At 09.46 20/11/2006, Raimund Specht wrote:
>[snip]
>Another less expensive solution would be to attach our CM16/CMPA P48
>ultrasonic microphone
>(<http://www.ultrasoundgate.com/microphonesP48.htm>http://www.ultrasoundga=
te.com/microphonesP48.htm)
>to your Sound
>Devices 722 recorder. This microphone employs the same microphone
>capsule as the above computer-based UltraSoundGate system. The
>drawback of this solution would be that you would have no acoustic or
>visual feedback while recording inaudible bat echolocation calls
>(except of the recording level meter on your recorder). Another
>problem might be the limited bandwidth of the recorder (ultrasonic
>sound components above about 60...70 kHz might produce unwanted
>aliasing effects).
Hi Raimund, I would correct the latest statement.
According to the frequency response of the 722 and its a-a filters
(see http://www.unipv.it/cibra/res_techtest_uk.html) you'll get
frequencies above 60 kHz attenuated (-5 dB at 80 kHz, -10 dB at
92kHz) whilst the risk of getting aliasing is limited to frequencies
above 96 kHz.
Frequencies above 96 kHz (maybe harmonics of echolocation signals
below 96 kHz) will be attenuated according to the frequency response
(dark line) and reflected back to the range below 96.
The pink line going from 96 kHz (-12dB) back to 50 kHz (-70dB) shows
the maximum level of aliased components that may occur for input
frequencies higher than 96 kHz if recorded at 0 dB!
Thus if you record a 120kHz tone at 0 dB (unlikely to happen) you get
an aliased component of 72kHz at -30dB, but if you get the same 120
kHz at, say, -30 dB (more realistic) you get the aliasing at -60 dB.
The aliasing risk could be further attenuated by the frequency
response of the microphone and of the amplifier you use. If the
global frequency response of the ultrasonic receiver you're using
falls down above 100 kHz the aliasing risk could become negligible.
Further, aliasing of sweeps is clearly identifiable: the first-fold
alias of an upsweep is a downsweep. By the way, with the equipment we
are talking about multi-fold aliasing will never happen; this risk is
limited to AD converters without anti-aliasing circuitry like those
of, for example, most NIDAQ boards. The widely used NI DAQ 6062
requires additional expensive filters to avoid aliasing, or the use
of very high sampling rate to largely exceed the bandwidth of the
signals to be recorded.
Gianni
--------------------------------------------------------------
Gianni Pavan
Centro Interdisciplinare di Bioacustica e Ricerche Ambientali
Universita' degli Studi di Pavia
Via Taramelli 24, 27100 PAVIA, ITALIA
Tel +39-0382-987874
Fax +39-02-700-32921
Web http://www.unipv.it/cibra
Email
|