Just to make sure we have it straight, Nige - when you say "class A", do you refer to true, pure class A, or are you referring to it as UK manufacturers did once, calling them class A amps when in fact those were class AB amps with a higher than usual operation in class A (e.g. Musical Fidelity)?
Because, if you happen to have some 200 or 250 Watt power trannies, I really see no point in stopping at 4.5 Watts, they should be good for around 20W/8 Ohms of straight class A operation.
4.5 Watts seems to me to be the extreme opposite of the power hungry Americans, just as I feel they tend to overdo it (although I assume their living rooms are much bigger than in Europe) in having unnecessarily much power, so 4.5 Watts seems to me unnecessarily far too low.
Regarding offset - twist, rock and roll it any way you want to, in the end, you don't want it. ANY of it.
A speaker cone, to the best of my knowledge, is not a semiconductor you can swing to more or less class A operation. If the DC offset pushed it 2 mm, then you have 2 mm less excursion than you could and should have. In my view, you'd be tampering with something that should not be tempered with. DC will also heat up the magnets, thus influencing their overall performance in just the way it should not be able to do.
Just my 2 cents' (dollar or euro) worth.
Because, if you happen to have some 200 or 250 Watt power trannies, I really see no point in stopping at 4.5 Watts, they should be good for around 20W/8 Ohms of straight class A operation.
4.5 Watts seems to me to be the extreme opposite of the power hungry Americans, just as I feel they tend to overdo it (although I assume their living rooms are much bigger than in Europe) in having unnecessarily much power, so 4.5 Watts seems to me unnecessarily far too low.
Regarding offset - twist, rock and roll it any way you want to, in the end, you don't want it. ANY of it.
A speaker cone, to the best of my knowledge, is not a semiconductor you can swing to more or less class A operation. If the DC offset pushed it 2 mm, then you have 2 mm less excursion than you could and should have. In my view, you'd be tampering with something that should not be tempered with. DC will also heat up the magnets, thus influencing their overall performance in just the way it should not be able to do.
Just my 2 cents' (dollar or euro) worth.
Hi DVV . The bad boy class A is in fact a good thing . 5 Watts good , 20 W bad ? You end up with nothing worth having if going to 20 W . Forget valves as that is a different world . The Yamaha class A/AB was OK . I strongly beleive by about 5 watts most of us regardless of speaker are no longer listening critically . We are then into dynamic excitement . Equally to say that the under 1 watt performance unimportant is absolutely and categorically untrue . Thus 0 to 5 watts is critical . If saying 5 watts class A and 100W C we keep the power supply out of saturation . Now if that amp is as the Yamaha of old a pure class A setting of perhaps 20W I don't object to that . I suspect all and everyone will prefer the class A/C version . My feeling is class G probably is better than sliding bias class A . I recently used sliding bias for a valve amp . It measures great , not sure I will use it in real life . I just don't trust it .
I was thinking in the <1 watt dissipation area for DC offset .
I was thinking in the <1 watt dissipation area for DC offset .
ok, since this turned to be one of the most successful (read: huge) threads on diyaudio (next to cables make no difference it seems), I'll take the chance and ask before searching... (forum police please be gentle).
two things are bugging me.
1. amp output current capabilities
I've used a low resistance shunt to measure actual instantaneous output current with a real load (my speakers) with real signals (music). at what would correspond to ~100W RMS at max input level (pretty close to realistic SPL levels with classical). 83 dB W/m speakers, impedance magnitude doesn't drop below 4 ohm, phase pretty well behaved (can post measured data).
I have found nothing that would indicate current starvation. I expected to find observable flat portions where max output current is reached but nothing even close.
in fact the 6 amps max I have measured equal max output voltage/min impedance magnitude.
I've heard the claims of insane instantaneous currents. what kind of speaker at what power and with what music cause that? I remember Ottala et al mentioning high current being required under special conditions but I can't isolate those conditions.
what am I missing?
