R=C/(2*BW) so:
c=2.997*10**8
bw=14.9*10**9
c/(2*bw) = .01 (m/Hz*s ?)
If the required precision is 0.01 meters and the method is based on
reflection. A wave should preferebly be completed with a wavelength of 0.02 m.
Given f=(c/lambda) and T=1/f, 1/(c/lambda) should give the cycle time. Which
in this case would be:
(1/(c/.02))*10**12 = 66.733 ps
To generate this a di/dT where T=16.683ps should be enough?
I can't answer this question. If you want to pursue this further, I think
you should start a new thread, but think in terms of ultra short pulse
generation and detection, since that is what you need to do.
Since I have now read some of the links other people put in this thread, I
see that the way to do this is to put a waveguide in the tank, and send
your pulse down the waveguide. You should get a reasonably strong
reflection from the surface of the liquid, and you might get another one
from the end of the wave guide (bottom of the tank).
Infact because what is measured is the the wavefront (I hope
it should be
possible to let the T of di/dT be 66.733. However I guess experimentation
will have show what actually works out.
If they can't measure it, it's not likely they will bother. Soil tend to be
a good conductor (compare to how deep ground radar goes).
Besides doing the measure every hour should be enough.
Right. What I was trying to say is that you are not going to turn this
into an approved product since it is an intentional radiator. But now I
think you could argue that it is not a radiator, since you would probably
use a wave guide to contain the radiation.
As for detection if echoes won't screw the thing up, and it should be ok
because hopefully the shortest path between tx and rx is straight reflection.
Impulses from other rf sources should be ok, considering the timespan is
short (kind of avoiding bullets in the matrix due speedy action
.
I agree. Also, you could take several readings and throw out any that seem
wild.
In the actual detection I though of letting a mosfet conduct immediatly
when the di/dt rise is done. Causing a capacitor to be "slowly" charged. And
when the wavefront hits the rx antenna another mosfet breaks the charge
through discharge of the capacitor of the input to the first one.
The capacitor charged between the tx and rx event should then correspond to
the time.
(Should I choose something else than mosfet?)
I don't know, and I don't think too many people are reading this thread
anymore. Like I said, start a new thread. I have seen this basic idea used
for precise timing, but not with MOSFETS.
As I recall, the circuit measured the time from when a pulse was received
until the next clock rising edge by charging a capacitor with a current
source during the intervening time. At the end of it, the voltage on the
cap was a measure of the time. But this was with ECL logic and some pretty
fancy calibrated current source. I didn't really understand it.
Another approach would be to setup some logic chips where the logic levels
is in a race condition based on input. And the resulting spike is integrated
through the means of a capacitor. A catch is if the gates will manage this
because they have a latency time of 3-4 ns on the fastest avail. But maybe
they get the propagation correct.
Again, I don't know.
(getting phase and amplitude for multiple quantas in more advanced apps would
be really tought in these timedomains
Like I said, maybe you should start a new thread. The basic parts
of the problem are:
1) design wave guide to put in tank
2) send sharp pulse (or step) type waveform down waveguide
3) detect first return pulse
4) somehow measure interval between pulse generation and pulse return.
I'm pretty sure that a very fast logic chip with a series resistor and a
high-speed oscilloscope could give you resolution of maybe 5 centimeters,
or so. But I'm not sure how to design a simple reliable detector that
could measure this.
The wave would look like this: (Use courier or something like that)
________________
/
/
______/
/
/
0V ____/
time ---->
The first rising edge occurs when the driver goes high. For a time,
all it feels is the transmission line impedance, so it flattens out.
When the reflection from the surface of the liquid hits the resistor,
we get a second rising edge.
Later, there would be a final rising edge, when we hit the open-circuited
end of the wave guide. Or the waveguide could be terminated in a
short-circuit. In fact, this is better. Then the reflection would be
negative and would cause the waveform to go back to zero volts.
It is important that the source impedance be very closely matched to the
transmission line impedance.
--Mac