This is index.rtf in view mode; [Download] [Up]
I have used their microphone with the line-in jacks. I would stream the signal into a file, and then bring the file into Mathematica for analysis.
For Mma 1.2, the following worked:
signal=ReadList[sndfile,{Byte,Byte,Byte,Byte}].
Then the ith datum for one line jack is
256*signal[i,1]+signal[i,2],
and for the other line jack is
256*signal[i,3}+signal[i,4].
(But Mma 2.0 has better import commands, so things may be easier.)I tested this with a good sine generator. Using a Mma non-linear fit package written by:
Pekka Janhunen, Finnish Meteorological Institute,
Geophysics Dept., email: pjanhune@finsun.csc.fi,
(thanks again, Pekka), I got fits like:
dc-offset = -524 +/- 23 counts
amplitude = 26523 +/- 35 counts
frequency = 99.84 +/- 0.12 (expected: 100.000 00 Hz),
with residuals that were approximately gauss-distributed, ranging from about -20 to +22, and with a std. dev. of about 6. (Sorry --- I haven't really carried out a proper analysis here). Using 6 counts as the noise figure and 26523 as the signal figure, we get 20*log(ratio) = 72 dB as the signal to noise figure. This is some 22 dB short of the theoretical figure corresponding to 1 bit in sixteen. I don't know where the missing 22 dB went. A good part may have been not adequate shielding on my part! And some part could have been jitter in Ariel's clock. Anyway, 72 dB was more than good enough for my needs, and I haven't dug deeper. The fit mentioned involved only 100 points, spanning about 1% of a full cycle, so I was not greatly disturbed over the difference between the known frequency and the fitted frequency: 100.000 0 - 99.836 = 0.164 Hz; anyway, this is comparable with the uncertainty: 0.12 Hz, so there is no statistically significant difference. There is no difficulty acquiring 10^5 points and bringing them into Mma; some sorts of further analysis with Mma may be too time-consuming with so many points... .
Altogether, the gadget has proven handy, and useful for our needs!
Regards,
Henning Leidecker
These are the contents of the former NiCE NeXT User Group NeXTSTEP/OpenStep software archive, currently hosted by Netfuture.ch.