This document was written in fall 2005 before my conversation with the
engineer at Samson. Be sure to note that my initial conclusion that
the mic works wonderfully under Linux was incorrect. Also note that
my later conclusion "The Bad News" was also not completely correct.
The fact, it seems, is that one can get OK recordings with the mic under
Linux with no special mixer settings/support. But to get the full performance
of the mic, somebody needs to write code that sends the optimal gain
parameters to the mic in response to the fader/slider setting in the
Linux audio mixer. I have no further information on this since I
don't own the mic anymore.
The Samson C01U USB microphone works fine under
Linux (kernel version 2.6.12-5).
I use amixer to control it.
It comes up muted by default on my system,
so it is necessary to unmute it (enabling capture) before recording.
I have confirmed that the dynamic range is at least 83dB.
It is probably more than that but I can't confirm that.
Theoretically, as a 16-bit device, it could be around 96dB.
The way I did this test was to adjust the input level
using amixer to a level that was just under overload on my test sound, which
was tapping hard on bottom of an empty paper coffee cup 20cm from the mic,
on axis.
Without touching any settings or unplugging the mic, I
wrapped the mic in my down jacket and rested it in a chair.
I recorded several 10 second files
of the signal from the mic while it was in this insulated configuration.
Sox was used to compute the RMS power of the signal in each file and lowest
figure was kept. The logic behind this is that all of the higher figures
were corrupted by ambient noise, and the lowest one is the least corrupted
and therefore the closest to the truth. It is unlikely that there are
noise processes in the microphone that come and go on a ten-second time scale.
The RMS amplitude
given by sox for the quietest recording was 0.000071 which is normalized so that a full scale square
wave will give an RMS power of 1.0.
20log10(1/0.000071) = 82.97 dB
This is pretty good. But it certainly isn't so good that we can just forget about
setting levels when recording and handle it later in software. If the dynamic
range were 120dB we could probably get away with that. I am not aware of any
theoretical noise floor in condenser mics. Dynamic mics have thermal noise.
By the way, dynamic range and signal to noise ratio in audio are very tricky
and can easily be misleading. For example the top end of the dynamic range
is defined by the onset of clipping. But in many devices, the distortion
starts rising well before it gets to the actual point of clipping. And
also most devices have frequency-dependent maximum levels. And also some
devices sound worse than others when they distort.
The main thing that I was interested in was whether Samson could keep the
digital parts of the mic from contaminating the low level analog input.
It seems that to a large extent they suceeded. I would be really interested
in taking the non-USB version of this mic, the Samson C01U, feeding it
to a good mic preamp and a good A/D converter, and doing side-by-side
comparisons of the performance in the same acoustic environment. But I don't
have any of that gear and don't plan on getting it any time soon.
And now, the bad news...
A few days after I wrote the section above, I used the Samson C01U USB mic in
a real-world recording situation. The instrument was solo piano. Of course I
wanted the usual things: low distortion, low noise, flat frequency response,
and reasonable sensitivity. The grand piano being recorded had its lid open and the mic
was on a straight mic stand next to the piano, pointing in. Under ALSA
the mic exports one control, which I set using amixer so that the loudest
peaks occuring during soundcheck just hit the -10db mark. Then I recorded
a few minutes of audio at the best audio quality that the mic can deliver.
The sound was slightly distorted and also had a high noise floor.
This was not digital distortion. The mic distorted long before it ran
out of digital headroom.
It seems like a flaw in the gain structure
of the mic, but it is hard to imagine an experienced audio company like
Samson making such a basic mistake. I wrote Samson and asked for more
technical information about the mic, but they ignored my email. So I
returned the Samson C01U USB mic to the store where I bought it.
Based on my experience, I would not recommend the Samson C01U USB mic for
high quality recording under Linux.
Update: Dec 2, 2005:
An engineer at Samson recently contacted me and explained the architecture
of the C01U USB. I feel that if my letter in August to Samson had reached
him instead of going to /dev/null, he would have supplied the information
necessary to make the C01U USB microphone well-supported under Linux.