Forums

44kHz radar time of flight, yes and no

Started by Jan Panteltje November 29, 2013
 http://youtu.be/dnrH-7DyGXI

This works, increasing the distance between transmitter and receiver increases
the delay of the received signal.
I used a Raspberry Pi to make the on/off pulse,
well of course some 555 timers could do it too.
A small 2 transistor switch keys the transmitter oscillator driven from a GPIO pin.

Youtube resolution is a bit low to see the scope,
top trace is transmitted 44kHz signal,
bottom trace is received signal,
trigger is external from the Raspberry key pulse.
10 ms on, 10 ms off.  

So far so good.

BUT
doing it in the 'wind tunnel' with varying air speed,
does not give any measurable differences in time of flight.
Setup:
 http://panteltje.com/pub/44kHz_radar_time_of_flight_test_in_wind_tunnel_IMG_4105.JPG

So, as I now have the windtunnel I will next test some thermal circuits,
simpler, lighter (coils are heavy), and maybe even better.
Will see.

 

 



In article <l7a9e9$ga3$1@news.albasani.net>, Jan Panteltje
<pNaonStpealmtje@yahoo.com> wrote:

> http://youtu.be/dnrH-7DyGXI > > This works, increasing the distance between transmitter and receiver increases > the delay of the received signal. > I used a Raspberry Pi to make the on/off pulse, > well of course some 555 timers could do it too. > A small 2 transistor switch keys the transmitter oscillator driven from a > GPIO pin. > > Youtube resolution is a bit low to see the scope, > top trace is transmitted 44kHz signal, > bottom trace is received signal, > trigger is external from the Raspberry key pulse. > 10 ms on, 10 ms off. > > So far so good. > > BUT > doing it in the 'wind tunnel' with varying air speed, > does not give any measurable differences in time of flight. > Setup: > > http://panteltje.com/pub/44kHz_radar_time_of_flight_test_in_wind_tunnel_IMG_41 > 05.JPG > > So, as I now have the windtunnel I will next test some thermal circuits, > simpler, lighter (coils are heavy), and maybe even better. > Will see.
Judging by the pictures of the experimental setup, I doubt that the time-of-flight is large enough to be detected, given the response time of the sensors, which look like standard narrowband piezeoceramic burglar-alarm units. The Q of such units is too high to allow sharp responses. Ultrasonic burglar alarms are CW. Anyway, I'd do some back-of-the-envelop computations, to see if the setup is even plausible, Joe Gwinn
On a sunny day (Fri, 29 Nov 2013 10:32:34 -0500) it happened Joe Gwinn
<joegwinn@comcast.net> wrote in <291120131032349826%joegwinn@comcast.net>:

>In article <l7a9e9$ga3$1@news.albasani.net>, Jan Panteltje ><pNaonStpealmtje@yahoo.com> wrote: > >> http://youtu.be/dnrH-7DyGXI >> >> This works, increasing the distance between transmitter and receiver increases >> the delay of the received signal. >> I used a Raspberry Pi to make the on/off pulse, >> well of course some 555 timers could do it too. >> A small 2 transistor switch keys the transmitter oscillator driven from a >> GPIO pin. >> >> Youtube resolution is a bit low to see the scope, >> top trace is transmitted 44kHz signal, >> bottom trace is received signal, >> trigger is external from the Raspberry key pulse. >> 10 ms on, 10 ms off. >> >> So far so good. >> >> BUT >> doing it in the 'wind tunnel' with varying air speed, >> does not give any measurable differences in time of flight. >> Setup: >> >> http://panteltje.com/pub/44kHz_radar_time_of_flight_test_in_wind_tunnel_IMG_41 >> 05.JPG >> >> So, as I now have the windtunnel I will next test some thermal circuits, >> simpler, lighter (coils are heavy), and maybe even better. >> Will see. > >Judging by the pictures of the experimental setup, I doubt that the >time-of-flight is large enough to be detected, given the response time >of the sensors, which look like standard narrowband piezeoceramic >burglar-alarm units. The Q of such units is too high to allow sharp >responses. Ultrasonic burglar alarms are CW. > >Anyway, I'd do some back-of-the-envelop computations, to see if the >setup is even plausible,
It should be, but with a lot of DSP processing: http://www.technik.dhbw-ravensburg.de/~lau/ultrasonic-anemometer.html But now I will test some of this: http://www.fonema.se/anemom/anemom.html I want low weight, no icing, reliable, simple :-) Add cheap, and then it is eeh usual problem :-)
>Joe Gwinn >
On a sunny day (Fri, 29 Nov 2013 10:32:34 -0500) it happened Joe Gwinn
<joegwinn@comcast.net> wrote in <291120131032349826%joegwinn@comcast.net>:

