Useful reading for those who refuse to recognize that this is one of the principal audible differences between high output impedance amps compared to low output impedance amps. Thanks for highlighting this.
14 responses Add your response
In consideration of this article, here is something to keep in mind:
The question essentially is: is the speaker intended for this kind of amp or isn't it? Conversely, is the amplifier intended for this kind of speaker? It all depends on which paradigm you subscribe to/which paradigm the designer(s) subscribe to.
This might seem like a complicated way to say: component matching matters. But it is also a tool for being able to sort some of that out before doing any auditions, so you can waste less time and money due to mismatches.
General Electric's tests conducted in the 1960s, amounts of only hundredths of a percent were found to be audible and irritating to the human ear.
Is there a reference or source for this research? I am not questioning the research - just purely interested from a self educational perspective.
Hundredths of a percent odd harmonic distortion is of the order of Total Harmonic Distortion specs for many SS amps and far lower than what I understand any loud speaker is capable of achieving (except for headphones). Although I do understand that the concern is about very high order odd harmonic distortion (5th and up which may be rare in mechanical devices anyway). I understand that push-pull designs with very large amounts of feedback tend to eliminate even harmonic distortion by virtue of their design and have increased risk of odd harmonics - especially when over driven. Just curious to understand when these issues become most manifest.....
The ringing induced by a negative feedback loop will cause trace amounts of odd orders to appear at almost any level and with almost any signal. Our ears are so sensitive in this area that it is really easy to hear.
I read about the GE study some years back and seen it referred to several times since. GE has done a *lot* of studies over the years! and that is a gross understatement. I'm interested in re-reading the study myself, if there is a way to sift through the thousands of studies that they have done since then!
There is an easy way to demonstrate the phenomena though:
take any functioning amplifier and a sine/square wave generator. Set the rig up so that a speaker is driven by the amp and a VU meter is available so the level can be observed. Set a 1000Hz sine tone at 0VU. Then change to square wave and decrease the level until you hear it sounding at the *same apparent* level. Then observe the new VU reading. Most people get to about -22 to -24db or so- less than 1/100th the power level!
Interesting. Your comments indicate that perceived loudness is strongly linked to higher order distortion. This may be related to the way the ear/brain seems to sum up the overall energy across various frequencies in deriving an impression of loudness.
Bob Katz is very critical of the prevalent problem in CD mastering where a more square shaped audio waveform is produced as a result of hyper compression of the audio signal. Many modern pop CD's show extensive flat topping on the digital waveforms...generally thiis stuff sounds loud harsh and awful....many greatest hits CD compilations sound much worse than the original Vinyl.; this helps perpetuate the belief that Vinyl is better and digital is awful.
Hi Bob, Of course amps with a higher output impedance will have tonal coloration on speakers *designed for a lower output impedance*. In effect, that was the point of the white paper link I provided.
There are speakers that are *intended* for amps of higher output impedance. Remember the old Acoustic Research AR-1? OK- you may not be that old :) but that was the first acoustic suspension speaker ever marketed. It was designed to sound its best with amps that were 3 ohms or higher! -and it did.
Many horn speakers are intended for higher output impedances. Horn loudspeakers and full-range high efficiency speakers tend to be highly reactive due to their more precisely focused voice coil gaps and the other things that make them more efficient. As a result there is greater 'back EMF' from the speaker. Amplifiers with a low output impedance (Voltage Paradigm) often get that impedance from the operation of negative feedback; when that back EMF gets into the feedback loop it messes with the amp big time, resulting in enhanced ringing/odd-ordered harmonic effects that are not present with lower efficiency speakers!
So amps with a low output impedance often sound shrill on horns. Horns have been blamed for sounding shrill a lot, but it is the *interaction*, not the horn and not the amp. The interaction does not exist with high output impedance (Power Paradigm) amplifiers, and all of a sudden the horn is capable of sounding amazingly good!
ESLs and magnetic planers are another example. Both types have an impedance curve unrelated to driver or box resonance. These types of speakers are also well-known to sound their best with an amplifier that is on the Power Paradigm, as the speaker is expecting constant power rather that constant voltage out of the amp. Magnetic planers will often operate within either paradigm quite well as they have relatively flat impedance curves. ESLs, OTOH will sound rather shrill with lower output impedance amplifiers, as that amplifier will try to maintain voltage as the impedance of the speaker decreases. ESLs have an impedance curve that decreases as frequency increases; this causes a lot of amplifiers to make as much as 5 or 6 times as much power in the highs as they can in the bass; in effect they are rendered unable to play full bass impact on the ESL. This has given rise to the idea that ESLs have trouble making bass, which is absurd.
So, yes, there are speakers that will be tonally neutral with amplifiers that have a 'high' output impedance. It is a matter of intention on the part of the speaker designer. In effect, the Stereophile article is a bit misleading. If they had graphed the power output of the amplifiers in question, you would see an entirely different picture!
Does this clarify the issue?
Yes the amp ouput impedance modifies the response of the speaker and can clearly make audible differences. The letter is a fair statement if you assume that low ouput impedance amplification provides the most common reference for speaker performance and testing.
For example refer to this impedance plot
It is almost a certainty that a 3 ohm ouput impedance amp will modify the response of this speaker relative to a much lower ouput impedance amp (perhaps you could say vice versa too - a low impedance amp will modify the response compared to a high impedance amp). I expect the 2Khz to 3 Khz "prescience" region of the upper mid range would be audibly accentuated with the higher impedance amp.
In essence a high ouput impedance amp coupled with a speaker that has large swings in impedance will cause a modification to the speakers ouput with frequency; usually most noticeable differences occur in the bass where impedance rises at resonance...however, this example shows that an audible effect can also occur in the mid range.
What sounds better? Is this a good match? That is for the listener to decide. Some speakers are generally regarded to sound much better with a higher output impedance amp.
Shadorne's example of the Verity is a good one. In the frequency response curves the speaker is notably down at the same frequencies that a large peak in impedance occurs.
The speaker is well known for favoring tube amps, and now we see one reason why- it appears that with an amplifier of higher impedance, the 2-3KHz dip exhibited by many transistor amps would simply go away.
Another example was for years the Wilson Watt. It had a resonant peak in the tweeter, that was compensated by a band frequency trap that made for a low impedance at that frequency. Transistors exacerbated the issue by making more power into the low impedance and the speaker developed a reputation for a sort of brightness in this account. By using a tube amp with a higher output impedance, one that would not dump power into the low impedance, got rid of the issue.
Its all a matter of intention on the part of the designer, whether it be amp or speaker.
Nelson Pass has had some fascinating articles on this subject in the last few years.
Atmasphere and Shadorne, thanks for the comments.
Regarding Shadorne's example... If we are to believe the listening studies by Toole & Olive, then wouldn't we consider that speaker somewhat of a poor design?
It's interesting that picking an amplifier with a highish output impedance would improve the frequency response in the 2K - 5K region.
This seems to be a strong argument for active speakers -- let the designer have complete control over the amp/speaker system.
I need more time to ponder Atmasphere's comments.