Best done with a swept measurement, to control output levels. Provision for level control of input and output is needed. I did it with stepped attenuators.
: Take a baseline freq resp measurement at the lowest level of interest. I used 0.1V RMS (drive level). That becomes the reference sweep.
: For each step (I used 2 or 3dB) increase the drive level and decrease the mic level.
: Subtract the reference sweep, and plot. No change = straight line.
: Repeat until you reach the upper level of interest. I typically used 10V RMS.
On a good system/driver, the lines will all overlap. On many of them you'll see level-dependent nonlinearities crawling out of the woodwork.
IMO, that's overthinking it. For me it was a go/no-go measurement. Drivers with significant nonlinearity were removed from contention for future use. You can't EQ out level-dependent nonlinearities..
Last edited:
Hi,
Assuming your reply was for me then please note that my point of view is that I don't think at all that in this thread philosophy has to do with somehow, but I'll not deepen here just because I think it has nothing to do with this thread and I would not quite like to have a further reply of the same tone that maybe brings us even further away.
However it could be even interesting elsewhere. Thanks anyway. 🙂
Back to the topic, an empirical acquisition is not a scientific acquisition, but it is necessary for science because if considered useful it can encourage scientists to find adequate explanations.
Without the curiosity, courage and experimentation of a few ones we all would never have had any scientific (and then technologic) progress.
But first it takes those who believe in it and then invest a lot of money in it and in research in order to earn many more money later...
IMO, of course
I wouldn't necessarily agree that a speaker's tendency to produce harmonic distortion affects it's ability to reproduce it... if that's what you're saying.
I suspect you are attaching too much importance to speaker harmonic distortion. I mean sure, you need to see it doesn't get out of hand and for that you need to understand it.
However Dr Geddes would likely suggest looking beyond harmonic distortion to things that matter more in a speaker.
Not exactly. I was sort of responding in a general way to a claim sometimes made by others to the effect that speaker distortion, given its relative magnitude, should mask any distortion from electronics. Can't say such has proven to be true IME.
Last edited:
Speakers don't produce crossover distortion, or higher orders of harmonic distortion which are more of concern.
Also https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6254101
Also https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6254101
Last edited:
Would you also buy amplifiers, DACs, CD players based purely on which has the lowest distortion?
I agree completely.
I am constructing one way fullrange loudspeakers with dsp assistance. The liberty you have to test and hear and get listening experience is an important help to construct a really good sounding loudspeaker.
And I made recordings of some music instruments just to play them in the listening room together with my loudspeakers. Because for fine tuning frequency response you need the feedback with the original sound. This concerns mainly the correct use of digital filters and optimization of the direct and indirect sound field.
If you do it right it is impossible to hear a difference between original sound and recording.
And believe me some days I think my loudspeakers sound bad. Then I put my test recording - I take my music instruments - and hear clearly that everything is correct but my daily listening fitness is tricking me.
You certainly need for doing this some measurement equipment. I guess without measurements you would need the rare talent of absolute listening capabilities.
Last edited:
It's hard to explain in a forum post, but to give an example, harmonic distortion in the right dose is something that some like and others not. Dr. Sean Olive made also a study for Harman to find out tendencies in personal preferences wich resulted in the infamous Harman Curve (that is not fla), that should fit most, but also in the most popular deviations of that curve. It's still a subject of research, where Harman puts a lot of time and money in. Earl Geddles was also busy with that.
But the factors that play a role in personal precerence are mainly frequency response, dispertion, harmonic distortion by low order harmonics and damping factor/dynamics they say. And all of those can be measured with a Kippel or similar systems. Who likes what in what ammounts is what we can't measure as it's subjective, but the factors that play a role are known. If you want to read more about it you will need an AES subscription to read the studies of Dr. Olive that are published there.
Ignorance is hard to file under arrogance. Ignorant people need to be excused + educated.
