Ruediger,

Re the output of a large speaker for potential use as a seismometer:

Your experiment with the 100x amplifier and speaker does show that
the moving coil can generate a voltage with indescernable motions,
like fanning air against the speaker. This, however, is a very large
signal when compared with seismic ground motions. In fact, even miniscule
thermal convection WITHIN the case of a seismometer can be a large
signal, and we have resorted to small (10 watt) heaters in the topmost
part of large seismometer cases to try to stratify (hotter at the top)
the air within the case. We also put insulation between the pier and the
bottom of the case, since earth heat transmitted through the pier would
cause a thermal inversion within the case. Some modern and pricy VBB 
sensors are installed in a vacuum bell.

Here are some numbers that may help clarify the sensitivity question.
I have measured the coil constant of several large speakers, and
the constants run from 10 to 20 Volts/meter/second. By comparison,
small high-frequency geophones run about 40 V/M/s, larger 1 hz and up
seismometers run 200 to 500, and fedback (electronic) seismometers
have an output of 1000 V/m/s and up.

In our normal use of a 1 hz seismometer with 270 V/m/s (the L-4C with
a 5500 ohm coil), remote station sites have required amplifier gains
of 60 to 78 db (x 1000 to x 8000) for optimal sensitivity to both
near-field and distant earthquakes. We use a damping/attenuation
network that provides 100V/m/sec into the amplifier, which after the 
gain of 4000, gives 0.4 volt/micron/second. With an electronic noise
level of around 1 millivolt, our resolution is about 2.5 nanometers/second,
and our maximum signal is 4 volts = 10 microns/second. (I have such
a "short period" telemetered station in the back field of the farm here).
The station is sensitive to any Mb 3.0 event within several hundred 
kilometers, and most 1-second p-waves from teleseisms.

So to use a speaker with 10 V/m/sec, an amplifier gain of about 10 times
what we use for the L4-C would be needed, or something like 8000 to
80,000. This could result in a fairly noisy amplifier since several stages
would be required (we use two stages for the 78db amplifier; the schematic
of which is posted). But this is workable with proper filtering.

But a low output from the coil/magnet is not the main problem with a
speaker. The compliance of the speaker suspension is rather stiff for
seismometer use. I chopped a speaker cone and the back webbing into
spider strips to still support the coil, but was unable to sense
the earth microseism background of about 1 micron at 6-seconds.
Since I had glued on about 10 each 1/4" lead shot balls as a mass,
eventually the speaker suspension sagged and the coil dragged.

I think that carefully winding a coil (ie forget the drill) for a
maximum number of turns within the magnet gap, but enough length
either side to have a linear output, will produce the best signal.
The output increases with the square root of the coil resistance
if all the other dimensions are the same (ie more turns means 
smaller wire in the same volume so the resistance increases; see
the Riedesel paper).  But given unlimited space, it is a function
of the number of turns within the magnet gap. Various trade-offs 
seem to favor using wire about #36 size to achieve coils of a few
thousand turns and several hundred ohms for strong magnets.

Regards,
Sean-Thomas

_____________________________________________________________________

Public Seismic Network Mailing List (PSN-L)