Loading ...
Sorry, an error occurred while loading the content.
 

"Tilt" Sensor

Expand Messages
  • Chuck Hackett
    I am building a machine to level 7.5 gauge railroad tracks (see: http://www.cpgrr.org/). The device grabs the rails, lifts them into place (height and
    Message 1 of 7 , Apr 10, 2013
      I am building a machine to level 7.5" gauge railroad tracks (see:
      http://www.cpgrr.org/). The device grabs the rails, lifts them into place
      (height and "level") and tamps the ballast around the ties. To do this, the
      machine needs a "level" reference. In curves the machine must also add
      "super elevation" (raise the outside rail up to about 0.250" from level),
      the amount of which depends on the curvature at that point. In my case the
      frame of the machine is kept level by a sensor that feeds a microcontroller
      (ATMega128). (Among other things) The ATMega128 controls hyd cylinders that
      keep the machine frame level (or tilted to add super elevation) even on very
      uneven track.

      I do not need to be able to sense 360 degrees. I only need to sense
      something like +/- 3/8" over about 8". As I calculate it this would be
      about +/- 2.5 degrees.

      I don't need exact linearity, I can calibrate its characteristics to my use
      within the microcontroller.

      I would like to have a resolution of something like 1/32" in 8" (~0.224
      degrees).

      The signal may be analog, digital, PWM, I2C, etc. Frequency response is
      very minimal (below 10 hz).

      I need a "proportional" signal, not just "level" (i.e.: on/off, or "bang,
      bang" control).

      I am currently experimenting with an Analog Devices ADXL202
      Tilt/Accelerometer (+/- 2g) but I fear that it will not give me the
      resolution I need.

      There are some real nice ones out there for chips in the $30-$50 range but
      I'd like to keep it closer to $5-$15 ...

      Suggestions?
       
      Cheers,

      Chuck Hackett
      "Good judgment comes from experience, experience comes from bad judgment"
      7.5" gauge Union Pacific Northern (4-8-4) 844
      http://www.whitetrout.net/Chuck
    • AlienRelics
      Plumb bob with capacitive proportional sensing of differential position? Or LVDT sensing of position of same? Steve Greenfield AE7HD
      Message 2 of 7 , Apr 10, 2013
        Plumb bob with capacitive proportional sensing of differential position? Or LVDT sensing of position of same?

        Steve Greenfield AE7HD

        --- In Electronics_101@yahoogroups.com, "Chuck Hackett" <egroupscdh@...> wrote:
        >
        > I am building a machine to level 7.5" gauge railroad tracks (see:
        > http://www.cpgrr.org/). The device grabs the rails, lifts them into place
        > (height and "level") and tamps the ballast around the ties. To do this, the
        > machine needs a "level" reference. In curves the machine must also add
        > "super elevation" (raise the outside rail up to about 0.250" from level),
        > the amount of which depends on the curvature at that point. In my case the
        > frame of the machine is kept level by a sensor that feeds a microcontroller
        > (ATMega128). (Among other things) The ATMega128 controls hyd cylinders that
        > keep the machine frame level (or tilted to add super elevation) even on very
        > uneven track.
        >
        > I do not need to be able to sense 360 degrees. I only need to sense
        > something like +/- 3/8" over about 8". As I calculate it this would be
        > about +/- 2.5 degrees.
        >
        > I don't need exact linearity, I can calibrate its characteristics to my use
        > within the microcontroller.
        >
        > I would like to have a resolution of something like 1/32" in 8" (~0.224
        > degrees).
        >
        > The signal may be analog, digital, PWM, I2C, etc. Frequency response is
        > very minimal (below 10 hz).
        >
        > I need a "proportional" signal, not just "level" (i.e.: on/off, or "bang,
        > bang" control).
        >
        > I am currently experimenting with an Analog Devices ADXL202
        > Tilt/Accelerometer (+/- 2g) but I fear that it will not give me the
        > resolution I need.
        >
        > There are some real nice ones out there for chips in the $30-$50 range but
        > I'd like to keep it closer to $5-$15 ...
        >
        > Suggestions?
        >  
        > Cheers,
        >
        > Chuck Hackett
        > "Good judgment comes from experience, experience comes from bad judgment"
        > 7.5" gauge Union Pacific Northern (4-8-4) 844
        > http://www.whitetrout.net/Chuck
        >
      • John
        ... I would go with an accelerometer if you can, it has the big advantge of no moving parts. I appreciate your price concern, but would have thought that the
        Message 3 of 7 , Apr 11, 2013
          --- In Electronics_101@yahoogroups.com, "Chuck Hackett" <egroupscdh@...> wrote:

