Thread Tools
Aug 09, 2006, 11:58 AM
Caution:Makes sharp left turns
Troy's Avatar
Thread OP
Discussion

Doppler - speed analysis


I have been playing with the programs available from Shrike 1/2A racing group and would like to start a discussion on it. The main issue I have with this particular program is that it only accepts .wav files and you have to convert other files to .wav before you can use it. There is an .avi to .wav converter there but it seems like most of the video files I see are either mpeg or .mwv.

I'd also like to see how you gather RPM measurements from this software.
Last edited by Troy; Aug 09, 2006 at 12:20 PM.
Sign up now
to remove ads between posts
Aug 09, 2006, 12:16 PM
Caution:Makes sharp left turns
Troy's Avatar
Thread OP
I did find a file converter here:
http://www.filetrial.com/convert/mpeg-2towav.html
Aug 09, 2006, 03:23 PM
hot air rises...
jfinch's Avatar
Ditto... Anyone know of any freeware conversion software that doesn't just demo?

Troy, I've used wavepad (www.nch.com.au/wavepad/) which will import any video or audio format and then resave as *.wav, but after 30 days the demo stops and it will no longer import video files (but it's still pretty good for editing audio files).

In alot of ways I like this program (spectrogram) better then Waveooscope. Easier to see what's happening and some interesting info can be found. I'll add to this when I get home... gotta go back to work.
Aug 09, 2006, 04:21 PM
Caution:Makes sharp left turns
Troy's Avatar
Thread OP
I also like Spectrogram better because it doesn't limit the length of the .wav file like Wavescope does. I can isolate the area of the sample and run just a particular pass of a 200 second long file.
Aug 09, 2006, 04:41 PM
Flying motor mount master
fly_boy99's Avatar
Troy-

Take the first video of F5D flights on the F5D.org page you showed me
earlier and run them through Spectrogram and tell me what you get.

Even with Spectrogram it's kind of a pain in the butt to figure out the
start and end traces.

B
Aug 09, 2006, 04:53 PM
Can you spot the giraffe?
daleksw's Avatar
Imtoo can be found here. It can convert most things to most other things, and extracting audio from wmv, mpg or mp4 is easy. The free download converts up to 5 mins of any file, which should be plenty for this application.

Edit: should have said, it's the mpeg encoder tool you want.

David
Aug 09, 2006, 05:36 PM
Caution:Makes sharp left turns
Troy's Avatar
Thread OP
Quote:
Originally Posted by fly_boy99
Troy-

Take the first video of F5D flights on the F5D.org page you showed me
earlier and run them through Spectrogram and tell me what you get.

Even with Spectrogram it's kind of a pain in the butt to figure out the
start and end traces.

B
I did play with those and they are not great audio samples. For one thing, they are way off to the side so the Factor is just a guess. I use 10 or 15 and in the most legible wave I can measure shows about 170mph and that's towards the end of the race. That's believable since we know they do on average about 175-180+mph on course from radar readings. Sample below. From what I know this was a 64 second run.

The biggest problem with some of these samples is that the camera follows the plane and messes up the doppler effect. The audio sample should be from a stationary microphone. In the "Pilot's view" videos I can't even get a noticable drop from doppler. It's just a big red blob. The plane needs to enter and exit the area. Throwing an over head turn in there messes up the effect.
Aug 09, 2006, 09:40 PM
hot air rises...
jfinch's Avatar
My doppler thesis... or what I've learned last couple days. This is going to be a long post, I hope you can bear through it.

I had argued in the past that it didn't make sense that rpm's could be measured using doppler, but I think I was wrong because I didn't really understand how the principle worked. I think everyone understands the general concept of doppler shift so I won't talk about that. But the equations that describe the shift for an object approaching you and leaving are:

Fa = Fm*Vs/(Vs-Vm)
Fg = Fm*Vs/(Vs+Vm)

where Fa is the frequency as the plane approaches, Fg is the frequency as the plane leaves, Fm is the natural motor frequency, Vs is speed of sound and Vm is the plane speed.

Solving these two equations for the two unknowns (Fm and Vm) you get:

Vm = Vs*(Fc-Fg)/(Fc+Fg)
Fm = 2/(1/Fc + 1/Fg)

Those two equations give plane speed and motor frequency. It also turns out that motor frequency is directly proportional to motor RPM (but I had to prove this to myself).

RPM = Fm*60/Octave

First let's look at motor rpm. I recorded various motors spinning a prop (static) at a known rpm as measured using a high quality tach on a cheap cell phone, then replayed this recording from the phone through a cheap microphone that came with the sound card on my computer to my hard drive. It's simply amazing that this even works with all the low brow audio transfers! Anyway, the first two pictures below show the screen shot of 20,200 rpm and 13,560 rpm. The base (lowest) frequency is the frequency that relates directly to motor rpm. And if it's difficult to see/pick the base frequency then just pick one that is good and clear. I'm not sure why the cursor doesn't show up in my screen shots, but you can tell what frequency it's at by looking in the info box. The 20,200 motor shows a clear base octave of 337 hz which is 20,220 rpm. Pretty close to measured, I'd say! The 13,560 motor didn't show a real clear base octave so I picked the second octave and calculated a motor frequency of 455 hz. This calculates to 13,650 rpm (remember to divide by the octave, 2). That too is very close!

Does it work on motors spinning props with more then one blade? I downloaded rvincent's video of his Mig showing the static rpms of his new EDF motor. His tach showed 52,800 rpm. The first picture shows a full the spectrum of that audio. I show this to show how many different octaves these recordings have. Anyway, the second picture shows a close-up of the base octave which has a frequency of 880 hz. That calculates to 52,800 rpm! Yowza, right on (there's some validation for you rvincent)!