>> does not give any measurable differences in time of flight. >> Setup: >> >> http://panteltje.com/pub/44kHz_radar_time_of_flight_test_in_wind_tunnel_IMG_41 >> 05.JPG >> >> So, as I now have the windtunnel I will next test some thermal circuits, >> simpler, lighter (coils are heavy), and maybe even better. >> Will see. > >Judging by the pictures of the experimental setup, I doubt that the >time-of-flight is large enough to be detected, given the response time >of the sensors, which look like standard narrowband piezeoceramic >burglar-alarm units. The Q of such units is too high to allow sharp >responses. Ultrasonic burglar alarms are CW. > >Anyway, I'd do some back-of-the-envelop computations, to see if the >setup is even plausible,
It is simple, if sound moves at 330 meter per second, so .33*3600 = 1188 km/h then if I fly at 120 km/h it makes only 10% difference. Measurable, sure, but needs more DSP. The air speed of the little home made windtunnel is probably more like 1m/s, 3.6 km/h, so hardly detectable. I think a thermal system will work better at that low speed. Target max speed is 160 km/h that I expect, so nothing is of the table as yet.
In article <l7ac9n$mcp$1@news.albasani.net>, Jan Panteltje
<pNaonStpealmtje@yahoo.com> wrote:

> On a sunny day (Fri, 29 Nov 2013 10:32:34 -0500) it happened Joe Gwinn > <joegwinn@comcast.net> wrote in <291120131032349826%joegwinn@comcast.net>: > > >In article <l7a9e9$ga3$1@news.albasani.net>, Jan Panteltje > ><pNaonStpealmtje@yahoo.com> wrote: > > > >> http://youtu.be/dnrH-7DyGXI > >> > >> This works, increasing the distance between transmitter and receiver > >> increases > >> the delay of the received signal. > >> I used a Raspberry Pi to make the on/off pulse, > >> well of course some 555 timers could do it too. > >> A small 2 transistor switch keys the transmitter oscillator driven from a > >> GPIO pin. > >> > >> Youtube resolution is a bit low to see the scope, > >> top trace is transmitted 44kHz signal, > >> bottom trace is received signal, > >> trigger is external from the Raspberry key pulse. > >> 10 ms on, 10 ms off. > >> > >> So far so good. > >> > >> BUT > >> doing it in the 'wind tunnel' with varying air speed, > >> does not give any measurable differences in time of flight. > >> Setup: > >> > >> > >> <http://panteltje.com/pub/44kHz_radar_time_of_flight_test_in_wind_tunnel_IMG_4105.JPG> > >> > >> So, as I now have the windtunnel I will next test some thermal circuits, > >> simpler, lighter (coils are heavy), and maybe even better. > >> Will see. > > > >Judging by the pictures of the experimental setup, I doubt that the > >time-of-flight is large enough to be detected, given the response time > >of the sensors, which look like standard narrowband piezeoceramic > >burglar-alarm units. The Q of such units is too high to allow sharp > >responses. Ultrasonic burglar alarms are CW. > > > >Anyway, I'd do some back-of-the-envelop computations, to see if the > >setup is even plausible, > > It should be, but with a lot of DSP processing: > http://www.technik.dhbw-ravensburg.de/~lau/ultrasonic-anemometer.html
This actually makes a phase measurement, according to the text. That's more plausible. If you run a two frequencies, one for N-S and the other for E-W, both sensors can be operated CW.
> But now I will test some of this: > http://www.fonema.se/anemom/anemom.html
Thermal does work, but needs some processing to linearize. The wires are either fragile or power-hungry.
> I want low weight, no icing, reliable, simple :-) > Add cheap, and then it is eeh usual problem :-)
Yep. Joe Gwinn
In article <l7aesb$sk4$1@news.albasani.net>, Jan Panteltje
<pNaonStpealmtje@yahoo.com> wrote:

> On a sunny day (Fri, 29 Nov 2013 10:32:34 -0500) it happened Joe Gwinn > <joegwinn@comcast.net> wrote in <291120131032349826%joegwinn@comcast.net>: > > >> does not give any measurable differences in time of flight. > >> Setup: > >> > >> > >> <http://panteltje.com/pub/44kHz_radar_time_of_flight_test_in_wind_tunnel_IMG_4105.JPG> > >> > >> So, as I now have the windtunnel I will next test some thermal circuits, > >> simpler, lighter (coils are heavy), and maybe even better. > >> Will see. > > > >Judging by the pictures of the experimental setup, I doubt that the > >time-of-flight is large enough to be detected, given the response time > >of the sensors, which look like standard narrowband piezeoceramic > >burglar-alarm units. The Q of such units is too high to allow sharp > >responses. Ultrasonic burglar alarms are CW. > > > >Anyway, I'd do some back-of-the-envelop computations, to see if the > >setup is even plausible, > > It is simple, if sound moves at 330 meter per second, so 0.33*3600 = 1188 km/h > then if I fly at 120 km/h it makes only 10% difference. > Measurable, sure, but needs more DSP. > The air speed of the little home made windtunnel is probably more like 1m/s, > 3.6 km/h, so hardly detectable.
So, the null result is to be expected. A phase-sensitive CW system can detect this. Arrange things such that at zero speed, the output of the mixer comparing the transmitted signal with the received signal is zero. As the wind speed picks up, the phase detector output will deviate from zero. The deviation follows a sine-wave function, and is linear for small deviations. There will be multiple non-zero windspeeds that also yield zero output from the mixer. There are various ways to solve this problem, to extend the range. Perhaps the simplest is to use a few mutually prime transmit frequencies (that all nonetheless are in the transducer passband), and use a variant of the Chinese Remainder Theorem to deduce the actual speed. Another common way is to amplitude modulate the 44 KHz signal with a low enough frequency that there is no ambiguity, and feed the demodulated TX and RX signals to the mixer instead of the 44KHz carrier. The modulation frequency must be low enough that the sidebands fit through the transducer bandwidth.
> I think a thermal system will work better at that low speed.
Hot wire anemometers do work well, and there is a huge literature on how to design them. But they are fragile and sensitive to dirt accumulation.
> Target max speed is 160 km/h that I expect, so nothing is off the table as yet.
Hmm. This If the house blows away, the airspeed will be lower than that. And there will be lots of flying junk in the air, and a hot-wire unit is unlikely to survive. Joe Gwinn
On a sunny day (Fri, 29 Nov 2013 17:18:39 -0500) it happened Joe Gwinn
<joegwinn@comcast.net> wrote in <291120131718391692%joegwinn@comcast.net>:

>> It is simple, if sound moves at 330 meter per second, so 0.33*3600 = 1188 km/h >> then if I fly at 120 km/h it makes only 10% difference. >> Measurable, sure, but needs more DSP. >> The air speed of the little home made windtunnel is probably more like 1m/s, >> 3.6 km/h, so hardly detectable. > >So, the null result is to be expected. > >A phase-sensitive CW system can detect this. Arrange things such that >at zero speed, the output of the mixer comparing the transmitted signal >with the received signal is zero. As the wind speed picks up, the >phase detector output will deviate from zero. The deviation follows a >sine-wave function, and is linear for small deviations.
I will try some phase detector on the CW Doppler setup. Have to think about what type. Maybe start with scope XY (Lissajous) to see what it does. http://en.wikipedia.org/wiki/Lissajous_curve
>There will be multiple non-zero windspeeds that also yield zero output >from the mixer. There are various ways to solve this problem, to >extend the range. Perhaps the simplest is to use a few mutually prime >transmit frequencies (that all nonetheless are in the transducer >passband), and use a variant of the Chinese Remainder Theorem to deduce >the actual speed.
Fun: http://en.wikipedia.org/wiki/Chinese_remainder_theorem
>Another common way is to amplitude modulate the 44 KHz signal with a >low enough frequency that there is no ambiguity, and feed the >demodulated TX and RX signals to the mixer instead of the 44KHz >carrier. The modulation frequency must be low enough that the >sidebands fit through the transducer bandwidth.
Yes, and there is a problem, as Q is high to get some high voltage with little power.
> >> I think a thermal system will work better at that low speed. > >Hot wire anemometers do work well, and there is a huge literature on >how to design them. But they are fragile and sensitive to dirt >accumulation.
Maybe when the filament is in a little tube, so it sees wind speed from one direction only, well back and forth of course, then it is more protected,
> >> Target max speed is 160 km/h that I expect, so nothing is off the table as yet. > >Hmm. This If the house blows away, the airspeed will be lower than >that. And there will be lots of flying junk in the air, and a hot-wire >unit is unlikely to survive.
We had a storm like that a couple of weeks ago, fence blew over, 2 wooden poles just broke off. Had to dig the remains out really deep (was in the groundwater) and build a new fence, still standing.