My loudspeaker measurements (work in progress)

[TimVG]

Trust me that I've tried to optimize the presentation within my time and location restraints

 

[Krunok]

@617 is not wrong. Please lose the holier-than-thou attitude.
Anechoic chambers cannot provide reliable data at the lowest frequencies. Nearfield or ground-plane can.
Blend those techniques with gated measurement for frequencies > 0.5-0.8kHz, and results comparable to anechoic measurements can be had.

Please lose that attitude yourself. ANSI 2034 (and spinorama) is not about sub-100Hz data as all graphs anyhow converge there.

 

[617]

Trust me that I've tried to optimize the presentation within my time and location restraints

As a rule of thumb, anything further than twice the baffle width or so is 'far field'. 50cm is far field for most speakers, at least according to the ARTA recommendations. I normally measure around 24" for small speakers to get the furthest room reflections.

 

[dc655321]

ANSI 2034 (and spinorama) is not about sub-100Hz data as all graphs anyhow converge there.

That doesn't sound correct.
Got any sources that corroborate a 100Hz limit for spinorama tests?
If this is so, I'd be very interested to learn why.

 

[TimVG]

As a rule of thumb, anything further than twice the baffle width or so is 'far field'. 50cm is far field for most speakers, at least according to the ARTA recommendations. I normally measure around 24" for small speakers to get the furthest room reflections.

24" is more than 50cm..

 

[TimVG]

Also you want to consider the wavelength of the x-over so that the combination can fully develop. 50cm for my purpose is good

 

[Krunok]

That doesn't sound correct.
Got any sources that corroborate a 100Hz limit for spinorama tests?
If this is so, I'd be very interested to learn why.

Does it really come as a surprise to you that those curves converge below 100Hz or you are just exepriencing a brain fart? Because all that directivity stuff gets super interesting below 100Hz, right?

 

[Krunok]

24" is more than 50cm..

He was obviously too lazy to check that before commenting. Same as suggesting gating before seeing IR.

 

[617]

Krunok, you're right on the money! I'm a lazy clown with a major attitude problem. I will refrain from sharing my paltry knowledge in this thread.

 

[dc655321]

Does it really come as a surprise to you that those curves converge below 100Hz or you are just exepriencing a brain fart?

Neither. Interpreted your statement wrong is all.

 

[Juhazi]

I don't want to argue which is better, semi-nearfiel fullrange or splitted measurement, but semi in a large hall is for sure easier and fair to see differencies. Soundstage uses them and say that they are unreliable below 80Hz, that's it - we must accept that there is some uncertainty and that we can't get perfect truth. One good point is that it sums port output automatically, but back-ported speakers loose a bit.

Smoothing is not important, raw data is ok for me, I can do smoothing in my head! Good point is that no smoothing shows edge interferences better, in critical 1-5kHz rang - but only in single measurements, not in averaged ones!

I am interested in distortion graphs, they tell me how stressed the drivers are. They can tell about cabinet resonance too. Absolute % numbers are not important. I can live without them, but when listening tests tell something about harshness and coloration, I want to know where xo points are and what slopes and how distortion varies per freq. They also tell how low the bass can blow without a subwoofer. It is easy to show ambient noise level too, just toggle that in REW.

 

[TimVG]

Since when has THD been correlated to statistical preference?

Like I stated. Mh graphs are not definitieve

 

[Juhazi]

Ok, you all can discuss statiscs and preferences here.
Bye!

 

[TimVG]

That's not the point at all. Shame It has to be this way, I outlined my measurement method from post #1. I do try to give a fair reprentation of each speaker I meausre.
In any case the laws of physics still apply. Resonances should be present throughout all of the measurements. Below 1000Hz they are not as accurate as a calibrated anechoic chamber, but I never claimed as such. I will see about adding very nearfield measurements of the individual drivers in due time.

 

[TimVG]

Some of my thoughts
For those of you that believe in speaker preference in relation to anechoic measurements:
Since human binaural hearing depends on our lateral set of ears, these lateral measurements are of more importance to us than the ones taken in the vertical plane (I did not say unimportant -*- less imporant). So the early reflections here, due to partially my own time constraints but also in relation to our hearing are performed on the horizontal axis. The front an rear wall reflections are missing, as are the floor and ceiling bounce - but since the lateral reflections are deemed more important anyhow, it is better then nothing from the manufactures.

 

[Krunok]

I don't want to argue which is better, semi-nearfiel fullrange or splitted measurement, but semi in a large hall is for sure easier and fair to see differencies. Soundstage uses them and say that they are unreliable below 80Hz, that's it - we must accept that there is some uncertainty and that we can't get perfect truth. One good point is that it sums port output automatically, but back-ported speakers loose a bit.

Smoothing is not important, raw data is ok for me, I can do smoothing in my head! Good point is that no smoothing shows edge interferences better, in critical 1-5kHz rang - but only in single measurements, not in averaged ones!

I am interested in distortion graphs, they tell me how stressed the drivers are. They can tell about cabinet resonance too. Absolute % numbers are not important. I can live without them, but when listening tests tell something about harshness and coloration, I want to know where xo points are and what slopes and how distortion varies per freq. They also tell how low the bass can blow without a subwoofer. It is easy to show ambient noise level too, just toggle that in REW.

ANSI 2034 is not aboout below 100Hz response, it is not about THD (especially not one measured with sweep when it is covered with noise so figures are totally deceiving) and not about smoothing as graphs are "smoothed" enough by doing averaging.

 

[Krunok]

Divided by and scaled in level. Substracting showed some very weird anomalies

This division is still bugging me. Can you please show substracted curves so we get the idea of "weird"?

 

[TimVG]

This division is still bugging me. Can you please show substracted curves so we get the idea of "weird"?

 

[Krunok]

I see.

I just checked with some of my measurements - you did it the right way using division.

Are spinorama charts available for any of the speakers you measured? It would be interesting to see the comparison, athough you index doesn't really correspond to any of those 2 spinorama DI indexes.

Also, can you please post IR of one of on-axis measurements?

 

[TimVG]