Originally Posted by Bill Mac
Below is a link to a thread over at the Emotiva Lounge. The thread is discussing the addition of an XPA-5 to a Denon 4520. The interesting part of the discussion are the thoughts that with the addition of the XPA-5 that overall SQ is improved. In that certain aspects of music have better clarity and that the soundstage is improved. It has been a very civil and respectful exchange of differing opinions. My main question is that if the volume levels are matched between using the 4520's amp section and then with the addition of the XPA-5 how does the SQ improve with just the addition of the XPA-5. Also how can one really do an accurate comparison when it takes a fair amount of time to add the XPA-5 or remove it for the comparison. I also asked if these differences happened at low to moderate volume levels and several in the discussion said that they heard SQ improvement at these volume levels as well.
So I would like to ask those in this thread what could the XPA-5 be doing to the signal to improve SQ in both clarity of specific instruments and improve the soundstage? I'm under the assumption that an amp like the XPA-5 should be transparent and not add anything to the signal or alter the SQ of the music one is listening to. But maybe I'm wrong in that assumption. With that I would like to know what in fact an external amp such as the XPA-5 could be doing to improve upon the SQ of the amp section of the 4520. I also would have to believe that the amp section in the 4520 is quite good and not lacking unless driven really hard with challenging speakers. So any thoughts on this discussion would be appreciated
This sort of thing comes up all the time. Someone buys a new amp, adds it to their system and suddenly they hear 'night and day' differences. Typically, they hear things on their discs that they 'never heard before', 'improved clarity', 'better soundstage', 'better imaging' and so on. Often their wives/girlfriends hear the differences too ;)
The one thing that all of these reports have in common is that they are subjective. They are always sighted tests, always in uncontrolled conditions. In every branch of science where A is tested against B, (eg medicine) these conditions for testing would result in immediate dismissal and incredulity. Sighted tests mean that the well-proven phenomenon of expectation bias cannot be ruled out of any apparent 'result'. And of course when someone has laid down significant coin for a new amp, expectation is high. If the test was conducted blind, the question we need to ask is if the same differences could be heard? The answer, in every properly conducted blind test that I have read about is that people cannot differentiate between any two modern SS amps that are not broken and are working within their design parameters.
Another problem with the 'tests' of the type you report on is that it is also well known that volume levels need to be very precisely matched if the test result is to be valid. This is due to a well-proven and documented psychoacoustic phenomenon where 'louder' usually sounds 'better' to the listener, and indeed when the signal is louder, it may also be possible to hear some things that could not be heard when the signal was less loud. It is impossible for people at home, without suitable equipment, to accurately set the volume levels on each amp to +/- 0.5dB, so again their results have no real value.
A further complication in this sort of subjective evaluation is that the tests are usually conducted by swapping Amp A out with Amp B and then listening again. But for a test to be valid, the swap must be instantaneous, using some form of switching device. Otherwise, 'auditory memory' comes into play and, again, it had been well-established that human auditory memory lasts only a few seconds. As it is impossible to swap amps in that time, the test is again invalidated by proven scientific research.
Finally, the conditions in which the 'tests' take place are not controlled in any way. In addition to the points raised above, people will often move speakers, change other components at the same time, listen days apart, use random audio tracks for testing, not necessarily the same on each item under test and so on. Given all of this, it is likely that their 'tests' will indeed show variations in SQ - but these variations are not a result of the difference between Amp A and Amp B.
There are very good reasons that, in other fields of science, tests are done in controlled conditions, following well-researched and understood protocols. In medicine, for example, when Drug A is tested against Drug B or a placebo, the trials are always blind (usually double blind). The idea of telling the patient which drug he was taking is laughable.
Coming to the second part of your post - what can account for any differences in Amp A and Amp B that will deliver the differences alleged by the posters you reference?
If both amps are not broken and both are working within their design parameters, and not clipping, then the answer is nothing. In every properly conducted blind test that I have seen, the results are always the same: people cannot differentiate between the two amps under test with any reliability more than pure random chance. Often they 'hear' differences between amps when it is the same amp that has been 'switched to'!
If you look at the specs of modern SS amplifiers of the type we are discussing, you will see that in all cases they all have levels of distortion well below the threshold of human audibility. IOW, they take the input signal and output it in identical form, but louder. It follows therefore that if all amps take the input and output the same thing, but louder, there can be no audible differences between Amp A and Amp B.
So when could an amp make an audible difference? Well, if Amp A is underpowered for the speakers it is driving and incapable of driving them cleanly to the desired SPLs, then replacing it with a more powerful amp will solve the clipping issues and it will therefore potentially sound better than the original amp. Hence my qualification above where I say that the amps must be not broken and working within their design parameters. It is also possible that a certain speaker presents a very difficult load to an amp and if the replacement amp has better capabilities with low impedance loads, then again there could be a noticeable difference. HST, you are specifically referencing the amps in a Denon 4520 and an Emotiva XPA-5 and none of this is likely to apply to those two units, both of which will drive most speakers to very high SPLs without clipping. Bottom line: if the amp can drive the speakers you own to the levels you require and do so cleanly and without clipping, then swapping for another amp that can do the same will yield no sonic differences.
If the XPA-5 and Denon 4530 were tested A-B or ABX in a properly conducted blind test, I would eat my hat without ketchup if a reliable difference in SQ could be identified between the two, given the caveats I mention above. So, unfortunately, the guys in the other thread are indeed 'hearing differences' - but they are perceived and not real, the result of improper testing procedures (eg no level matching) or expectation bias (as would be expected in a sighted test). IOW, their results have no real value to anyone else.
Oh yes, one final thing - if they are claiming that their new amp has improved soundstage, then that is usually a good clue that they are 'hearing things'. Amps don't impact on soundstage - that is a function of the speakers and their placement in the room.
Disclaimer: if anyone wants to get into a never-ending discussion about the validity of ABX tests etc, saying that the tests themselves are not valid rather than the results derived from them, then pelase know that I will not respond nor be drawn into that discussion. It is a waste of everyone's time - the 'believers' will not change their mind no matter what the scientific evidence is that is presented. There are numerous other threads here on AVS devoted specifically to that sort of discussion and it seems to me to serve no purpose to derail this XPA-5 thread with all that.