2. slew rate
it's a known fact that real music doesn't contain lightning-fast rise times and nothing remotely close to step signals can be found in it. I've looked at waveforms of very dynamic sounding music in sound editors and even at large scale orchestra "explosions" the rise slope is pretty slow.
still, high bandwidth and high slew rate are said to correlate with better sound.
is there any objective reasoning behind that? RedBook doesn't contain anything beyond 20kHz and my amp seems to be able to reproduce a CCIF IMD test signal (19k+20k) at 100W with reasonable distortion (intermodulation products below -80dB WRT fundamental). subjectively I'd say that it somehow seems to struggle on loud and crowded passages.
again, what am I missing?
two things are bugging me.
1. amp output current capabilities
I've used a low resistance shunt to measure actual instantaneous output current with a real load (my speakers) with real signals (music). at what would correspond to ~100W RMS at max input level (pretty close to realistic SPL levels with classical). 83 dB W/m speakers, impedance magnitude doesn't drop below 4 ohm, phase pretty well behaved (can post measured data).
I have found nothing that would indicate current starvation. I expected to find observable flat portions where max output current is reached but nothing even close.
in fact the 6 amps max I have measured equal max output voltage/min impedance magnitude.
I've heard the claims of insane instantaneous currents. what kind of speaker at what power and with what music cause that? I remember Ottala et al mentioning high current being required under special conditions but I can't isolate those conditions.
what am I missing?
2. slew rate
it's a known fact that real music doesn't contain lightning-fast rise times and nothing remotely close to step signals can be found in it. I've looked at waveforms of very dynamic sounding music in sound editors and even at large scale orchestra "explosions" the rise slope is pretty slow.
still, high bandwidth and high slew rate are said to correlate with better sound.
is there any objective reasoning behind that? RedBook doesn't contain anything beyond 20kHz and my amp seems to be able to reproduce a CCIF IMD test signal (19k+20k) at 100W with reasonable distortion (intermodulation products below -80dB WRT fundamental). subjectively I'd say that it somehow seems to struggle on loud and crowded passages.
again, what am I missing?
Last edited:
Where the man makes a mistake is in his too narrow view of class AB output stages, which have been pushed more towards pure class A.
At 50 mW-2W/8 Ohms, my own output stages have a distortion of less than 0.005%, as inside that envelope they are working in pure class A mode. At full rated power into 8 Ohms, 28.3 Vrms, their THD is less than 0.02%, 20 Hz to 20 kHz. But, in my view MOST important, their THD decreases linearily as the power levele is reduced.
I use around 130...150 mA bias current per trannie in a 4 pair output stage, which means that my pure class A range is approximately 2.9/5.8 Watts into 4/8 Ohms.
Therefore, implicitly, I agree with you that the first 5 Watts are the most critical. I reckon I spend around 98% of the time below the 5W/8 Ohm line with my speakers, in my room, under normal everyday listening levels. I have to really gun it to go further and by then, the sound is uncomfortably loud in my small room.
Obviously, if one's room is twice and more times the size of mine, and if one's speakers are less efficient and less well behaved than mine, more power will be needed, no question about that. With a really big room, and some speakers behaving badly, one might well need oodles of power.
However, Nige, remember that big sound can only come from big things. Big sound will come from big speakers, and that priceless feeling of an amp sounding like it has infinite power, even if it's only a sweet illusion, is priceless. 5 Watts per channel may sound great, but it will not sound as big as an equally well thought out higher power amp. In that respect, the Americans are right, despite a few oddties here and there belying their actually smaller power envelope.
Then there's the matter of properly handling low impedance loads, which bodes well for an amp's overall stability of operation under normal conditions. Otherwise you end up with a lab sample fo a small power amp which chokes on anything less than 6 Ohms of impedance, thus effectively ruling out something like at least 60% of all loudspeakersout there, and we haven't even touched the matter of efficiency yet.
No, sorry, I disagree with you overall. I agree that the first 5 Watts are exceptionally important, but limiting your output there is a zero option for me.
Let me give it a go, from my point of view.
