Reply by Mark December 13, 20122012-12-13
On Dec 12, 2:41&#4294967295;pm, Tauno Voipio <tauno.voi...@notused.fi.invalid>
wrote:
> On 12.12.12 7:52 , Randy Yates wrote: > > > > > > > > > > > Tauno Voipio <tauno.voi...@notused.fi.invalid> writes: > > >> On 12.12.12 8:51 , Randy Yates wrote: > >>> Gentle comp.dsp readers, > > >>> I've read a few places around the web that FMCW radar is good for > >>> ranging. Why? Why wouldn't a "standard" pulse radar be just as good? >
it probably would be "just as good" but not as easy to do.. there is a lot more average power in a CW signal compared to short pulses the RF analog hardware is a lot easier to to implement to create an FM CW carrier compared to very short high power pulses no T/R switch needed does the FM system need 2 antennas? probably a bunch of other implementation details Mark
Reply by Tauno Voipio December 12, 20122012-12-12
On 12.12.12 7:52 , Randy Yates wrote:
> Tauno Voipio <tauno.voipio@notused.fi.invalid> writes: > >> On 12.12.12 8:51 , Randy Yates wrote: >>> Gentle comp.dsp readers, >>> >>> I've read a few places around the web that FMCW radar is good for >>> ranging. Why? Why wouldn't a "standard" pulse radar be just as good? >>> >>> I understand that as the range becomes short, the time difference >>> detected by a pulse radar would become very small. In that case an FMCW >>> radar, which detected range via delta f, might be easier or more >>> accurate, especially with a relatively high sweep rate. But then a >>> relatively high sweep rate implies a large bandwidth and the associated >>> computational complexity. >>> >>> Or how about using a pseudorandom sequence? If we know the round-trip time is >>> less than tau, then generate length-N PN sequence, >>> >>> N >= ceil(tau / Tb), >>> >>> where Tb is the chip period and use that to correlate and find time >>> delay. >> >> >> In aircraft radio altimeters, the signal is a linearly swept FMCW. >> The wide-band processing is in the analog part: The received signal >> is mixed with a sample of the transmitted signal, and the difference >> signal is processed. The difference is proportional to the amount >> of frequency change change of the FM signal during the travel time. >> >> It is a simple matter to measure the frequency of the LF signal >> and display the result in units of distance. >> >> The typical range of a radio altimeter is 0 to 2500 ft. > > Thanks for the information, Tauno. This agrees with what I've read. > Although I wonder what they use for other altimeters - I can clearly > remember the altimeter in my Dad's Cessna Centurion (back in the 70s) > registering up to 10,000 feet at least.
The standard altimeter is just a barometer with adjustable zero point. The relation between pressure and altitude is somewhat curved. In the classical altimeter the curve is created with a crank and lever mechanism. The curvature is defined by the ICAO atmospheric model, with 15 degrees C and 1013.2 hectopascals at sea level (your thing may have had inches of mercury, 29.92 inHg). For the formulas, google for 'barometric formula'. -- Tauno Voipio
Reply by Eric Jacobsen December 12, 20122012-12-12
On Wed, 12 Dec 2012 09:20:01 +0200, Tauno Voipio
<tauno.voipio@notused.fi.invalid> wrote:

>On 12.12.12 8:51 , Randy Yates wrote: >> Gentle comp.dsp readers, >> >> I've read a few places around the web that FMCW radar is good for >> ranging. Why? Why wouldn't a "standard" pulse radar be just as good? >> >> I understand that as the range becomes short, the time difference >> detected by a pulse radar would become very small. In that case an FMCW >> radar, which detected range via delta f, might be easier or more >> accurate, especially with a relatively high sweep rate. But then a >> relatively high sweep rate implies a large bandwidth and the associated >> computational complexity. >> >> Or how about using a pseudorandom sequence? If we know the round-trip time is >> less than tau, then generate length-N PN sequence, >> >> N >= ceil(tau / Tb), >> >> where Tb is the chip period and use that to correlate and find time >> delay. > > >In aircraft radio altimeters, the signal is a linearly swept FMCW. >The wide-band processing is in the analog part: The received signal >is mixed with a sample of the transmitted signal, and the difference >signal is processed. The difference is proportional to the amount >of frequency change change of the FM signal during the travel time. > >It is a simple matter to measure the frequency of the LF signal >and display the result in units of distance. > >The typical range of a radio altimeter is 0 to 2500 ft. > >-- > >Tauno Voipio
In ancient times when I worked on radar for a living what you're describing was called "de-ramp on receive", where the receive LO swept at the same rate as the Tx ramp so that the output Rx frequency was proportional to range. An FFT of the output of the mixer provided the range profile. Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by Randy Yates December 12, 20122012-12-12
Tauno Voipio <tauno.voipio@notused.fi.invalid> writes:

