Depends on who you talk to, Scott.
Here, the two current freaks are Wayne, who has to be since he owns speakers with a 1 Ohm impedance, and me, who has been poisoned for life by Harman/Kardon's philosophy of maximum load tolerance, meaning high impulse currents.
Although, and this is my favorite beef with H/K, while it's all very nice to quote VERY impressive "HCC" ("High Current Capability") figures of say 35A for a SEPP output stage, this means prescious little unless it is backed up by a clear explanation of how it was measured, and especially for how long. Judging from the quote specs, it's fairly obvious this refers to a very short time period, like 1 mS or so, there's no way you'll squeeze 35 Amps from a single pair of 150W output devices. Others make such claims as well, for example, Electrocompaniet quotes its power output capability into 2 Ohms for 1 mS as well.
I regard this as guided logic. 1 mS is simply far too short to have any true meaning, in my view. Even IEC's impulse power spec, lasting 20 mS, is in my view too short, twice that starts to become useful. The clicks and pops of an LP, which can cause tremendous instantaneous power requirements, are usally not high up above 100 kHz, so they last very shortly, rather they can be down low, well inside the midrange.
Overall, I agree with you that this must be taken into account. That it often isn't is quite simply due to the economics of the business and the manufacturers' reluctance to invest in power supplies and output stages and their desire to spend that money on shiny buttons and functions few of us need, or lately into metal case sculpture artwork.
Well, I'm intrigued that those curves aroused such comment! They looked pretty good to me, most amplifiers are much worse than this - for example, the Krell tested by Stereophile was quite atrocious in comparison.
The only thing I would take from the curves is whether the remaining distortion and noise is audible: if S/N is -90dB at 1W, driving 90dB sensitivity speakers, then the artifacts are at 0dB - therefore, they can't be heard ...
There are far nastier things causing audible problems, to be worried about - I'll leave it to others to build 'perfect' power amps ...
The only thing I would take from the curves is whether the remaining distortion and noise is audible: if S/N is -90dB at 1W, driving 90dB sensitivity speakers, then the artifacts are at 0dB - therefore, they can't be heard ...
There are far nastier things causing audible problems, to be worried about - I'll leave it to others to build 'perfect' power amps ...
Its relative , if distortion decreases as you increase power then the 1 watt is considered high , actually, if they posted the Mw output you would really see how high it really is . Amplifiers that have lower distortion levels in the mw area while increasing as power increases were acknowledged to sound more natural , especially on high sensitivity speakers , low feedback amps for eg ..
audio power amps with high feedback around a high bias "First Watt" Class A stage are typically "unmeasurably" linear at mW output too - so you've said nothing interesting
the "harmonic multiplication" math is often shown with 10% starting distortion, and shows worst behavior is with 10-20 dB feedback
start the math with <1% distortion in the loop and then work from >40 dB loop gain and the very 1st "multiplied" harmonic is way below the noise floor
Last edited:
A friend uses Quad ESL 63's and an Art Audio 300 B amplifier . I suspect the THD combined to be about 1% and 5% when loud . Tom measured the distortion of amps and speakers via a microphone . Generally it was very bad . Estimates of 20 % being typical . I have not been able to repeat the tests . I supect he is more right than wrong .
I use older Mangneplanars and often a mildly modified Quad 303 . I would assume my distortion to be < 0.1 THD when CD for the entire system if below 90 dB @ 1m . The really odd thing is both systems are alike . Both do not make a bad source worse . Both are supremely warm without obvious loss off HF .
The Quad is like the girl next door . She has always been there , I do not take her for granted . Lacking slightly in detail . I will soon finish my Hypex amps . I suspect the Quad will not be replaced . She gets prettier each year and was perhaps a very Plain Jane when I first knew her ( a long story and a little sad ) . I will be honest and say I wouldn't have looked at her when I was young , only poverty and sentimentality kept her in my house . Mildly modified is that I don't want to make her into my design . Some BDY 56's if I did ( Cricklewood Electronis have them , Early Naim NAP 250 ) . She has top grade 3055's and slightly better caps and is separately bi-wired with high grade coupling caps . Slight PSU upgrades ( bypassing ) . Bias and centre voltage . The bias is with discrete resistors , values taken from the pots .
I am feeling too lazy to calculate Dvv 130 mV bias . About 1,5 watt class A is it ? If so it is I guess a class A amp as best as human ears can tell ? Above that it is too loud . Someone proved to me these facts are almost universal . People buy speakers that need about 5 watts . All that changes is room size and that allows bigger boxes to compensate .
I use older Mangneplanars and often a mildly modified Quad 303 . I would assume my distortion to be < 0.1 THD when CD for the entire system if below 90 dB @ 1m . The really odd thing is both systems are alike . Both do not make a bad source worse . Both are supremely warm without obvious loss off HF .