Ad 1. Please take a look at, for example, B&W's sales literature. You should note there that they give (or used to) both the nominal impedance, as say 8 Ohms, and the minimum impedance, as 3.3 Ohms in one case some years ago. There is a reason why a speaker's impedance is referred to not as a line, but as a modulus.
You might also notice that quite a number of speakers have their worst case phase shifts at about the same point where their impedance dips the most. As an example, say the above B&W speakers behaves like that, and say its wort phase shift is -60 degrees (not frequent, but not unheard of). When you work it out, -60 degrees means double the current, and that at an impedance of 3.3 Ohms. In effect, that's asking the amp to deliver power into what is effectively a 1.65 Ohm load.
Of course, that is not likely to occur at full power delivery, so it's not as bad as it may initially appear to be, but is still taxing for the amp and its PSU.
Unless of course you are related to brother Wayne here, whose speakers have a NOMINAL impedance of something like 1.5 Ohms.
Did I get it right, Wayne? That kind of power throws even John here well into overtime labor.Ad 2. The key reason, in my view, why wide bandwidth power is good for you is simply because it reduces phase shifty in the upper spectrum of the audible range, meaning at 20 kHz.
A decent slew rate helps with possible sound abberations (e.g. LP "clicks", which can be lightning fast) to keep the amp fully stable even under not so common operating conditions.
Here's the "but": it's not at all the same thing HOW you get any of this. And here we have two camps. The first are the traditionalists, who believe in global NFB being the cure-all. If the amp's open loop bandwidth is say 4 kHz, after enough global feedback is applied, it's closed loop response may well be over 1 MHz
The second are those who believe, as Otala does, that it's more logical to solve problems right at the place where they occur, i.e. locally. First you get it as clean as you can using only local feedback, and the apply some global feedback, sort of as an icing on the cake. Here, John and I are the great low global NFB proponents, others also agree I think, but we have our share of the "fundamentalists" who love global NFB above all other options.
And we have the fun gun brigade, like Nigel, who do it all at once so it's hard to keep track.
But we love him anyway. One last thought. Some time ago, we had a discussion on slew rate and it seems we agreed that about 40 V/uS for power levels up to say 100 Wrms/8 Ohms was quite sufficient. In practice, this could well be more, but the point is, one is unlikely to actually need more. 1 V/uS per every peak volt of output.
P.S. That's a Romanian flag, right? So we're neigbors, I'm in Serbia.
yes, it is. don't be a stranger if you plan to visit.
and let's say I agree with the first part, but the second... the answer is really not objective
let's say I don't listen to vinyl.I read a Stereophile interview with Nelson Pass and he specifically said that he only found relatively large slew rates on the Ron Tutt/Jim Keltner Drum and Track disc and then only with rim shots. relatively meaning in the 0.5V/us order
and then he went on saying that Mr. Curl (I'm assuming the same person you're "irreverently" referring to as "John") advocates a 10x safety margin. and I was like "whoa, where did that 10 come from?" I'm really not disagreeing but I also fail to find an explanation for that.yes, it is. don't be a stranger if you plan to visit.
and let's say I agree with the first part, but the second... the answer is really not objective
I read a Stereophile interview with Nelson Pass and he specifically said that he only found relatively large slew rates on the Ron Tutt/Jim Keltner Drum and Track disc and then only with rim shots. relatively meaning in the 0.5V/us order
Ditto if you happen to come to Serbia.
Let me ask you this - what happens when you play 24-bit 192 kHz sound? Please don't tell me you don't listen to this either, and do not plan on listening to it in the near future.
Yes, "John" is indeed John Curl. Any "irreverence" on my part would be unforgiveable for two simple reasons: 1) John is my senior, meaning he is entitled to respect by defualt, not to even mention the fact that he has "made his bones" a long time ago, and 2) I respect and like John (although we've never met in person, but hey, lots of life before us yet), a rare combination for me, and I would like to keep it that way in his case and in case of most others here. If I thought differently, I wouldn't be here.