Ok, so I wonder if this will work on a motor without a prop at all. No-load. I tried three different motors at no-load and it was not as cut and dry as with a prop. For example one motor didn't show any trace that I could see until the 20th octave! Don't know how you could know that beforehand (I knew only because I already knew the rpm). But the other two motors showed better results. You can easily see the second octave (base octave not shown) for a motor spinning at 37,970 rpm (1266 hz) which calculates to 37,980 rpm.

I'm still shaking my head at this all, but it does appear that you can measure motor rpm using this software.

Ok, back to measuring plane speed. Attached is the full frequency response of a high speed pass of an electrified vendetta. As you can see there are many octaves to pick from. The base octave isn't too clear, so I picked the second octave to zoom in on. Being able to zoom in on a particular frequency is what really set's this software apart from Waveosscope, imo. The final picture shows a close-up of that second octave. The approaching frequency is 1274 hz and leaving it's 827 hz. This gives a plane speed of 164.5 mph and motor rpm of 30,088 rpm.

The final two attachments are the audio file of the 20,200 rpm motor if you want to play with it in spectrogram. And the last file is an excel spreadsheet I put together to calculate speed and rpm. Have a go at it and see what you think.
Last edited by jfinch; Feb 18, 2012 at 11:17 AM. Reason: updated spreadsheet
Aug 10, 2006, 10:12 AM
Caution:Makes sharp left turns
Troy's Avatar
Thread OP
How exactly are you zooming in on the octaves?
Aug 10, 2006, 11:01 AM
hot air rises...
jfinch's Avatar
After loading file go to file>parameters>change. There are lots of stuff you can change on this sceen.

Some of the parameters that I change around.

Scale (dB) will change how intense (colorful) the scan is.
Time Scale (msec) will either stretch out or shrink the x-axis (time) scale.
FFT Size (Points) is what you change if you can to zoom in on things. 512 is like a wide angle lens and 16384 is maximum magnification. At the higher magnifications you have to select the Band (Hz) that you are interested in and Freq Resolution (although I don't typically mess with the resolution). And don't forget that you have to use the scroll bars in the window.

Below is the analysis of part of one of the F5D vids. You can clearly see the three planes making laps and even calculate speeds for each plane, although the camera was in the wrong location for doppler. It's still pretty interesting though.

Waste 30 minutes messing around with those settings and it'll all make sense.
Aug 10, 2006, 04:25 PM
Caution:Makes sharp left turns
Troy's Avatar
Thread OP
Can someone make a .wav file from this awesome putfile video? I tried it with the Microsoft recorder but I don't have a microphone to record the sound. This thing was cooking before it disintegrated.
http://media.putfile.com/fai-crash
Aug 10, 2006, 04:52 PM
UN Earth peoples true enemy
treehog's Avatar

I could do with a rpm reader


I often take the laptop flying to down load the vidio feed from camera from filming or aerial work as memory is limited to 15 minutes

A program to measure RPM would be good

Also I want to set up the laptop so that within one second of passing the microphone the speed is calculated

hopefully this can be put into a program which could then calculate the point to turn at the end pylon and with some suitable signal noise or light or similar that indicates the turn and I could then turn the plane as well as having a caller of which I havent got anybody crazy eneogh to do this job

yeah that plane was smoking before it shed something
mind you the girl looked better than the laptop

I hate the put files as I cant save them

but then again eventualy mayby the laptop could replace the girl when it comes to turn turn turn

I also saw that doppler can be used for DS as the whistle is suffent to make a doppler affect
Aug 10, 2006, 06:30 PM
Registered User
down_shift's Avatar
Quote:
Originally Posted by Troy
Can someone make a .wav file from this awesome putfile video? I tried it with the Microsoft recorder but I don't have a microphone to record the sound. This thing was cooking before it disintegrated.
http://media.putfile.com/fai-crash
I like the wide-angle capture. Sorry, but the plane is way too far away to obtain a clean and reproducible Doppler shift. However, measuring the prop's RPM (or the frequency of any flutter) would be easy.

Try to capture a sharp Doppler shift like this video clip. (only 1.3meg)
Aug 11, 2006, 12:04 AM
hot air rises...
jfinch's Avatar
Yeah the FIA Crash video doesn't have a good enough pass (the couple passes he did make were ~200 mph but he was also turning away from the camera). The pass that his wings fell off would have been a good pass. The motor was spinning at 23,580 at take-off.

down shift, that video would be much better for doppler if it didn't have the music over the plane.
Aug 11, 2006, 07:39 AM
Registered User
down_shift's Avatar
Quote:
Originally Posted by jfinch
down shift, that video would be much better for doppler if it didn't have the music over the plane.
I added the music later.

Here are the results from the first pass and pictures of the two strongest Doppler shifts (one being the airframe, the other being the propeller).


Quick Reply
Message:

Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Doppler Speed Help?? c/f Flying Wings 16 Feb 26, 2010 11:06 PM
Discussion Doppler speed measurement astroboy403 High Performance 12 Jun 15, 2008 12:14 AM
Question doppler speed program.. rjarois High Performance 4 Aug 05, 2006 05:26 PM
Question How to shoot/calculate doppler speed? meteor Electric Ducted Fan Jet Talk 5 Sep 10, 2004 09:11 AM
Doppler shift speed analysis realdeal Power Systems 5 May 01, 2002 12:28 AM