re: help sensing building pressurazation
8 oct 1996
paul milligan wrote:
>email@example.com (guy macon) pondered, and wrote:
>>what basicly happens is that instead if storing the energy as pressure
>>in the elastic volume of air between the gravel, you store it in the
>>elactic capacitance of the rc filter.
> ok, i got the buffer ( or 'averaging' maybe is a better word ) effect
>- you're doing a trickle-charge rc. so, it's in series with a
>pressure transducer ? this is the part i didn't get from your post.
guy suggested a low pass filter with an rc time constant of 10 seconds,
as i recall. is that long enough? wouldn't a minute or two work better?
rc is the product of the resistance in ohms and the capacitance in farads,
in this case 100k ohms ie 100,000 ohms ("danger! 100,000 ohms!") and 100
microfarads, ie 100 x 10^-6 or 0.000100 farads, and it comes out in units
of seconds. if there were only one pressure sensor, and its amplified output
signal were 5 v for a long time, charging up a large capacitor through a
large resistor for a long time, with the other end of the capacitor grounded,
and then the pressure sensor signal suddenly changed from 5 v to 0 v,
the capacitor voltage would start to slowly decrease, following this formula:
v(t) = 0 + (5-0)exp(-t/10), where t is in seconds, and exp is the "e-to-the-x"
button on your $15 casio fx-990 calculator. after 10 seconds, the voltage
drops to about a third of the original, after another 10 seconds, a ninth,
and so on, until after 5 time constants, it's almost 0, but still decaying,
if you were to look extremely closely. it never quite reaches 0.
i could see this being built with a largish capacitor with one end connected
to a signal ground and the other end connected to a few low-impedance
high-level (eg, 0-5v, after amplification) pressure sensor outputs through
series 100k resistors. that point, ie summing node, might also be connected
to a high impedance voltage follower input, or this might all be done
less noisily in software.
or we could do this with a big jar (how big?) with some holes and tubes
(how big woudl the holes be, to make a 1 minute time constant?) how do
we calculate the time constant of that pneumatic rc circuit?
if we really wanted a long time constant, ie integration time, we might
use a microprocessor to do all this, or more interestingly, to me at
least, turn the pressure sensor outputs into heat, and heat a box full
of water, and sense the water temperature to measure the average wind
pressure. what is the time constant of an 8' cube of water with r20 walls,
say a plywood-lined box full of water surrounded by 6 inches of fuzzy
fiberglass insulation? the thermal resistance r is r20/(6x64ft^2) f-hr/btu
and the 512 cubic feet x 64 lb/ft^3 = 32,768 lb of water has a thermal
capacitance c of 32768 lb x 1 btu/f-lb of water = 32768 btu/f, so the
rc time constant is 20/(6x64)32768 = 1707 hours, ie 10 weeks.