Time-of-flight laser distance sensor

Discussion in 'The Projects Forum' started by cmartinez, Aug 15, 2013.

  1. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
    I'm starting a project trying to measure short distances (less than one meter) using a laser pulse to illuminate a target and later measuring how long it takes for its reflection to activate a photodiode placed just beside the laser source.

    I found this diode that has a 2ns response time:
    http://www.semicon.panasonic.co.jp/ds4/PNZ331CL_AED_discon.pdf

    Now:
    A) light travels at 300,000,000,000 mm/sec. That means that it takes 3.33 x 10-12 sec for it to travel one mm

    B) The distance I'd like to measure is, say, 500mm away (one meter, back and forth). That would be one meter back and forth. So light would take 1.67 x 10-9 seconds to cover that distance.

    So, if I wanted to measure distances with a 500mm resolution (one meter, back and forth), all I would need is a microcontroller working at a little more than 600 Mhz that would count the clock ticks it takes between its activating the laser signal and detecting the photodiode's response. That is, taking into account also the laser's ramp-up time, which should have fairly constant repeteability (yeah, right)
    That would also mean that all I would need to do to improve resolution would be to increase the MCU's frequency, which would be fairly easy, right?

    BUT... something tells me that I might have made a very basic mistake in these assumptions, and that the problem might be far more complicated than it seems.

    I'd like to hear anyone's opinion on this problem.
     
    Last edited: Aug 15, 2013
  2. wayneh

    Expert

    Sep 9, 2010
    12,085
    3,024
    Light travels at 3x10^8 meters per second, not mm. So there's 100X error right there.
     
  3. BobTPH

    Active Member

    Jun 5, 2013
    782
    114
    Ultrasound is much more feasible for this distance.

    Bob
     
  4. sirch2

    Well-Known Member

    Jan 21, 2013
    1,008
    351
    Yup, ultrasound, you can buy modules that run from microcontrollers pretty cheaply.

    If you need a point measurement then the little Sharp IR distance sensors (google it) are pretty good.
     
  5. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
    Wayneh, sorry, my bad, you're absolutely right. I discovered my mistake 2 minutes after I posted my thread, and later corrected it. You must've read the incorrect version first.
    I have nothing against ultrasonic sensors, it's just that I need a very quick response time (possibly faster than ultrasound can give) and I need to measure distance to an area of 1mm max. My intention is to map a steel surface with a 1mm resolution in distance and a 1x1mm resolution in area. I found the sensor you're referring to:
    https://www.sparkfun.com/products/242

    Don't you think that the area it would cover would be a little to big for what I want? Plus, the steel area I'm trying to map could be a little rusty or stained, and therefore not as reflective as the sensor might need.
     
  6. sirch2

    Well-Known Member

    Jan 21, 2013
    1,008
    351
    What resolution do you want?
     
  7. wayneh

    Expert

    Sep 9, 2010
    12,085
    3,024
    Another stupid question: Why not just buy one? Is there something about the commercial devices that you don't like? If it's just price, you may find it's costlier to make your own.

    You may want to read the wiki on this technology. To my eyes, it sounds like a triangulation strategy might be superior to time-of-flight for your application.
     
  8. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
    I want to be able to obtain a resolution of 1mm in distance, over an area of 1x1mm, I can place the sensor as close as 25mm away from the area to be measured. When you consider this, it's actually a very rough measurement, but it has to be done at a high speed.
    I've considered using inductive linear sensors, but I cannot place them close enough together to obtain that resolution, since the smallest commercially available sensor of this type is about 5mm in diameter (3mm in some brands)
    That's why I've been considering a time-of-flight sensor, since I can output a laser beam through an optic fiber and input it back to the sensor through another optic fiber (with a microlens at its tip)
    That way I can place, say 500 optic fibers for the laser, and another 500 for the sensors. The fibers would be placed alternatively in a row and that would allow me to rapidly scan the area I'm interested on.
     
  9. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
    There is no such thing as a stupid question, my friend. And thanks for trying to help me.
    I don't like the commercial ones because they're Expen$ive and slow, especially the triangulation ones (such as the ones manufactured by Keyence)... and I would need too many of them to get what I want. Besides, that kind of sensor would be overkill, considering the resolution I need. That's why I'm considering designing my own.


     
  10. wayneh

    Expert

    Sep 9, 2010
    12,085
    3,024
    Very interesting. Thanks for letting us see that you have been doing your own research.

    So you need a way to detect light time-of-flight with a precision down to 3 pico-seconds or better. Like reading 1mm record grooves at 333GHz. That's out of my comfort zone, way way out! I wonder if reverse engineering on of the commercial devices would get you a good start.
     
  11. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
    It's interesting, alright.
    Commercially available stuff is bulky and expensive, and I found a chip manufacturer that sells a specialized chip for this application.
    I've also been considering using a very high frequency counter (such as http://www.micrel.com/_PDF/HBW/sy10-100e137.pdf) that would be activated by the laser being turned off. (Something tells me that the ramp-down time would be faster and easier to manage than the ramp-up one)
    But that chip only works up to 2Ghz... so it's still about two orders of magnitude below what I need.. I need a counter capable of running at 400 Ghz, that way I would obtain a 2.5 ps resolution, barely enough for what I want... plus there would always be the issue of repeatability... a complicated problem indeed...
     
  12. sirch2

    Well-Known Member

    Jan 21, 2013
    1,008
    351
    What about using structured light or interferometry?
     
  13. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    Laser TOF ranging to 1mm resolution is a massive undertaking, even for someone skilled in RF systems and optics design. Doing it FAST (without averaging lots of readings over lots of time) ups the difficulty another (big) notch.

    Just google for "DIY 3D scanner", "my 3D laser scanner" etc to see the many ways people do this kind of thing much easier. Generally by using angle trig and video processing, where the straight laser is detected by a camera at an angle;

    [​IMG]
    :)
     
  14. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
    Yes, I've considered using trig measurement, but can this technique read multiple points at once?
    I have a couple of ATMEL linescan cameras (they have a single row of 4096 pixels) with me that have direct digital outputs that can easily be connected to an MCU and start experimenting... but how would I set them up in order to get what I want?
    You are right, TOF for this application would be a huge undertaking, and of course I'm still considering other technologies (such as high resolution radar: http://www.gizmag.com/high-res-radar-raindrops/23136/), but I'm running out of options here.
     
  15. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
  16. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    I think you should give up on the laser TOF ranging. :)

    You said you wanted to scan or map out a large steel surface with the measurement 1x1 mm sqare and 1mm resolution. That requires many thousands of readings and a lot of precise XY scanning, if you only measure 1x1mm dot at a time.

    Check out the "david 3D laser scanner" for a system that uses a laser LINE and camera, and can map a complex surface quickly and cheaply.

    There are a number of private projects using similar fast and easy methods for 3d surface scanning.
     
  17. cmartinez

    Thread Starter AAC Fanatic!

    Jan 17, 2007
    3,552
    2,490
    Thanks for the suggestion. Yes, maybe TOF is still in its infancy and the technology needs to be developed a lot further (by someone else that would actually get PAID for it) before it becomes more practical.

    I'll check on the 3d laser line scanner and let you know what I decided to do. Thanks.
     
Loading...