Go Down

Topic: the smallest self-balancing robot (Read 2162 times) previous topic - next topic


robtillaart

#1
Oct 18, 2015, 09:55 am Last Edit: Oct 18, 2015, 09:56 am by robtillaart
moderatore: crosspost removed
Rob Tillaart

Nederlandse sectie - http://arduino.cc/forum/index.php/board,77.0.html -
(Please do not PM for private consultancy)

MarkT

[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

tristanscomics

What is the ping sensor? All of the ones I see detect things to the centimeter, which definitely won't be enough for a project like this.

DuaneDegn

#4
Oct 28, 2015, 05:15 am Last Edit: Oct 28, 2015, 05:15 am by DuaneDegn
What is the ping sensor?
https://www.parallax.com/product/28015

I think the resolution of the Ping (and similar ultrasound sensors) is just under a centimeter. If I did the math right, the wavelength of sound at 40KHz is 8.5mm. I think the resolution of an ultrasound sensor is limited by the wavelength. Medical ultrasounds use much higher frequencies.

I'm not sure is the wavelength really limits an ultrasound resolution. I've heard people I consider knowledgeable about this topic disagree with this claim.

Abhishek_M

Hi,
I am also building something similar so i hope you could help me.
I'm using an Adafruit 10DOF IMU. (http://www.adafruit.com/products/1604)
I can't get the accelerometer raw data converted from Gs[m/s^2] to degrees.
Could you suggest a way/help/guide to do that?

Johnny010

https://www.parallax.com/product/28015

I think the resolution of the Ping (and similar ultrasound sensors) is just under a centimeter. If I did the math right, the wavelength of sound at 40KHz is 8.5mm. I think the resolution of an ultrasound sensor is limited by the wavelength. Medical ultrasounds use much higher frequencies.

I'm not sure is the wavelength really limits an ultrasound resolution. I've heard people I consider knowledgeable about this topic disagree with this claim.
Well no, ultrasound works on a delay between propagation and echo. This is dependent on the speed of sound in a medium...not its wavelength.

The resolution will mostly be limited by the precision/resolution of the timing device.
The reliability is an issue, destructive echoes, external noise etc.

Speed of sound =~ 350ms-1

A 10cm distance from sensor = 20cm total.
time = 0.2m / 350ms-1 = ~6E-4s or 0.6 milliseconds.

0.6ms = ~16kHz

Add some issues with reliability. Systematic error such as de-constructive wave patterns on the echo and random error such as noise and you are left with a fairly tetchy way to use sound as navigation.

Unless you have the brain power of a bat and can analyse signals a little better.
PS. A bat uses 40kHz - 100kHz waves. They can spot a small mosquito...so the resolution is not wavelength based empirically.

DuaneDegn

Well no, ultrasound works on a delay between propagation and echo. This is dependent on the speed of sound in a medium...not its wavelength.
I'm not convinced either way myself. Having the resolution based on the wavelength matches with my experience using ultrasound sensors.

It's relatively trivial to time the echo pulse to the nearest clock cycle of an 80MHz microcontroller. At 80MHz a single clock pulse is 12.5ns. I'll double to 25ns this since we are likely to have some error at the beginning and end of the echo.

At 350m/s sound will travel 8.75um. This is well below one millimeter resolution but the resolution one can get from an ultrasound is much closer to the 8.75mm wavelength of a 40KHz transducer.

I'd be very interested in any experiments one could do to show resolution better than the wavelength.

It seems reasonable to me, for the transducer to require a full wavelength, of the sound being analyzed, to arrive, in order to detect the sound.

jremington

#8
Dec 04, 2015, 03:43 pm Last Edit: Dec 04, 2015, 03:44 pm by jremington
Of course the resolution is limited by the wavelength. This is true of any technique that involves wave propagation.

As DuaneDegn points out, for a sensor to pick up a pressure wave signal, a significant fraction of one wave cycle must act upon the transducer. Furthermore, most ping sensors send out a pulse train, and you might not even detect the first pulse in the train.

Go Up