//
I have come to the conclusion that it could be easier to rank speakers/drivers quality on the basis of a set of measurements than listening tests
The main measurements ? FR of course ... Dispersion vs Hz ... Spectral Decay vs Hz ... and distortion
I have to say that distortions measurements are by far the least popular and i am honestly surprised because in my mind they tell a lot indeed
Maybe because most of the results on commercial speakers and drivers are very poor ? eye does not see heart does not hurt ?
yes of course Maybe the real problem is how the measurements are carried out However even with a pure resistive load they say something
Much better on a complex load simulation a real load The amp speakers matching is very important
In some cases i understand some amps even took fire ? i do not want that
Last edited:
Hi and thanks a lot for your valuable advice There are so many stories of blind tests giving unexpected results
Our senses are not limitless
I really appreciated your reply and I thank you for that.
I don't think I'll make that subscription that you suggest because I've a lot of difficulty reading from English while Google Translate, I don't know why, it seems to worsen day by day...
However if I may, I would just be a little more cautious on certain statements when they appear to be categorical instead of "possibilistical" (that's which does not exclude the possibility of compromises).
In principle I prefer the conditional way, at least until something is really demonstrated and universally accepted.
Thanks again.
What is really sure is that the object of all our efforts will then be presented to our ears.
The measurements are indispensable and extremely useful to try to make mistakes as little as possible in that presentation to our ears, but if you think the measurements are everything (ALL) then I don't agree because in my opinion they are not yet enough. although the technologic progress there is and is undoubted.
I think the real ending is not to know that your system is "perfect" in measurements, but that it is perfect for you.
That is, your PLEASURE in listening. 🙂
Hello, distortion is not that critical as you may think. Not every distortion ist audible if you get to a certain amount of high fidelity.
At "Hifi-Selbstbau" there is a tool (for members) where you can make with your headphones (good quality is a must) a blind listening test which distortion is audible and which not. Title is "distortion - how much is too much?"
https://www.hifi-selbstbau.de/grund...nmenu-70/239-klirrfaktor-wie-viel-ist-zu-viel
I derived from this webpage that if you build a loudspeaker the most important are the mid frequencies like 1 khz where the ear is most sensitive. At 5 khz and 100 hz - the blind listening showed that the ear is 10x less sensitive to distortion.
So your loudspeaker should be in the mid frequencies until 80 db of loudness less distorted than 1% - this is for K2. They measured also K3 and other distortion but only visible for members.
However this is a hint - what is audible?
Attachments
Last edited:
Hifi Selbstbau had a result:
Result: By consistent application of a standardized calculation rule, the "allowed" (because inaudible) distortion factor could be determined for a wide excitation level range for various distortion components.
The results correspond very well with the "rule of thumb", but also give quantitative information. In general, the following applies:
THD sensitivity is greatest in the mid-frequency range around 1 kHz at 100 Hz or 5 kHz the distortion sensitivity is about 10x lower for medium level ranges from 70 to 90 dB the distortion sensitivity is greatest
For "unbelievers" there is also a small program for the audibility of the distortion factor, which can be carried out as a blind test. For subscribers, there are a number of further links at this point.
Result: By consistent application of a standardized calculation rule, the "allowed" (because inaudible) distortion factor could be determined for a wide excitation level range for various distortion components.
The results correspond very well with the "rule of thumb", but also give quantitative information. In general, the following applies:
THD sensitivity is greatest in the mid-frequency range around 1 kHz at 100 Hz or 5 kHz the distortion sensitivity is about 10x lower for medium level ranges from 70 to 90 dB the distortion sensitivity is greatest
For "unbelievers" there is also a small program for the audibility of the distortion factor, which can be carried out as a blind test. For subscribers, there are a number of further links at this point.
I agree with that. What I found out building phase linear fullrange loudspeakers is that after doing fine tuning I emphasize the mid frequencies by a half decibel and I use not too much damping in the bass reflex enclosure. That makes a loudspeaker sound a little bit more vivid than nature is. However listening tests with dsp showed that 2 decibels too much in the high frequencies (over one to two octaves) will drive you crazy on the long run. And for a certain listening position or room you always can adjust the bass right to your taste - the limit I found out is plus minus 2 decibels are tolerable. But it makes your loudspeaker sound "soft" and "warm" or "dry" and "tight" in the bass.
If I put my frequency and time coherent / linear loudspeaker fullrange loudspeakers in a room close to the back wall with a low roof with early reflections from the ceilink - all information concerning instruments in space and colour of instruments was lost. In such a position every loudspeaker would do. So usually well away from all walls the recorded information was good to hear. Maybe in this listening position a FIR filtering could help but I never tried their effect.