The 20 dB figure for a room comes from scarce informations. The speaker shop in my town used to say that they measured a +18 dB peak in a client living room. And one or two readings on the web seemed to confirm this value as a maximum deviation, but I have no documented reference.
For headphones, we have the headphone.com measurments
. It has always puzzled me to see that a a given frequency in the treble, the Sennheiser HD600 was 15 dB below the Beyer DT990, which are two high end models.
I compared them in real life, and well, this seems to be right on the spot !
For speakers, I have seen some measurments here and there. It seems that good speakers manage to stay within +/- 2dB in a wide frequency range in medium frequencies. But in low frequencies, this is another story.
Amplifiers are nearly never measured under real load. Resistive loads do not account for their sensitivity to the speaker's impedance. But here is a serious study, with a real curve : http://www.soundandvisionmag.com/ass...rInterface.pdf
I have never thought of measuring the frequency response of mine.
I have experimented with speaker cables in Hydrogenaudio, with other forumers here :A collection of anti-hifi [ripoff] information - Hydrogenaudio ForumsWhy does the gauge of speaker wire matter? - Hydrogenaudio Forums
I can restore the pictures if you want.
I have seen or performed some RMAA measurments on various soundcards and CD players. The accuracy of most of them is in fact 0.1 dB, not 0.01 dB.
Here are some, with four different players : Hifi, Triangle, et Gamme ES [Topic] - HiFi & Home Cinema - Video & Son - FORUM HardWare.fr
Here is a vertical zoom on my CD Player's frequency response, seen from the analog input of a DAT deck (Yamaha CDX860 -> Sony DTC55ES) :
The level difference comes from the DAT stereo input volume setting. On one hand, the roll off is -1 dB at 20 kHz. On the other hand, the flatness from 20 to 2000 Hz is as accurate as 0.02 dB.
Last, for interconnects, I have measured a whole set of them, from the 2 € spaghetti to a 560 € pair of Tara Labs, with the same setting as above, and substracted the response from it, in order to show the difference between cables.
You can see all the results at the bottom of the account of our double blind test here : homecinema-fr.com • Afficher le sujet - RÃ©sultats du test en aveugle - cÃ¢bles de modulation
From the top graph to the bottom graph :
-The raw frequency response of the measurment setup, with cheapo interconnects (same as above).
-Difference with itself, measured again : shows the accuracy of the playback and recording process.
-Difference with itself, measured again after all the other measurments : shows the drift due to heating or other slow variations.
-Difference with an SB64 cheap soundcard output, just to show a standard frequency response.
-Difference between standard and ACR interconnects
-Difference between standard and Audioquest interconnects
-Difference between standard and DIY interconnects
-Difference between standard and DIY interconnects (unshielded)
-Difference between standard and RG179 coaxial cable
-Difference between standard and standard + 5 meters extension of standard interconnects
-Difference between standard and Tara Labs interconnects
-Zoom in the treble end of the raw frequency response of the measurment system.
Interpretation : the one pixel deviations can be assumed to be rounding errors in the measurment software. The high frequency peaks occur at a very low output level according to the raw frequency response, thus can be caused by differences in the background noise from cable to cable.
Thus the only significant deviation that can be associated with interconnects is the 0.02 dB roll off that occurs at 20 kHz when 5 meters of cheap extention cord is added behind the standard interconnects, which makes a total of 5+1 = 6 meters of standard cable, with a set of additional CINCH connectors in the middle.
Nota bene : contrary to a common belief, thin standard 2 or 3 euros interconnect are actually coaxial shielded pairs.
Nota bene : in french, interconnects are called "câbles de modulation".