'OT: LC oscillator and PIC as capacitor AtoD'
I am attempting to make a seismometer using a slightly different approac
The sensor is a capacitor plate where one plate is fixed to the earth (case) and
the other plate is attached to a spring; hence when the earth moves the
capacitor spacing and therefore the capacitance changes. Normally the capacitor
sensor is placed in a bridge circuit but our idea is to use the sensor
capacitor as the capacitor in an LC oscillator. As the earth moves, the
capacitance changes and so does the frequency of the LC oscillator.
(Here comes the part where the PIC comes into play.) I then wish
to use a PIC to count the oscillations. The number of counts for a set
period of time (determined by the PIC's clock) will be purportional to the
capacitance spacing. This seems to me a simple way to digitize a
capacitor meter, yet I've never heard of people using this approach. Is
there some fundamental reason that this won't work that I'm missing?
Horowitz and Hill state in their book, The Art of Electronics, that
an LC oscillator is only stable to 1 part in 10,000. Can anyone tell
me why that is? Seems to me that an LC oscillator should be extremely
stable (assuming one is using a fixed rather than a variable capacitor).
Can anyone enlighten me as to why an LC oscillator is only stable to one
part in 10^4?
Thanks in advance for your help,
Uwe Koenneker (KNN)
For L/C-oscillators the biggest influence is given of the
temperature-coefficient of the L (not of the C).
So you will definetely have a temperature-drift, which will exceed the
influence of the capacitor-plate-
You might consider to compensate the tempereature-coefficient of the L
by using capacitors with the
opposite coeffiecient in parallel to your C. Another approach might be
to put your complete setup in a
"regulated oven". As the temperature-drift is quite small you might also
try to compensate it by other means.
But I believe it will be difficult!
|Slow-pass filtering (in effect) should handle this - he's after the
sudden changes in frequency, say 0.5 Hz or faster, for seismology; a
slow frequency drift over minutes or hours, shouldn't do anything but
change the comparison low-passed frequency at the output of the low-pass
filter. (Maybe a 10 second Tau there?) I'd be lots more concerned
about getting the thing down deep enough into bedrock to reduce road
noise, etc. than worrying about something that the software can handle
The way (for example) to detect windshear for aircraft safety can
involve low-pass filtering of barometric pressure, temperature, Angle of
Attack, etc. data (which, of course, ALL automatically changes as your
aircraft is in the air!) - it's the sudden trends, compared to the
low-passed information, all combined, not the drift of the inputs, that
are a trigger for that "Dee, Doop! WindShear! Windshear! Windshear!"
alarm, that hopefully will save all on board the airplane.
Uwe Koenneker (KNN) wrote:
More... (looser matching)
- Last day of these posts
- In 1999
, 2000 only
- New search...