Good morning everyone
I have always read with great interest and tried to understand the lab reports showing the results of measurements carried out on loudspeakers
With reference to the distortion parameter I see different approaches:
1) those who make measurements
2) those who make measurements but remain doubtful about their importance
3) those who do not make measurements at all and never discuss the topic
I wanted to have some advice on the subject Some opinions
Lately I saw a video of an old Infinity Renaissance series loudspeaker that I liked mainly for its form factor
I found praise online with reference to its remarkable sound quality
But I also found an old review in Italian magazine that confirms the excellent sound quality but caused me some doubts
https://www.audioreview.it/prove/diffusori/infinity-renaissance-90.html
In essence this magazine has developed a test in which is reported the max SPL at one meter obtainable at various frequencies with intermodulation distortion not exceeding 5%
Unfortunately the procedure is not clear
The graph that comes out (see below ) is quite disappointing especially at low frequencies even if a special woofer is used
Can graphs like this be useful?
If so, why are they an exception rather than a rule?
I have always read with great interest and tried to understand the lab reports showing the results of measurements carried out on loudspeakers
With reference to the distortion parameter I see different approaches:
1) those who make measurements
2) those who make measurements but remain doubtful about their importance
3) those who do not make measurements at all and never discuss the topic
I wanted to have some advice on the subject Some opinions
Lately I saw a video of an old Infinity Renaissance series loudspeaker that I liked mainly for its form factor
I found praise online with reference to its remarkable sound quality
But I also found an old review in Italian magazine that confirms the excellent sound quality but caused me some doubts
https://www.audioreview.it/prove/diffusori/infinity-renaissance-90.html
In essence this magazine has developed a test in which is reported the max SPL at one meter obtainable at various frequencies with intermodulation distortion not exceeding 5%
Unfortunately the procedure is not clear
The graph that comes out (see below ) is quite disappointing especially at low frequencies even if a special woofer is used
Can graphs like this be useful?
If so, why are they an exception rather than a rule?
It tells you if the speaker itself add to the output or not. If one think it is important I suppose depend on if one want to hear the original signal or have something added.... Now if you take away all distortion, would you hear the original... not necessarily or perhaps hard to judge...
... but plainly - would you like your speaker to obviously add something by itself to what you play through it?
//
... but plainly - would you like your speaker to obviously add something by itself to what you play through it?
//
Hi ! thank you very much for your very helpful advice
Moreover i wonder if i can establish the range of good operations of a speaker measuring its IMD vs Hz and SPL
i guess so
Once fixed the max acceptable level of IMD at 5% i can see that the speaker above has no bass below 70 Hz
I was very surprised because the special dual coil woofer is considered a very high quality unit
but the IMD measurements tell another story
And then i think again to the excellent listening reports speaking of a thunderous bass
What do people hear ?
or could it be that below a certain Hz the ear becomes so less sensitive that cannot perceive distortion ?
this last point is very interesting to me
of course not I would like an accurate rendition of the recording
Moreover i wonder if i can establish the range of good operations of a speaker measuring its IMD vs Hz and SPL
i guess so
Once fixed the max acceptable level of IMD at 5% i can see that the speaker above has no bass below 70 Hz
I was very surprised because the special dual coil woofer is considered a very high quality unit
but the IMD measurements tell another story
And then i think again to the excellent listening reports speaking of a thunderous bass
What do people hear ?
or could it be that below a certain Hz the ear becomes so less sensitive that cannot perceive distortion ?
this last point is very interesting to me
Last edited:
Leaving out any subjective thoughts on the matter, distortion graphs (or any measurement) is a data point. Alone, it may not mean much but with a collection of other data points, a person gets a better idea of the unit as a whole. Any data point can be used to prioritize a unit against others if one so chooses depending upon trade offs since it’s always about balancing trade offs.
Professionally, I would consider it bad practice to ignore any data point even if one does not act upon a data point. It’s also bad practice, typically, to base a decision on only one data point unless extreme prejudice is required for that particular trade off.
Professionally, I would consider it bad practice to ignore any data point even if one does not act upon a data point. It’s also bad practice, typically, to base a decision on only one data point unless extreme prejudice is required for that particular trade off.
The tolerance for HD / IMD / THD varies with frequency it seems. But I believe it's still there - it's just that we are som used to high HD in the bass. On higher frequencies, it kills the music - high levels males sound turn dull and grey which might not be what one expects...
