude and the square of frequency, then large-amplitude, high-frequency sine wave distortion tests handle that nicely.
Hmm, lots of assumptions there. The worst PSRR of an amp is in the front end. If you run the front end from a boosted, regulated supply, then it won't see this variation, and it seems likely to me that your assumption of the audibility of this would be less likely to hold.
scott wurcer said:
Yes, but maybe the 0.001% 7th does make an amp unlistenable - but not because it is hideous 100% of the time (to speak generally), rather because it is audible at precisely the wrong time(s)? It would seem that under some specific conditions that 0.001% 7th along with other harmonics might indeed be quite noxious... there may be some discussion of that exact %ile figure.
How about at the magic 1 watt level? Then?
Indeed, I seem to recall that Dr. Geddes shows that a sufficiently loud fundamental tone masks a fairly wide range of upper and lower harmonics, the louder, the more masking range. Now in a typical speaker, the louder the louder the harmonics (distortions) and the greater the IM products. At what point do the newly produced harmonics reach a loudness sufficient for them to act like new "fundamentals" thus masking (in the case of higher harmonics) nearby lesser harmonics? If say the 4th got loud enough would it mask the 5th, 6th and 7th? How about if the 5th got loud enough would it mask the (lesser amplitude) 7th?
Might this be a mechanism whereby the LOWER distortion speaker might actually be more revealing of higher order distortions? Thereby the "better speaker" would actually "sound worse" than one might expect? Does this ever happen?
Back to the amp case... And there is the opportunity to suggest that if compared to an amp not exhibiting the same sort of (what shall we call it?) "distortion" and with appropriate non-masking source material and the rest of the signal chain being sufficiently whatever it is that makes it technically feasable to hear this that then one might not like the amp with the 7th at all?
Does Dr. Geddes' findings make anyone want to use a class B output stage if they have the alternative of a Class A, etc? Preamp or amp?
As far as what Dr. Geddes chose to use for his test set up, I will not comment, as I don't think that it speaks to the crux of the matter... I'm not interested in debating Dr. Geddes (I think he has made a significant contribution) or dissecting his work looking for flaws. I think that just the fact that he can justify the idea and as you say "say that THD does not matter" is important.
I think the question for us is what meaning(s) does this have in terms of design considerations for things like the Blowtorch and other things?
Still very puzzled...
_-_-bear
Metalman, we both know that 'tin whiskers' is a more accurate term, but how could we scare anyone with that? Clue them in, if you can, as they won't listen to me on this subject. Heck, they won't even 'Wiki' it.
andy_c said:
Glad you too arrived at the 1 watt idea... 😉
Perhaps the "nutters" with the silly SE tube amps don't look quite as wacked? (skip the headroom question for the moment...)
Have you or anyone else (bringing it back to the Blowtorch, and some issues we already batted about) ever built an amp or preamp and made variations upon the power supply? Especially merely changing regulators? Heard a difference? Why? How? Isn't the change made when swapping out "good" regulators way below this "threshold" of audibility?
"...are we having fun yet?..." - B. Crump
_-_-bear
Isn't that where the input stages get unbalanced
and all them odd order harmonics start to "rear their
ugly heads"
(blameless amp..ha)OS
Dunno, just made it up myself... 
Actually the bear's research has determined that it is more likely either 1.14141414141414141414 watts precisely, or else 3.14159 watts, but not both.
wait, maybe pi/2 watts?
Some have postulated that it is <sqrt -j> watts, but I dunno about dem fancy ciphers none...
_-_-
Actually the bear's research has determined that it is more likely either 1.14141414141414141414 watts precisely, or else 3.14159 watts, but not both.
wait, maybe pi/2 watts?
Some have postulated that it is <sqrt -j> watts, but I dunno about dem fancy ciphers none...
_-_-
anatech said:
Chris,
They are not exempt. They get more time to transition, because it is to expensive to convert to lead-free in just a few years. When I was still doing NATO logistics, this was a big issue. Imagine all those billions of dollars in inventory in military warehouses and repairshops, let alone current production.
So, the military (and medical people) just got some more time.
Jan Didden
bear said:
I would rather get higher 😉
In case of a push-pull class A and 1A idle current, we get 16W in a class A into 8 ohm, but 8W into 4 ohm and 4W into 2 ohm. For 500mA idle, it is only 2W into 4 ohm (and here would be the transition, see graphs hereabove). Now, the speaker's sensitivity plays its game. So I would say 700mA at least, for PP output stage; 1A better, or more if possible, but everything above 1A is very difficult to cool.
What amazes me is that I have 100's to thousands of working amps out there, based on the principles that I have put forth here, recently. For some reason, they work pretty well, and measure pretty well, too. In any REAL design, there is some compromise. Yes, students, one must bend a little too get the unit to work for the majority of applications. My emitter resistors were originally 0.1 ohm, BUT we found problems with thermal runaway, so we moved them to .15 ohms. The Asian builders wanted to use .22 ohm, but I refused. It would have made things easier to build, BUT it would have exaggerated the Gm doubling, so highly thought of by Bob Cordell. I also use 10 ohm base resistors and this also causes problems, BUT it also prevents spurious oscillation with difficult loads, a useful tradeoff.
However, IF one actually got a JC-1 power amp and measured it, you would probably find that the Class AB-2 setting to be slightly lower than the marketing department states, and almost perfectly tracking with the .15 ohm emitter resistors. Still, the high current position sounds better, for most critical applications, and I recommend it, because the signal will not often go though the transition with real music, and even so, the changeover is not very noticeable as far as higher order odd harmonic production is concerned.
However, IF one actually got a JC-1 power amp and measured it, you would probably find that the Class AB-2 setting to be slightly lower than the marketing department states, and almost perfectly tracking with the .15 ohm emitter resistors. Still, the high current position sounds better, for most critical applications, and I recommend it, because the signal will not often go though the transition with real music, and even so, the changeover is not very noticeable as far as higher order odd harmonic production is concerned.
john curl said:
This is important and understandable, because 0.15ohm Re would require a bit different idle than the 0.1ohm one.
I use 0.1ohm, but less pairs. Thus higher idle per pair. Of course, not produced in hundreds, regarding power amps. But preamps .... they were already produced by hundreds, hundreds of DIYers 😉
Isn't the point that the signal goes through the transition on every cycle at relevant levels? How do you know the efficiency or listening level of the user? What am I missing?john curl said:
- Status
- Not open for further replies.