          > I am currently experimenting with an Analog Devices ADXL202
          > Tilt/Accelerometer (+/- 2g) but I fear that it will not give me the
          > resolution I need.
          >
          > There are some real nice ones out there for chips in the $30-$50 range but
          > I'd like to keep it closer to $5-$15 ...
          >
          > Suggestions?
          >  

          I would go with an accelerometer if you can, it has the big advantge of no moving parts. I appreciate your price concern, but would have thought that the cost of even the best accelerometer would be a very small proportion of the cost of the entire machine.

          John.
        • Chuck Hackett
          ... Or ... Disadvantages of pendulum: - Takes lots of room - Required critical damping to settle it quickly after a disturbance (i.e.: vehicle moves, vehicle
          Message 4 of 7 , Apr 11, 2013
            > From: AlienRelics
            >
            > Plumb bob with capacitive proportional sensing of differential position?
            Or
            > LVDT sensing of position of same?

            Disadvantages of pendulum:
            - Takes lots of room
            - Required "critical" damping to settle it quickly after a disturbance
            (i.e.: vehicle moves, vehicle stops, hydraulics tilts vehicle towards
            "level", etc.). This probably requires placing the pendulum, or parts of
            it, in oil to dampen it, and the oil bath must be enclosed to avoid
            spilling, etc., etc.
            - Still must build a "pickoff" device/sensor which adds complication

            > From: John
            > ....
            > I would go with an accelerometer if you can, it has the big advantage of
            no
            > moving parts. I appreciate your price concern, but would have thought that
            > the cost of even the best accelerometer would be a very small proportion
            of
            > the cost of the entire machine.

            I have added caps to the accelerometer I have to drastically lower its
            bandwidth (down to about 10 hz) and as the bandwidth goes down the
            resolution goes up (or the ability to detect the 'real' value out of the
            'noise'). It looks like I will be able to achieve a sensitivity of a bit
            more than 0.2 degrees.

            In my search I uncovered several nice devices at SparkFun.com, even a nice
            unit that combines a 3-axis accelerometer with a tilt-corrected compass in
            the same chip.

            BTW: The "cost of the entire machine" is pretty low considering that its
            parts (including solenoid valves, hyd pump, motors, steel, etc.) came from
            my "scrap box" :-)
             
            Cheers,

            Chuck Hackett
            "Good judgment comes from experience, experience comes from bad judgment"
            7.5" gauge Union Pacific Northern (4-8-4) 844
            http://www.whitetrout.net/Chuck
          • John Beaty
            Hi Chuck, I have been mulling over a somewhat related project with a 2g accelerometer. That range of resolution would work very well.   How many bits of
            Message 5 of 7 , Apr 12, 2013
              Hi Chuck,

              I have been mulling over a somewhat related project with a 2g accelerometer.
              That range of resolution would work very well.  

              How many bits of resolution do you use for an A/D converter to successfully read at that level?
              How much time is needed between readings to "see" the 0.2 deg change?

              Thanks

              John


              ________________________________
              From: Chuck Hackett <egroupscdh@...>
              To: Electronics_101@yahoogroups.com
              Sent: Thursday, April 11, 2013 9:31 PM
              Subject: RE: [Electronics_101] Re: "Tilt" Sensor



               
              > From: AlienRelics
              >
              > Plumb bob with capacitive proportional sensing of differential position?
              Or
              > LVDT sensing of position of same?