//
//
Hi! thank you very much for your valuable advice What surprises me is that distortion measurements in lab reports are an exception
i understand that could be different views about how to perform these measurements
imho multitones tests can be very telling for instance
The weird things is that while they are pretty common for amps they are almost never seen on loudspeakers
In general it seems that tests routinely performed on any unit up to the power amps will never be performed on speakers
as i think of speakers as any other link in the audio chain i find this approach at least weird
Why speakers are treated differently ?
Measuring distortion to high precision in electronics is more straightforward, for the same reasons we discussed over here:
www.diyaudio.com/community/threads/one-humble-question-about-phase-coherent-speakers.420385/post-7855846
www.diyaudio.com/community/threads/one-humble-question-about-phase-coherent-speakers.420385/post-7855846
Hi thank you very much for your valuable advice
In the case mentioned above we have a speaker that has a powerful bass, an excellent woofer of particular construction with a double voice coil, enthusiastic reviews and then when measured it shows that it has a notable distortion at low frequencies
this confuses me because it makes me doubt that people either like distortion or they don't hear it at all
I see a distortion of 5% at 70Hz and 90dB/1 meter
I would expect more from a high end woofer
In the case mentioned above we have a speaker that has a powerful bass, an excellent woofer of particular construction with a double voice coil, enthusiastic reviews and then when measured it shows that it has a notable distortion at low frequencies
this confuses me because it makes me doubt that people either like distortion or they don't hear it at all
I see a distortion of 5% at 70Hz and 90dB/1 meter
I would expect more from a high end woofer
Alas, speakers add a LOT more than harmonic & intermod distortion to the output.
Harmonic & intermod is rather further down the scale in audible importance than other stuff. I have several cases where speakers with high measured THD SOUNDED less distorted than (good) speakers with very low measured THD in DBLTs.
https://aes2.org/publications/elibrary-page/?id=2476 This paper isn't only about Intermod.
Probably the most important audible 'fault' in a speaker is its Room Interface Profile. https://aes2.org/publications/elibrary-page/?id=3798
Hi i am attaching for reference the graph of a speaker with much lower distortion
https://www.audioreview.it/prove/diffusori/jbl-everest-dd66000.html
above all other things i am trying to understand which type of measurements (common ones aside of course) could be helpful in the drivers selection process that is a very initial step
Some designers even suggest a listening test with specific tracks This sounds to me a little unscientific and subjective
For some people all speakers sound very good Other ones are extremely picky
maybe there is no answer
https://www.audioreview.it/prove/diffusori/jbl-everest-dd66000.html
this is very interesting Could you remember one bad sounding speaker with a very low THD ?
i see
above all other things i am trying to understand which type of measurements (common ones aside of course) could be helpful in the drivers selection process that is a very initial step
Some designers even suggest a listening test with specific tracks This sounds to me a little unscientific and subjective
For some people all speakers sound very good Other ones are extremely picky
maybe there is no answer
It's so expensive and difficult to do DBLTs that we don't test bad sounding speakers. So we are comparing speakers which already sounded good on sighted tests.
Probably for drive units, the 'waterfall' aka KEFplot is the best indicator. But you have to interpret this and very few people can do this properly. Of course a drive unit is used to make complete speakers and how this is done, makes a HUGE difference to the sound.
I used to say the BEST measurements can distinguish bad speakers from good speakers. But an experienced listener can tell this in about 10s in "a listening test with specific tracks". What the best measurements can't distinguish is an EXCELLENT speaker from a good speaker. I don't see any evidence this Millenium to change my mind.
And the test signal for this listening test is important too. Pink noise is the most audibly sensitive to the stuff that 'waterfalls' show. But it is TOO sensitive. About 10-20dB more sensitive than music so would throw stuff out that would make EXCELLENT speakers for music
BTW, the waterfall on the JBL Everest is OK though not in the highest class. Of course this only applies above 1kHz as the usual waterfall is nonsense at lower frequencies. So it's probably a good speaker. Only a DBLT can show if it is excellent.
Last edited:
I read the german diy loudspeaker magazines like Klang und Ton and Hobby Hifi for years.
They test new drivers and show measurements of their designed boxes:
its quite normal for many loudspeakers to go up in distortion below 100Hz. Smaller drivers below 10inch/25cm easily get 10% distortion in the bass region at 90db and more.
The ear is 10 times less susceptible to distortion at 500hz and 5khz than at 1khz.
Read here
They test new drivers and show measurements of their designed boxes:
its quite normal for many loudspeakers to go up in distortion below 100Hz. Smaller drivers below 10inch/25cm easily get 10% distortion in the bass region at 90db and more.
The ear is 10 times less susceptible to distortion at 500hz and 5khz than at 1khz.
