Sound radiation of a cello, where does near field end? 2

[mkoijn]

I'm trying to measure the sound radiation of a cello by tapping the cello bridge with an impact hammer mounted on a pendulum, in the same direction it would be bowed and measuring the response with a microphone. My question is, how far must I place the microphone away from the cello to be sure I am in the far field? If I place the microphone too far away from the cello the signal becomes weak in relation to the effect of the modes of the room I'm working in (I can't afford an anechoic chamber)and you can only hit the cello bridge very lightly before damage occurs. Is it possible to calculate where the near field stops and the far field starts?, the first mode that lies in the playing range of a cello that has large amplitudes of in and out of phase components (and therefore canceling out effects in the radiated energy because this mode is well below the critical frequency) occurs at approximately 150 Hz, so the wavelength in room teperature air is around 2.25 meters for that frequency. The length of the cello body is approximately 74 cm. Is this enough information to calculate near field far field distances or do I have to buy an intensity probe and map the particle velocity vectors? A colleague of ours is measuring at a distance of 0.5 meters from the cello, but I think at that distance he could still be in the region where canceling effects due to in and out of phase components are occuring. ( but I'm not sure!) Any help would be greatly appreciated.

Best regards
finnigan

 

[GregLocock]

Excellent question. Lots of detail and you've thought of the right approach.

Unfortunately all that hard work has to answered by saying, no, you can't calculate it. In a semi-reverberant room there isn't really such a thing as the far field (where all frequencies would drop off at 6 dB/doubling distance for a single source).

However, if you measure the absorption of the room and use enough microphone positions you should be able to estimate the sound POWER of the source, and if you measure the input POWER mag(F).mag(v).cos(phase difference) then you should be able to get meaningful results.

As you say intensity is another approach but it is very difficult to do with transient signals.

Failing that a couple of wavelengths away may be good enough for practical purposes!



Cheers

Greg Locock

 

[GregLocock]

I forgot to add that as proposed you are measuring what we would call a P/F function. This is VERY directional as well, so you would need to be very careful in making a measurement along the intended listening axis. Cheers

Greg Locock

 

[MikeyP]

If you are using an impact excitation then it should be possible to see from the resulting time domain mic signal where the first reflection occurs. If the space in which you are carrying out the test is big enough then there may be enough time before the first reflection occurs to obtain some information about the direct radiation. Of course the room will still "load" the instrument to a degree but you should be able to elimiate most of the reverberent effects.

(a more rigorous way of doing this is to measure the P/F function - see greg's post - then carry out an inverse fourier transform to obtain the impulse response function, remove everything after the first reflection and then fourier transform back again to get the new P/F function)

Everyting I have read in this area suggests that the radiation patterns from the instrument are very complex and so it is questionable whether you will obtain any meaningful results without an anechoic environment (still worth trying though!). And then there is the effect of the shadowing of the sound field by the player ...

Good luck

M

 

[mkoijn]

Thanks for the replys Greg and Mike. You are right, from my measurements so far, (with microphone at 2.5 meters away from the instrument) string instruments are very directional in their radiation patterns. However as a listner would receive both directly radiated sound and reflected sound we've tried to measure at a number of points around the instrument to try and classify an instrument by it's "radiation pattern signature". Every instrument apears to be very different in it's radiation pattern. What we are trying to do is to compare the sound radiation of valuable old Italian instruments with our newly made instruments to try and see analytically the differences we hear between the instruments,

Could you tell me why it's difficult to measure sound intensity using impact excitation? (obviously a scanning method is out of the question). I was (am?) seriously considering buying an intensity probe as I thought this would also elliminate some of the problems we have with reflected sound blurring detail in our sound radiation maps.

I will try the method suggested by Mike to elliminate reflections by removing the reflected radiation response part from the impulse response function.

Best regards

finnigan

 

[GregLocock]

I knew you were going to ask that!

Before I do try and answer it, Mike's suggestion will remove the direct echoes, but (as he says) will not affect the difference in radiated noise due to the impedance of the room. I'm not sure how to estimate its effect, but it is a very big issue when testing loudspeakers. However, you may well not be worried about 1 or 2 dB differences (I suspect you are more interested in harmonic structure than amplitudes). You should probably try and read up on concert hall acoustics, they spend a lot of time balancing direct and indirect noise paths.

OK, why doesn't conventional 2 microphone intensity work well with transient signals? I don't know for sure.

Theoretically, of course, it does. In practice it is difficult to get good acoustic measurements of transients, you usually run into dynamic range and S/N problems. The two microphone approach relies on p1(t)-p2(t) to estimate the particle velocity, and p1(t)~p2(t) so it will be very sensitive to noise on each channel.

Having said that there is no need to buy an intensity probe /in the first instance/, just use two identical mics, calibrate out the phase and gain errors by comparing their responses in a nice free field broadband soundfield, and make some measurements using that setup. If you aren't worried too much about high frequency one good way to do the calibration (well, it is a bad way but it works) is to tape the two mics together with duct tape and then record a couple of hours of wandering around the office!

I have no experience of hot wire velocity systems, they eliminate the difference between two similar signals issue, but probably introduce other problems.


Cheers

Greg Locock

 

[mkoijn]

MikeyP, Thinking about you're suggestion more carefully I realise that the lowest resolvable frequency with this method is

1/(time of arival of first reflection - time of arival of direct signal)

Unfortunately for me, floor to ceiling is 3 meters in our workshop which means with microphone placed 2.5 meters away midway between floor and ceiling, lowest resolvable frequency would be 244 Hz, which is not low enough.

You are right that it's probably useless to try and measure stringed instruments outside of an anechoic chamber, however I must try! any other suggestions would be gratefully received.

Best regards

finnigan

 

[MikeyP]

Just done a quick search on "intensity probe hire" and the first result was this

http://www.gracey.com/

who will supply an intensity probe and analyser for 60 UKP per day (minimum 7 days). It sounds remarkably cheap to me and I don't know if that sort of price is available in your location. It could certainly be useful for a "try before you buy". A proper intensity probe MIGHT just enable you to get results from an impact excitation as the S/N properties will be much better than a home made probe. Perhaps trying Greg's method first and then comparing it with a genuine intensity probe would be the best approach to see exactly what you need.

Of course the alternative approach is to hire an anechoic suite (many universities and government facilities with anechoic chambers are glad of the extra income). I suspect this would cost more for 1 day than the hired probe would cost for 2 weeks.

M

 

[GregLocock]

Or wait for a sunny day and do it outside! You'll be lucky to find a cheap anechoic chamber that will go down to 150 hz, that is a pretty big chamber.

To be realistic - why not set the cello up in a concert room and put the mic at the audience position?



Cheers

Greg Locock

 

[maxh]

Sorry to change the line of discussion but, some of this sort of work has already been done.
Suggest you look at
"Sound Intensity" by F Fahy
ISBN 1-85166-319-3
Where this type of measurement is described with respect to a double bass.

This book also references the following paper
Tro.J. Sound radiation from a double bass visualised by intensity vectors. Report No. STF44 A83088, ELAB, Elektronikklaboratoriet ved NTH, N-7034 Trondheim-NTH, Norway, 1983.

Hope this helps
Maxh

 

[GregLocock]

Good old Frank. If he's done it then read what he says very carefully - he was singlehandedly (almost) responsible for the entire acoustic intensity 'revolution' in the UK.
Cheers

Greg Locock