DIY Radar.. tip?

[Mike Monett]

Mac wrote:

[...]

I guarantee you that if you ask any instrument radar person what the
resolution is of a linear FM homodyne (i.e., FM Continuous Wave) radar,
he (or she) will say it is C/(2 * BW).

--Mac

Then how do you get the typical specs of 3 mm resolution over 30
meter distance with the products shown in Speff's links? FFT won't do it,
that's 10,000 bins - and sifting through some 800 FMCW patents (many were
duplicates) didn't show any breakthrough that would give this resolution.

I have, however, figured out an inexpensive way to get this level of
resolution for a first-surface radar without using FFT and without
concern for vco linearity. The only problem is I can't figure out if this
is the method used by these manufacturers, or if it's a new way of doing
it.

But the resolution equations given in different texts can be broken.
Obviously, as shown from Speff's links.

Mike Monett

 

[Mac]

On Fri, 15 Oct 2004 00:00:06 -0400, Mike Monett wrote:

Mac wrote:

[...]

I guarantee you that if you ask any instrument radar person what the
resolution is of a linear FM homodyne (i.e., FM Continuous Wave) radar,
he (or she) will say it is C/(2 * BW).

--Mac

Then how do you get the typical specs of 3 mm resolution over 30
meter distance with the products shown in Speff's links? FFT won't do it,
that's 10,000 bins - and sifting through some 800 FMCW patents (many were
duplicates) didn't show any breakthrough that would give this resolution.

Don't forget to search for homodyne, too.

Anyway, I would assume they just use zero padding, but I certainly don't
know.

In case you (or other readers) don't know what I'm talking about,
here's how the zero padding thing works.

Let's say you digitize the IF for the entire chirp duration (without
violating Nyquist). Now you append a boatload of zeros to the end of the
data. Now do your FFT. Now pick the first bin with a major peak and that
is your range estimate. There is no reason why you couldn't do this with
8192 or 16384 or even 32768 bins. A 32k point FFT doesn't take long at all
nowadays.

This zero padding will produce an apparently higher resolution range plot,
although it really just amounts to interpolation in the range domain, as I
said before.

I have, however, figured out an inexpensive way to get this level of
resolution for a first-surface radar without using FFT and without
concern for vco linearity. The only problem is I can't figure out if
this is the method used by these manufacturers, or if it's a new way of
doing it.

Apply for a patent! Then once you have it, you can tell us all about it.

But the resolution equations given in different texts can be broken.
Obviously, as shown from Speff's links.

Only because we have a priori knowledge of the target's nature. That is,
we know it is safe to interpolate in the range domain because we know
there are no other scatterers near the surface of the liquid in the tank.

If there were scatterers closer together than the resolution, then zero
padding would not help resolve them. They would still appear as a single
broad scatterer. I suppose this could happen in a tank if there were two
immiscible fluids, and they were separated by less than C/(2*BW), and they
had different dielectric properties.

Another way of saying this is that the radar cannot _resolve_ objects
smaller than the resolution size, but, as I am now realizing, it _can_
provide more precise range estimates than the resolution would suggest,
as long as certain assumptions (mentioned above) hold up.

The resolution is still an important and useful concept, however.

Mike Monett

--Mac

 

[Scott Stephens]

Rich Grise wrote:

On Monday 11 October 2004 02:06 pm, Scott Stephens did deign to grace us
with the following:

Rich Grise wrote:

If you pulse or chirp radar down the 'leaky' waveguide it works real
good too!


Actually, the two concentric pipes are just a capacitor. You run it at
normal RLC meter frequencies - in fact, you could use an RLC meter,
as long as it can resolve fractional pf - I don't know how many it
would be, but the formulas are out there. When I was in the USAF,
on the way to/from the flight line I'd hear guys from the engine
shop talking about the fuel quantity indicator and picofarads in
the same breath. And I saw a thing that looked very much like a
tube, and I'm pretty sure it wasn't a float. ;-)

Hey, how about a float, and "full" and "empty" switches?
Or "Full", "Buy Oil Now" and "Freeze Tonight" switches? ;-)

I would suspect for a linear length of line, a linear capacitance would
be measured. Might well be simpler to measure capacitance with clean
fuel, rather than needing a radar that is insensitive to inhomogeneous
slurries like sewage.

--
Scott

**********************************

DIY Piezo-Gyro, PCB Drill Bot & More Soon!

http://home.comcast.net/~scottxs/

Those who sow excuses shall reap excuses

**********************************

 

[Mac]

On Mon, 11 Oct 2004 05:31:28 -0700, Andrey wrote:
[fixed top-post]

"Spehro Pefhany" <speffSNIP@interlogDOTyou.knowwhat> wrote in message
news:c00km0hu18uojdgku1ulv6618q9mhfeuph@4ax.com...
On 11 Oct 2004 03:09:08 GMT, the renowned dbowey@aol.com (Dbowey)
wrote:

But now try to build one. This *was* a request for a DIY project.

Don

Ok, here's a 10.525GHz (X-band) Gunn diode transceiver with 50MHz of
varactor tuning.

http://www.shfmicro.com/10ghz.pdf

Anyone want to take this further?


Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers:
http://www.trexon.com
Embedded software/hardware/analog Info for designers:
http://www.speff.com

To do FMCV out of it, Varactor should be povered up with rump voltage - this
is simple.

Then you need mixer which makes "F transmitted minus F reflected". The
difference frequency is function of distance. You will also need some kind
of blanking circuit for detector to cut away moments when rump swings back

The faster you can re-tune the gunn, the shorter distance you can measure

Andrey

I think what you are talking about is an FM homodyne radar.

There is an equation for FM homodyne radar resolution. It is R=C/(2*BW).
This can be found in any radar text book.

R is resolution, C is the speed of light in the medium, and BW is the
bandwidth of the transmit chirp.

So if C is 300 Mega meters per second, then 50 Megacycles of chirp
bandwidth will give you 3 Meter resolution. Not too good for a tank level
indicator.

I don't think fast retuning of the Gunn matters at all. You can chirp
quite slowly without affecting range resolution at all. And a slow chirp
will give a lower IF which will be easier to deal with. The idea is that
you transmit and receive at the same time. You have to do this anyway,
because otherwise you won't have anything to mix against (they call it
homodyne because the transmit pulse is mixed against the receive pulse).

One (of the many) problem you haven't addressed is that there could be a
fairly strong return from the bottom of the tank, depending on how
strongly the fluid attenuates the RF signal. This may be a stronger signal
than the surface level, and thus will dominate "F reflected."

To avoid this problem, you can do an FFT on the IF, and you will then have
a data set representing return strength vs range. In fact, this can be
thought of as a 1-d image.

One of the other problems might be that the RF will bounce around inside
the tank many times, leading to a whole host of phantom returns in the 1-d
image. Sorting this all out could be difficult. The easiest way to fix the
problem would be to put RF absorber material inside the top dome of the
tank. Then the reflections bouncing off of the fluid surface will be
absorbed at the top of the tank and will not bounce back down toward the
fluid surface.

--Mac