Read here
Last edited:
👍
hi thank you very much this is exactly what i was asking I was sure that instrumental tests are key in the sound quality assessment procedure
imho the quality of raw drivers set the max performance that can be obtained Clearly any part like an inductor in series with a woofer will alter its behaviour How much will depend on the inductor characteristics But if i can measure a woofer i can also measure the system woofer plus inductor
Unfortunately the interviews to designers are often intended to provide general informations
Instead when i hear statements like the cone material affects the sound i almost immediately dream of a way to measure this different behaviour
Human senses are limited by nature
no human can provide an accurate freq response for instance
thanks again very interesting In general there is always and ideal behaviour that translates in ideal measurements
like a flat FR a very clean CSD the absence of IM products in a multitones test etc. The ideality is the goal
Then real units approach the ideal behaviour more or less
yes i see now Not very clean indeed And i dont like it considering that this speaker should be quite a pinnacle in quality
moreover i would keep an open mind The fact that a measurement is intended for amps it does not mean that it cannot be performed also on speakers
The fact that the result on speakers can be much worse shows how critical speakers are
My interest recently has shifted completely to speakers Almost 100%
Hi thanks a lot this is very important I see many speakers today using small woofers usually with tower form factor cabinet
they must have some intrinsic limits for SPL below 100Hz as you say
coming from years of listening to small speakers i did not know what i was missing a lot This resulted clear when i listened to a more full range speaker
Even listening to a big PA 2 ways i felt like something was missing then i understood that it had no deep bass like most PA speakers
this is also very important and fixes a range that should be treated very carefully
my goal is to find a good driver to cover from about 300 to 3kHz properly it must be a cone driver clearly
its quality could make or break the overall result
We have likely heard many good speakers in our lives. Reality the distortion might have been anywhere from 2 to 6%
even higher depending on listening levels.
speaker types have a ideal bandwidth and the lower and often upper range is high distortion.
The impedance peaks dont completely define it, but usually a good clue where it is going to happen.
Following the usual practice of 2 or 3 ways reduce distortion by staying away from the upper and lower limits of driver types.
You do see lower cost speakers that will perform or outperform many higher priced units. Or often see brand specific choices made.
So they get lost it the cost as I say.
Otherwise so yes it is good to understand distortion data or more detailed distortion data.
And given the idea yes, some drivers test very very well for distortion. And will be assuring it is actual "HiFi"
as opposed to getting lost in the cost.
Good to review data not get lost in numbers of a driver you have actually heard.
So you can see the true relation of what data shows and in reality what it will sound like.
even higher depending on listening levels.
speaker types have a ideal bandwidth and the lower and often upper range is high distortion.
The impedance peaks dont completely define it, but usually a good clue where it is going to happen.
Following the usual practice of 2 or 3 ways reduce distortion by staying away from the upper and lower limits of driver types.
You do see lower cost speakers that will perform or outperform many higher priced units. Or often see brand specific choices made.
So they get lost it the cost as I say.
Otherwise so yes it is good to understand distortion data or more detailed distortion data.
And given the idea yes, some drivers test very very well for distortion. And will be assuring it is actual "HiFi"
as opposed to getting lost in the cost.
Good to review data not get lost in numbers of a driver you have actually heard.
So you can see the true relation of what data shows and in reality what it will sound like.
@ginetto61 I'm not sure that this sort of distortion information is all that useful. The 5% threshold is somewhat arbitrary, and a bit high in my opinion. When the power input results in distortion below 5%, is it at 4%, 1% or 0.1%? Can it ever be as low as 1% or 0.1%??? It's better in my opinion to know the distortion level as a function of drive signal. Why? This information let's you, the designer, decide what threshold is acceptable and then choose the band limits for the driver.
One very important distortion measurement is for tweeters. It is primarily the distortion profile that determines how low in frequency a tweeter can be used, assuming you can EQ the response to be acceptably flat in the lower part of its passband. There is really not another better parameter for choosing where (how low) to cross over to a tweeter...
One very important distortion measurement is for tweeters. It is primarily the distortion profile that determines how low in frequency a tweeter can be used, assuming you can EQ the response to be acceptably flat in the lower part of its passband. There is really not another better parameter for choosing where (how low) to cross over to a tweeter...
EDIT to above post: my apologies I did not understand that these measurements are of a LOUDSPEAKER and not a speaker driver.
In that case, yes, these measurements are useful to know how loud you can play the loudspeaker before the sound becomes "grainy" from distortion. For this purpose, the 5% level is appropriate.
In that case, yes, these measurements are useful to know how loud you can play the loudspeaker before the sound becomes "grainy" from distortion. For this purpose, the 5% level is appropriate.
