[sdiy] semi ot: a little report on room acoustics and speaker measurements

[Czech Martin]

This is about room acoustic and speaker transmission.
I.e. natural echo and reverb.


This weekend had dry weather, even sunny. So it was time to measure the microphone/speaker system impulse response in as anechoic conditions as possible.
Someone from the list proposed to burry the speaker flat into the ground,
pointing to heaven, ideal half space condition (thanx!).

That´s what I did.

Few obstacles:

1.: measure how much input to the amp until amp distorts too much
2.: same for the speaker

This can be done using white flat noise having spectral notches,
i.e. bandstop regions. When these regions begin to fill in the microphone
recording then there is intermodulation/distortion.

Then I could´nt find the shovel, and hoe, so bougth them. Then I missed the microphone stand, so return again and fetch it. Glad I choose a place not far from home. It is a little river valey that means lots of stones! Damm it!
Needed 30minutes to dig this little hole in the ground.

But then the measurement could be started.
It is hunting season. So there was someone shooting as it was getting dark.
Felt uncomfortable! And the recording was also to be taken anew.


So it went on , from 18:00 to 20:00.
All  possible angles, burried and flat on the asphalt and with stand.

The results are quite good, but there are problems:

1.: the valley is polluted with infra sound, from the main road.
High frequency sound is absorbed, but the low stuff will travel for miles.
This scrambled the data below 80Hz or so, certainly wrong below 30Hz.
2.: the coaxial chassis has two notches in the on axis response, I thnk it is caused by the tweeter structure and perhaps woofer/tweeter interference.
Of axis it seems to be better.

>From 100Hz to 20000Hz the response is quite smooth.

Compared with a measurement in my living room the curves are much alike,
but the room response will smear and wobble a lot arround the anechoic.
And there is room gain below 80Hz.

Two ways to deal with the low frequency problems:
1.: make a new recording without environmental noise, perhaps find a better place in France, on this side of the Rhine there´s too much roads and railway.
Do it at night.
2.: the actual behaviour at such low frequencies is pretty much ideal, so ideal characteristics from TS parameters could be filled into the spectrum below fc.

And: the car was parked about 20m away. It was easy to see the reflection from this large object in the wave form. This makes 0.1 s travel time, but with this background noise the response was already burried in noise. So the response could be cut of before 0.1s. But a larger woofer could still be moving, so for larger chassis any object should be 100m away for safety.

Ok, this was to gain some experience, wait for better weather and do it again.
But before I have to test the S/N in actual rooms.

There is a nice pedestrian underpass with nice echos where I will try to
get some response. Due to the traffic above it is quite noisy (of course
also low frequency). I want to see if the 12cm chassis is able to exite that space enough in order that the impulse response can be extracted in a reasonable way.
Below 100Hz the response will fall with 12dB/oct....

If this is not possible I´d have to use a big additional woofer.

Most buildings in the city are drowned in traffic noise, the method must be able to work under these conditions. The underpassing is a worst case test.

There seems to be more interest in natural responses since ever more
convolution tools come up.

m.c.