thylantyr said:I have been told THD under 1% is inaudible
and even Mr. Pass has amplifiers rated for 1% THD.
D. Self's book focuses alot on lower distortion perhaps
for technical reasons, because as an engineer it's
the cool thing to do.....not necessarily for audible reasons.
This might have been said some 30 years ago. The naked true is unfortunately completely different. Because character of amplifier's distortion is very different from loudspeaker's distortion, as a result you hear the superposition of both. You should have a chance to hear amplifiers (must be perfectly stable) with different THD and IMD spectra covering some 0.001%, 0.01%, 0.1%, 1% etc. with one, two, three or more lines in the distortion spectrum and you will never say that 1% is inaudible. It is of course a question of an audio chain quality as a whole - the better chain the more you hear. Under some quality limit THD of 1% is really inaudible. But take a speaker with low distortion (like Wilson Audio Maxx, e.g.) and you will be very surprised.
The perfect test is to listen to the philharmonic orchestra. Complex music is very sensitive to any kind of distortion. What you do not hear with one sine signal is perfectly audible with complex music signal because of unpleasant THD and especially IMD products.
Re: further thoughts...
'feedback' in this context does indeed refer to global negative feedback, from the output back to a differential input, just like an op-amp.
One of the more, er, controversial topics in transistor amplifier design is the idea that negative feedback is a Bad Thing. This probably has its origins in valve amplifiers, where negative feedback was (and is) somewhat optional.
Anyway, some amp designs try to make the forward amplification path as linear as possible, and then use no global negative feedback. To do this really requires a class-A output stage, as the crossover region in class-B outputs is unavoidably non-linear.
So "non-feedback" amps are almost all class-A, but class-A amps aren't necessarily "non-feedback".
Hope that helps.
Cheers
IH
johnzimm said:
'feedback' in this context does indeed refer to global negative feedback, from the output back to a differential input, just like an op-amp.
One of the more, er, controversial topics in transistor amplifier design is the idea that negative feedback is a Bad Thing. This probably has its origins in valve amplifiers, where negative feedback was (and is) somewhat optional.
Anyway, some amp designs try to make the forward amplification path as linear as possible, and then use no global negative feedback. To do this really requires a class-A output stage, as the crossover region in class-B outputs is unavoidably non-linear.
So "non-feedback" amps are almost all class-A, but class-A amps aren't necessarily "non-feedback".
Hope that helps.
Cheers
IH
Is 1% audible or not?
The THD number is only a vague hint that something is going 'wrong' in the system, and gives a rough estimate of how big the effect is. What it doesn't tell you is what's causing the problem, and some problems may be more objectionable than others.
For instance, some things (e.g. valves, loudspeakers, human ears) have their nonlinearities at the signal peaks, and behave well with small signals. Other things (underbiased class-B amps, D-to-A converters) have their nonlinearities around the zero-crossing point; they behave poorly with small signals.
So, a THD figure on its own doesn't really tell you much; you need extra information (such as how it varies with signal level) to form a proper judgement.
Cheers
IH
The THD number is only a vague hint that something is going 'wrong' in the system, and gives a rough estimate of how big the effect is. What it doesn't tell you is what's causing the problem, and some problems may be more objectionable than others.
For instance, some things (e.g. valves, loudspeakers, human ears) have their nonlinearities at the signal peaks, and behave well with small signals. Other things (underbiased class-B amps, D-to-A converters) have their nonlinearities around the zero-crossing point; they behave poorly with small signals.
So, a THD figure on its own doesn't really tell you much; you need extra information (such as how it varies with signal level) to form a proper judgement.
Cheers
IH
Of course. In case you click on my www you will find THD spectral measurements beginning at 1W, in fact I do it from levels of 100mV and that is why I like class A with low distortion - it does not make sonic problems with small signals (sharpness, agresivity) that are quite common with AB class).
Hi,
Well designed Class A amps never sound aggressive with small or large signals.
Class AB amps suffer from x-over node distortion that's blatantly obvious when largish voltage swings are required.
This also explains why endowing these with brute force PSUs delays the phenomenon temporarily 'till they hit the limit and they ultimately do cave in under the load.
Applying this to tube amps works thanks to the fact that tubes clip ever so gracefully, don't expect this from a sandswitch.
Class A is "la classe royale" amongst poweramps and audibly so by a large margin.
Cheers,
Well designed Class A amps never sound aggressive with small or large signals.
Class AB amps suffer from x-over node distortion that's blatantly obvious when largish voltage swings are required.
This also explains why endowing these with brute force PSUs delays the phenomenon temporarily 'till they hit the limit and they ultimately do cave in under the load.
Applying this to tube amps works thanks to the fact that tubes clip ever so gracefully, don't expect this from a sandswitch.
Class A is "la classe royale" amongst poweramps and audibly so by a large margin.
Cheers,
I have made rod elliots "death of zen" class a amp, and must say that it has a superb sound...
Only thing to worry on class a is the heat dissipation, and so you need VERY VERY VERY VERY BIG heatsinks on the output transitors to keep the output transistors relative cool...
So class a runs pretty hot, and power consuming is "way over top" but the big reward is a very nice sound indeed...
Only thing to worry on class a is the heat dissipation, and so you need VERY VERY VERY VERY BIG heatsinks on the output transitors to keep the output transistors relative cool...
So class a runs pretty hot, and power consuming is "way over top" but the big reward is a very nice sound indeed...
Now class, let's not see just the same hands
JC-3 toshiba input jfets
hi,
i'm delighted to see this thread
came across this schematic a few days ago, it looks wonderful.
my thoughts on transistor replacements were:
2SK389 and 2SJ109 for the input fets
MJE15031/32 drivers (the ones MikeW lists aren't in my supplier's catalog)
2SC3281/2SA1302 or 2SC2987/2SA1227 outputs
toshiba drivers would be nice.. any suggestions?
i would use only one pair of output devices and bias for only 5W class A, so about 600mA. 12w dissipation per device..
Mr. Curl, what is your opinion of this design of yours?
Regards,
keyne
hi,
i'm delighted to see this thread
came across this schematic a few days ago, it looks wonderful.
my thoughts on transistor replacements were:
2SK389 and 2SJ109 for the input fets
MJE15031/32 drivers (the ones MikeW lists aren't in my supplier's catalog)
2SC3281/2SA1302 or 2SC2987/2SA1227 outputs
toshiba drivers would be nice.. any suggestions?
i would use only one pair of output devices and bias for only 5W class A, so about 600mA. 12w dissipation per device..
Mr. Curl, what is your opinion of this design of yours?
Regards,
keyne
keyne,
5200/1943 chips are nearly the same as 3281/1302, but newer and improved. I use them in my amps. Datasheets are available (try google). In case you have any problem with these datasheets, e-mail me and I can send you PDF files. Another advantage of 5200/1943 is that there are not so many clones and fakes as for 3281/1302. I use BD139/BD140 as drivers for 5200/1943.
Pavel
5200/1943 chips are nearly the same as 3281/1302, but newer and improved. I use them in my amps. Datasheets are available (try google). In case you have any problem with these datasheets, e-mail me and I can send you PDF files. Another advantage of 5200/1943 is that there are not so many clones and fakes as for 3281/1302. I use BD139/BD140 as drivers for 5200/1943.
Pavel
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.