Regarding Nelson Pass' view of the slew rate, I beg to differ. I can agree that MOST musical material is not too taxing, but one doesn't make a product for "most" times. If you want what is not "most", and would like to test your system without electronically manufactured torture signals, try The Blue Man Group, fisrt CD, piece No.8. About 6 or 7 minutes into it, one man takes a 2 metre run and hits a drum with a radius of over 6 feet (app. 2 m). If your system can reproduce it to the point where your liver moves aside by an inch, it's a good system.
Obviously, I agree with John's views. Statistically, in most cases, we haven't had a disagreement yet, but it's only a matter of time.
The explanation you need is better safe then sorry. No designer ever knows what the owner of his product will play, and is thus forced to think in the worst of events. And that Nelson's 0.5 V/uS was where exactly? Because if it was in the incoming signal from a TT/LP. or even a CD player, do remember that it will be amplified, and amplitude is a critical factor of slewing.
Assuming a full peak voltage output at 20 kHz, which is most unlikely in the real world, into an 8 Ohm impedance requires just over 7 V/uS. In that light, 40 V/uS is not quite so outlandish as it may appear to be at first sight, that's about 6 times over the perceived worst case.
More will never hurt; too little may.
John Hello . We went through this some time ago , this was the question I never asked . If one had to rescue an amp for a friend and didn't want to get too involved would input bandwidth limiting be enough ? Beauty of that is the friend with virtually no knowledge could find a solution in his spare time . He would feel he broke into our world just a little . Lets assume this guy loves/understands music so is well equipped to be a human oscilloscope .
I remember about some NZ guy who broke land speed records at Bonneville using a pre WW2 Indian . His slew rate looked wrong ( engine ) . He tweaked his aerodynamics to the point of burning his leg . If you see the film it is far too good to be true ( I wasn't there so can not say , the records speak for themselves I guess ) .
Like everything in life, more is better, most is too much ......
Like hell!
Too much food cuases gout and obesity.
Too much sweets causes diabetes.
Too much spending causes bankruptcy.
Too much sex causes heart attack.
Thank you for saying that, John, that's exactly what I've been telling Nige. Also, he does it the wrong way around, he simplifies things which should be more complicated for better effect, and makes more complex things which should be kept as simple as possible.
For example, the second stage of voltage amplification is usually the one which does most of the work, and while you can use a simple transistor, it would be much better to use a proper cascode stage, or some such. Yes, it does make matters more complex, but for the same amplification factor its natural distortion will be lower to much lower, reducing the need for so much global NFB.
Not necessarily so, Nige. A smaller but better charged battery may be better yet. Also, it greatly depends on the quality and power of the starter, and here yopu can see a wild variance in what people use. Lastly, the matter of your alternator is not an insignificant one, as it is directly related to how full your battery is at any moment.
Some manufacturers use skimpy little alternators, which take forever to charge the battery, causing it to be really stressed every time you want to switch the engine on if you haven't driven the car for a couple of day.
Personally, I view car batteries in exactly the same way I view capacitors in an audio device. The first chance I get, I kick the supplied battery and throw in a Bosch or Varta silver based battery, using the biggest one I can possibly fit into the given space. I never regretted this small extravagance.
Yeah, that too. Remember what Will from Stratford said:
Familiarity breeds contempt.
Also, far too often, it's too repetative. Lack of imagination kills it dead.
Perhaps . I do think adjusting input bandwidth limiting would transform some amps . It would be a nice bit of entertainment . The tightrope between doing something good and something bad could be explored . No broken amplifiers to fix . Happy friend .
I have many friends who have hi fi that has become almost part of the furniture . Often the introduction of something new casts doubts upon the viability of the system . Usually at this point I switch FM radio on and we sit and listen . It sounds OK . I then proceed to modify the amp to accept the newcomer . How many pieces of hi fi have been chucked out because the matching was poor ? Only then for someone to say I preferred my old system except .... We can engineer like it is a Volvo motorcar , mostly we don't need to . One old Hitachi receiver I did this for was so transformed as to be summer and winter . So many little things the designer could not foresee just pushed the right way . Eventually it was given away to his son , sad as I would have liked to own it . It ran the dreaded Bose 901's .
- Status
- Not open for further replies.