loudspeaker measurement techniques.

Status
Not open for further replies.
Jump to Latest
dipolar

dipolar

Member
Joined 2006
#1
I have read and known a lot of techniques on how loudspeakers are measures.but have near fail results while implementing the same at home

I have a ecm8000(non calibrated), alternatively wm61 mic(just for the sake of it) both are powered by series of 9volt batteries and goes to audio card.
and a-weighted spl meter Parts express type.


I have used REW and some few trail-wares to begin with
I would like to know following.

let us say my mic is always at 1 meter from loudspeaker and 3db mic error is also considered.

1.What is an impulse measurement, how is it carried out and what setup and precautions are required.

2.what is a waterfall spectrum and what is its significance.(all I see is mountain-scapes)

3. How is room reflections obliterated? I get different graphs while placing at different places in room..

4.How is phase measured ?

5.What weighting is supposedly used to measure ?
 
#3
Thanks Ra7,

But Please do help me with basic idea.A humble request from DIYer.

I fail to visualize and interpret what a water fall spectrum is for.

I agree the softwares are doing it automatically but without some indepth knowledge I feel high and dry all the time.

Regards

PS: My question was not 'how to measure a loudspeaker' but for the under laying concepts being explained in laymans language.
 
Last edited:
#4
A frequency response plot shows how "linear" the response in dB is at different frequencies. But a loudspeaker should not only have a flat response but also stop emitting sound when the music stops. The water fall plot is one way of graphically represent how well it stops att different frequencies. If the speaker stops instantly the waterfall would be a flat drop like a wall.

This driver have a problem at 8 kHz it shows up as an amplitude peak but also as a ridge toward the wiewer, indicating that it has trouble stopping in this frequency range.



If I change the scale of ringing from milliseconds to number of cycles before stopping the high frequency resonanses is more pronounced as they have more cycles per ms. So here you can see how many cycles the cone keeps going after the signal stops, in the previous graph it was in milli seconds instead.


Sometime problems that are hard to catch with amplitude (frequency response) show up in time response (or harmonic distortion or intermodulation distortion). How any of these measurable abberations correlate with human perception of sound is an other matter.
 
Last edited:
#6
Dr.Boar,
Thank you ,thank you ,thank you.

It did help me somewhat.please continue.

I guess the source was a white noise? and measured by omnimic as I could depict from graph. if so, how is it confirmed that it is not some other kind of artifact unrelated to the driver?

The decay on viewer plane seen is it, post switching off signal .

What is 'cycles' here? (cycles/sec is frequency.)

Regards.
 
Last edited:
#7
Cycles is a way of normalizing the ocillations of the cone moving. 1 milli second is 10 cycles of a 10 kHz wave but only one wave of 1 kHz ( and 1/10th of a 100 Hz wave).
Waterfall plots of the drivers are made really close up as the measurement does not separate late soundwaves emitting from the driver and reflections, diffraction from edges or surfaces.

So at 1 cm mesurement distance you get the driver as such at 50 cm things like diffraction comes gradually into the result further out reflections from the room take over. In room measurements tells you how the speaker-room-listening/measurement position work. Good for taming room problem with shuffling speaker and listener around as well as fiddle with bass traps, reflection absorbtion/scatter. However, it tells little about the intrinsic performance of the loudspeaker. To measure the speaker you have to stop the measurement before all those reflections from the room reach the microphone, this is called gating and turning of the measurement at 5 ms is common. The reason is that common ceiling height is about 2.4-2.5 m so it hard do get all surfaces much further than 1.2 m away indoors. With sound traveling at a third of a meter each milli second it works like this: With the mic close to the speaker (suspended midway up to the ceiling and away from walls) the first reflection arrives about 8-9 ms later increasing the distanse to the microphone will reduce this value asymtotically towards 3-4 ms. If you do not get why, it is time to dust of Phytagoras tehorem, smile 🙂
 
#9
DrBoar said:
Cycles is a way of normalizing the ocillations of the cone moving. 1 milli second is 10 cycles of a 10 kHz wave but only one wave of 1 kHz ( and 1/10th of a 100 Hz wave).
Click to expand...
It sounds like an expanded scale,Am I right?

The reason is that common ceiling height is about 2.4-2.5 m so it hard do get all surfaces much further than 1.2 m away indoors. With sound traveling at a third of a meter each milli second it works like this: With the mic close to the speaker (suspended midway up to the ceiling and away from walls) the first reflection arrives about 8-9 ms later. increasing the distanse to the microphone will reduce this value asymtotically towards 3-4 ms. If you do not get why, it is time to dust of Phytagoras tehorem, smile 🙂
Click to expand...
I like geometry very much 🙂
what troubles me most is 'poles 'and 'zeros' and calculus:t_ache: I may never be able to understand them.


So an impulse of full audio bandwidth would be used here.
Can a normal pc with a soundcard be quick enough to shoot and catch those ?


I have measured frequency response with white/pink noise umpteen times,but the impulse measurement I wasn't sure of. I always believed that my windows pc may not be quick enough to plot it reliably.

I am referring from Linkwitz Lab website and his work,lot of things goes above my head.
 
Status
Not open for further replies.
Top Bottom