The Quad is like the girl next door . She has always been there , I do not take her for granted . Lacking slightly in detail . I will soon finish my Hypex amps . I suspect the Quad will not be replaced . She gets prettier each year and was perhaps a very Plain Jane when I first knew her ( a long story and a little sad ) . I will be honest and say I wouldn't have looked at her when I was young , only poverty and sentimentality kept her in my house . Mildly modified is that I don't want to make her into my design . Some BDY 56's if I did ( Cricklewood Electronis have them , Early Naim NAP 250 ) . She has top grade 3055's and slightly better caps and is separately bi-wired with high grade coupling caps . Slight PSU upgrades ( bypassing ) . Bias and centre voltage . The bias is with discrete resistors , values taken from the pots .
I am feeling too lazy to calculate Dvv 130 mV bias . About 1,5 watt class A is it ? If so it is I guess a class A amp as best as human ears can tell ? Above that it is too loud . Someone proved to me these facts are almost universal . People buy speakers that need about 5 watts . All that changes is room size and that allows bigger boxes to compensate .
This is the very first comment of this type I have ever read anywhere.
Well, that's 130 mA (not mV) times 4 in my case, works out to about 2.2W/4 Ohms, or so. Enough, but not overboard, I think.
Last edited:
Misinterpretation, misunderstanding. You speak about noise and not about distortion. You need to learn to understand simplest measurements.
Do you have link showing such .....
Never heard or experienced that ...
Yep ........
If I may, let me say to the uninitiated that Mr Pavel Macura is rather well known and respected in Europe for his designs in audio.
I have heard only one of his line amps for a preamp, a friend made it for himself, and I must say it's one of the really good guys.
Good to see you here, Pavel.
I have heard only one of his line amps for a preamp, a friend made it for himself, and I must say it's one of the really good guys.
Good to see you here, Pavel.
Last edited:
Yes , how dare me ask for a link , absurd to say the least ...
This paper discusses the "harmonic multiplication" effect at length. You would need to manually treat the 10% or more open-loop distortion cases with their equations since most of the cases they show start at 1% or less.
http://www.its.caltech.edu/~musiclab/feedback-paper-acrobat.pdf
Both of these guys are strongly invested in the "hearing above 20kHz" stuff. If you read Jerry's CV on his Wiki page you might be surprised.
Last edited:
For that I would buy you a pint of cider and maybe two . This makes your idea clear . I would say with less currant more feedback would be a reasonable thing to do .
I use 100 mA with MOSFET's . It seems it is as Hitachi said about optimum . They are weird in being usable with zero bias ( circa 3 mA typical ) . Sort of washed out but almost OK .Contrary to Douglas Self's teaching gain doubling is not a major problem if bias is this high . He argues no amount of adjustment over comes the imprecise biasing of the MOSFET's . Equally setting it too high creates the muddy sound sometimes associated with MOSFET's . It could be said setting a bipolar amp this high is a deliberate attempt to have the worst possible amplifier if a technical believer . This ignores how we hear . I would suggest the 2.2 watts is already above the threshold and WE then become the MAJOR distortion problem in the system . Set it higher still to create serious heat problems . 2.2 watts is about right .
Over my old Quad 303 I would expect a more detailed sound from your amp ( the class A bit ) . The Quad is so dam near perfect that's all I would expect . Drive my Maggie's harder and say goodby to them . The Quad is very gentle with speakers . The Krell is not my cup of tea I should say . However I suspect it has always been rich people who have shown them to me . They seldom seem to get hi fi to work well . Jimmy Hugues's Krell was very OK .
Most amps measure like op amps I find . The first watt is a great story . Alas hard to say it tells me much . Respecting it is good sense , build to optimizes this ( a La Dvv ) . The Hypex has very good sub 1 watt . In fact it goes into noise before showing a great change . Noise is a storm in some ways when class D . It took me a month to build a test rig to see that . Also the switch-mode confused the problems . I am tempted to say don't use the switch-mode . I had to because space was limited . As you might remember I call amps like this Cake Mix . I have modified the Hypex and built systems around it . My Cake Mix uses free range eggs perhaps ? I haven't listened to it yet because some mechanical engineering needs to be right first . Christmas would be ideal . I solved the switch problem . Plan A done better .
Yes , how dare me ask for a link , absurd to say the least ...
Sparrow's nest, so forgive 100Hz multiples.
Attachments
134mW (you can read it from the image), 6.8 ohm.
I have digged out the board from the amplifier grave, just to show something. Normal is below 0.001% at 100mW. There are some Messiahs that like to speak about high distortion at low levels, these stories might have been true in the seventies or eighties. Or for a cheap underbiased design.
I have digged out the board from the amplifier grave, just to show something. Normal is below 0.001% at 100mW. There are some Messiahs that like to speak about high distortion at low levels, these stories might have been true in the seventies or eighties. Or for a cheap underbiased design.
The fact that the two are frequently presented as a single entity doesn't help matters. And, from my point of view noise is distortion, it is signal injected by the playback mechanism which is not part of the original waveform, the fact that it is largely random in nature is neither here nor there - distortion is distortion ...
- Status
- Not open for further replies.