Patrick :
I got the same crazy peaks and troughs you mention when using sine waves to
test speakers. I even got big fluctuations in readings if I stood within a
few feet of the speaker and moved my arms around. I thought these were
fluctuations caused by the room, so I gave up doing measurements.
So your pink noise and band pass filter method interests me. Do you feed
the pink noise through the filter then to the speaker, and measure the
result ? or do you put the filter on the output of the mic ?
By the way, since the original question was about equipment, what I use for
a microphone is the analogue output from a RadioShack 33-2055 digital sound
level meter. It has its own amplifier etc. Just connect an audio voltmeter
to the output. There is a mod on the web somewhere showing how to extend the
lower frequency response and remove the weighting to get a uniform frequency
response.
Thanks
"Patrick Turner" wrote in message
...
Fabio Berutti wrote:
Dear RATs,
a question posed by Mr. Turner about my new LS is now puzzling me: what
kind
of stuff is needed to test a loudspeaker? Is it possible to do it with
some
VERY cheap microphone and the sound card included in any PC? I don't
want
to spend a lot of $ in Bruel&Kjaer professional testing equipments....
Ciao
Fabio
I am presently using an electret mic insert purchased from
Hi Q products in Melbourne for aud $1.20 each.
I bought 10, and all gave the same flat response as per their spec sheets
for
that model of electret mic.
I fitted the insert into the end of a 12 mm copper pipe 300 mm long full
of
wool.
Its clamped to a mic stand.
The response is only flat from about 20 Hz to 8 kHz, and then rolls off at
6 dB
/ octave.
So I placed an RC emphasis circuit in the mic amp, and made the recovered
signal
flat to 15 kHz,
and after measuring many speakers, it must be about right enough to tell
me if
there
are drastic problems.
The test signal is pink noise, or white noise eq'd by 3 dB / octave across
the
band.
My home brew tunable 3 decade range bandpass filter for measuring bands
has a Q
of about 12
at any F chosen, so if centred on 1 kHz,
the poles are at 41 Hz each side of 1 kHz.
The filter and noise source and mic amp are in the one box
and use opamps to achieve what I want.
There are calibrated mikes out there which may cost $100, and would
probably be
better,
as would the use of a PC and program to analyse AF, 'Speclab' comes to
mind.
I do it the old fashioned way with a book and pencil with several takes in
the
room.
These can be averaged, and a resultant response line drawn.
The PC will be a lot faster.
The results always seem to co-relate to what I hear.
Drivers can be tested separately, then together, and the response of each
driver
plotted.
When the phase of a driver is reversed, the results show up in the
response.
Its real shoe string budget stuff what I have done, but its a lot better
than
nothing at all.
I have my eye on better gear, but the bills keep me poor, and I have to
save up
more.
Even with better gear, the room effects still give +/- 3 dB graph
undulations.
Don't ever test with pure sine waves; you get a crazy +/- 12 dB variation
with
maybe 50 peaks and troughs, and a totally meaningless outcome.
The more carpets and cushions, large bags of chopped foam in the room,
the flatter your measurements will be. And the better the music will be.
Its not as good as an anechoic chamber.
If most music cannot be improved by adjusting the amplifier tone controls,
then the speakers might be about right.
I said might be. The pink noise room test will find the
harder to hear other problems.
Patrick Turner.
|