> On 12.12.12 8:51 , Randy Yates wrote: >> Gentle comp.dsp readers, >> >> I've read a few places around the web that FMCW radar is good for >> ranging. Why? Why wouldn't a "standard" pulse radar be just as good? >> >> I understand that as the range becomes short, the time difference >> detected by a pulse radar would become very small. In that case an FMCW >> radar, which detected range via delta f, might be easier or more >> accurate, especially with a relatively high sweep rate. But then a >> relatively high sweep rate implies a large bandwidth and the associated >> computational complexity. >> >> Or how about using a pseudorandom sequence? If we know the round-trip time is >> less than tau, then generate length-N PN sequence, >> >> N >= ceil(tau / Tb), >> >> where Tb is the chip period and use that to correlate and find time >> delay. > > > In aircraft radio altimeters, the signal is a linearly swept FMCW. > The wide-band processing is in the analog part: The received signal > is mixed with a sample of the transmitted signal, and the difference > signal is processed. The difference is proportional to the amount > of frequency change change of the FM signal during the travel time. > > It is a simple matter to measure the frequency of the LF signal > and display the result in units of distance. > > The typical range of a radio altimeter is 0 to 2500 ft.
Thanks for the information, Tauno. This agrees with what I've read. Although I wonder what they use for other altimeters - I can clearly remember the altimeter in my Dad's Cessna Centurion (back in the 70s) registering up to 10,000 feet at least. -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
Reply by Bryan December 12, 20122012-12-12
Yikes it's too early.

* "...a quick search shows me it's typically a linear FM chirp..."
Reply by Bryan December 12, 20122012-12-12
On Tuesday, December 11, 2012 11:51:01 PM UTC-7, Randy Yates wrote:
> Gentle comp.dsp readers, I've read a few places around the web that FMCW radar is good for ranging. Why? Why wouldn't a "standard" pulse radar be just as good? I understand that as the range becomes short, the time difference detected by a pulse radar would become very small. In that case an FMCW radar, which detected range via delta f, might be easier or more accurate, especially with a relatively high sweep rate. But then a relatively high sweep rate implies a large bandwidth and the associated computational complexity. Or how about using a pseudorandom sequence? If we know the round-trip time is less than tau, then generate length-N PN sequence, N >= ceil(tau / Tb), where Tb is the chip period and use that to correlate and find time delay. -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
Can you post the "few places around the web"? I'm not completely familar with FMCW, but a quick search shows mean it's typically what a linear FM chirp with no time between pulses. If you want to compare linear FM chirps with "standard pulses", it can be an optimization problem if you have dynamic control over the pulse parameters. For a standard pulse, you can only control the duration of the pulse. This control will allow one to optimize for lowering the variance of the estimate of either the range or the doppler shift but not both simultaneously (shifting the energy of the ambiguity between domains). For a linear FM chirp on the other hand, you have two parameters in your control: the duration and the chirp rate. Alternatively, the maximum frequency in the sweep and the pulse duration. This allows for control over the variance of both parameters.
Reply by Tauno Voipio December 12, 20122012-12-12
On 12.12.12 8:51 , Randy Yates wrote:
> Gentle comp.dsp readers, > > I've read a few places around the web that FMCW radar is good for > ranging. Why? Why wouldn't a "standard" pulse radar be just as good? > > I understand that as the range becomes short, the time difference > detected by a pulse radar would become very small. In that case an FMCW > radar, which detected range via delta f, might be easier or more > accurate, especially with a relatively high sweep rate. But then a > relatively high sweep rate implies a large bandwidth and the associated > computational complexity. > > Or how about using a pseudorandom sequence? If we know the round-trip time is > less than tau, then generate length-N PN sequence, > > N >= ceil(tau / Tb), > > where Tb is the chip period and use that to correlate and find time > delay.
In aircraft radio altimeters, the signal is a linearly swept FMCW. The wide-band processing is in the analog part: The received signal is mixed with a sample of the transmitted signal, and the difference signal is processed. The difference is proportional to the amount of frequency change change of the FM signal during the travel time. It is a simple matter to measure the frequency of the LF signal and display the result in units of distance. The typical range of a radio altimeter is 0 to 2500 ft. -- Tauno Voipio
Reply by Randy Yates December 12, 20122012-12-12
Gentle comp.dsp readers,

I've read a few places around the web that FMCW radar is good for
ranging. Why? Why wouldn't a "standard" pulse radar be just as good?

I understand that as the range becomes short, the time difference
detected by a pulse radar would become very small. In that case an FMCW
radar, which detected range via delta f, might be easier or more
accurate, especially with a relatively high sweep rate. But then a
relatively high sweep rate implies a large bandwidth and the associated
computational complexity.

Or how about using a pseudorandom sequence? If we know the round-trip time is
less than tau, then generate length-N PN sequence, 

  N >= ceil(tau / Tb),

where Tb is the chip period and use that to correlate and find time
delay. 
-- 
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com