Sure, let's look at some areas that are missed by those measurements:
A) Asymmetric clipping.
Music isn't sin waves; A lot of musical waveforms are highly unsymmetrical. When clipped by a DC coupled amplifier, this produces significant DC offset at the output. So for example, a loud trumpet blast could push the speaker's voice coil half way out of the gap. As an aside, this is one of those things valve amps just naturally get right; the OPT prevents any DC or VLF from getting to the speaker.
Anyway, a bit of extra distortion from the speaker doesn't really matter, if it's only during very loud moments when the amp is clipping anyway. However there's a very easy, very popular way to make things worse:
A lot of amps use a cap in the feedback network which helps with DC offset and also adds a very low frequency pole to the response.This causes the DC offset to be corrected within 1/4 second or so after the start of the trumpet blast, but causes another large cone excursion after the end of the trumpet blast. So now we have unnecessary distortion from the speaker during a quieter passage, when the amp is not clipping.
It's possible to design an SS amp that has good bass response and rapid recovery from self-inflicted DC offset, but it's not the sort of thing you see every day. Most often a simple RRC network is used with a large time constant because "You have to set the pole at a very low frequency to minimize the signal voltage across the electrolytic capacitor, and you have to use an electrolytic capacitor due to the large value required" or some such circular logic.
Suggested test:
Start with a suitable test signal e.g. a 1Khz "square" wave with a 3 to 1 mark-space ratio, and play a 2 second burst of that through the DUT at a level of say 3 to 6 dB above clipping. Connect a 2'nd order low pass filter with the rolloff at about 40Hz to the output off the amp.
The output from the filter approximates the cone excursion one could expect from a speaker. Never mind the excursion during the tone burst, but look at the magnitude and duration of the excursion after the tone burst.
B) Asymmetry without clipping.
Consider a typical class AB BJT complimentary EF output stage. Normally a Vbe multiplier is used for temperature compensation to try to keep the bias "optimal". Just mounting the Vbe multiplier (or other temperature sensor) on the same heatsink as the output devices doesn't give very good tracking though, so a standard recommendation is to mount the temperature sensor on top of one of the output devices. This gives much better tracking. For sin waves.
A funny thing about asymmetrical waveforms (such as music), is that the power dissipation of one output device tends to be much higher than that of the other, so there will be a significant temperature difference between them. The question to ask now is whether your temperature sensor is monitoring the hot device or the cool one. Either answer is wrong, of course. Expect crossover distortion to fluctuate wildly, depending what's playing.
Testing:
We'll need a suitable test signal that causes differential heating, but doesn't contain high order harmonics. Perhaps equal parts fundamental and 1'st overtone, played at a fairly high level? The trick then is to look for a change in the higher-order distortions produced, when the polarity of the test signal is reversed.
C) Other issues with differential heating.
Some other devise parameters vary with temperature as well, so for example we can expect the current gains of the two output devices to be fluctuating independently while the amp is playing music (as opposed to test tones). This could cause distortion, especially in designs where the output stage is driven from a high impedance. I haven't thought of a fair test for this sort of thing though.
~~~~~~~~~~~~~~~~~~~~~~~~
There's a few other things I can think of that are missed by traditional measurements, but this post's too long already. More later, time permitting.
Hi,
But who control's the controller? (Sorry, could not resist)
I agree. Any test needs controls, both positive and negative.
My biggest gripe with the blind tests who's results are so often called upon to support "all amplifiers sound the same" is the combination of bad statistics and lack of any controls, which quite frankly make it very bad science at best...
Ciao T
But who control's the controller? (Sorry, could not resist)
I agree. Any test needs controls, both positive and negative.
My biggest gripe with the blind tests who's results are so often called upon to support "all amplifiers sound the same" is the combination of bad statistics and lack of any controls, which quite frankly make it very bad science at best...
Ciao T
A car has rather more long-term energy storage than a speaker. Even so, it you drove an electric car twice with the same terminal voltage on the same road under the same conditions I think you would find it would cover the same distance. All I am claiming is that the system is time-independent, not necessarily linear - so the same input produces the same output each time.CopperTop said:
As it is the amplifier we are changing and testing, not the speaker, then it seems easiest to look for changes in the amplifier output when loaded by the speaker or a dummy load. 'Back emf' or anything else will change the output voltage (unless the amp has zero output impedance).
An amplifier is a voltage source, with a finite non-linear frequency-dependent output impedance. The voltage, and the impedance value, have some relationship to the original input signal. The impedance may also be modified (being non-linear) by the output current drawn. That is really all I am saying. Same terminal voltage must mean same sound output, because the load is time-independent.
Don't you ever get tired of saying that? For the perhaps 10,000th time, there is no-one who has ever claimed that.
I take it as read that its being used as shorthand for the non-falsifiable assertions of those with an objectivist bent. Such generally include some qualifying words like 'competently designed' whilst being careful to avoid pinning down what the qualifying phrase actually means in practice.
Sy,
Please pardon my pithiness, I guess I should have written:
"All amplifiers that measure the same, similar or beyond certain minimum limits using basic traditional measurements (FR, THD+N, IMD, DF) sound the same."
with the corollary that:
"Amplifiers that sound different must be defective."
Past that I notice you have declined to answer my four little easy questions so far...
Ciao T
Please pardon my pithiness, I guess I should have written:
"All amplifiers that measure the same, similar or beyond certain minimum limits using basic traditional measurements (FR, THD+N, IMD, DF) sound the same."
with the corollary that:
"Amplifiers that sound different must be defective."
Past that I notice you have declined to answer my four little easy questions so far...
Ciao T
I thought it was just erecting a straw man, to divert attention from the non-falsifiable assertions of those with a subjectivist bent. Such generally include some qualifying words like 'musical' whilst being careful to avoid pinning down what the qualifying phrase actually means in practice.
well, I discouvered what it really is, the other day
its when you hear the drame in the music
typically classical music
isnt that exactly what you experience at a live concert
and what often is missing in reproduced music
that is if you can hear it, or think it matters at all
tho its a speciality of classical music, its clear that its positive to other kinds of music as well, ofcourse
resulting in what you might call musicality
that to me is the genuine musicality, as opposed to the 'faked' kind
with non acoustic instruments and amplified music its harder to tell whats up or down
but the difference is still easily heard
So then if 'musical', as an adjective applies to a percept (the experienced sound) and 'competently designed' applies to an object (an amplifier) then my and DF96's statements lack a vital symmetry.
Apparently you guys never get tired of having the same argument over and over again, so I've created a dedicated thread for it here: http://www.diyaudio.com/forums/everything-else/204360-pointless-argument-thread.html
Perhaps one of the mods will be good enough to move all posts in this thread, from SY's last post onwards, to the new thread?
Perhaps one of the mods will be good enough to move all posts in this thread, from SY's last post onwards, to the new thread?
Which is still an inaccurate summary of the engineering viewpoint. Ask questions with false premises and there's indeed no point in answering (thank you, Godfrey!).
- Status
- Not open for further replies.