              Disadvantages of pendulum:
              - Takes lots of room
              - Required "critical" damping to settle it quickly after a disturbance
              (i.e.: vehicle moves, vehicle stops, hydraulics tilts vehicle towards
              "level", etc.). This probably requires placing the pendulum, or parts of
              it, in oil to dampen it, and the oil bath must be enclosed to avoid
              spilling, etc., etc.
              - Still must build a "pickoff" device/sensor which adds complication

              > From: John
              > ....
              > I would go with an accelerometer if you can, it has the big advantage of
              no
              > moving parts. I appreciate your price concern, but would have thought that
              > the cost of even the best accelerometer would be a very small proportion
              of
              > the cost of the entire machine.

              I have added caps to the accelerometer I have to drastically lower its
              bandwidth (down to about 10 hz) and as the bandwidth goes down the
              resolution goes up (or the ability to detect the 'real' value out of the
              'noise'). It looks like I will be able to achieve a sensitivity of a bit
              more than 0.2 degrees.

              In my search I uncovered several nice devices at SparkFun.com, even a nice
              unit that combines a 3-axis accelerometer with a tilt-corrected compass in
              the same chip.

              BTW: The "cost of the entire machine" is pretty low considering that its
              parts (including solenoid valves, hyd pump, motors, steel, etc.) came from
              my "scrap box" :-)
               
              Cheers,

              Chuck Hackett
              "Good judgment comes from experience, experience comes from bad judgment"
              7.5" gauge Union Pacific Northern (4-8-4) 844
              http://www.whitetrout.net/Chuck




              [Non-text portions of this message have been removed]
            • Chuck Hackett
              ... In my case the accelerometer puts out two signals for each (x & Y) axis. One is the raw analog signal and the other is a PWM signal. I ll be using the
              Message 6 of 7 , Apr 14, 2013
                > From: John Beaty
                >
                > Hi Chuck,
                >
                > I have been mulling over a somewhat related project with a 2g
                > accelerometer.
                > That range of resolution would work very well.
                >
                > How many bits of resolution do you use for an A/D converter to successfully
                > read at that level?

                In my case the accelerometer puts out two signals for each (x & Y) axis. One is the raw analog signal and the other is a PWM signal. I'll be using the PWM signals and feeding it to a timer pin on the ATMega to measure the pulse width.

                The data sheet seems to say that using the PWM you get a bit more resolution than what is available in the analog signal but it's not clear to me why.

                The part I'm using (ADXL202) is no longer available (at least I don't see it at Mouser or Digikey) but I'm sure there are newer chips that would work for you.

                Also check out http://SparkFun.com as they have several with eval/breakout boards. (they also gave gyros and compasses if you need them).

                > How much time is needed between readings to "see" the 0.2 deg change?

                With the added low-pass filters to reduce the response frequency and increase resolution the settling time is still well under sub-1/4 second which works for my application.

                Cheers,

                Chuck Hackett
                "Good judgment comes from experience, experience comes from bad judgment"
                7.5" gauge Union Pacific Northern (4-8-4) 844 http://www.whitetrout.net/Chuck
              • John Beaty
                The manufacturer s website suggests the ADXL212 which looks to be very similar and looks like it uses the PWM interface. Haven t studied it in detail, but is
                Message 7 of 7 , Apr 15, 2013
                  The manufacturer's website suggests the ADXL212 which looks to be very similar and looks like it uses the PWM interface.
                  Haven't studied it in detail, but is is an interesting approach that I had not seen.

                  Thanks for the information and suggestions.

                  John.


                  ________________________________
                  From: Chuck Hackett <egroupscdh@...>
                  To: Electronics_101@yahoogroups.com
                  Sent: Sunday, April 14, 2013 4:26 PM
                  Subject: RE: [Electronics_101] Re: "Tilt" Sensor



                   
                  > From: John Beaty
                  >
                  > Hi Chuck,
                  >
                  > I have been mulling over a somewhat related project with a 2g
                  > accelerometer.
                  > That range of resolution would work very well.
                  >
                  > How many bits of resolution do you use for an A/D converter to successfully
                  > read at that level?

                  In my case the accelerometer puts out two signals for each (x & Y) axis. One is the raw analog signal and the other is a PWM signal. I'll be using the PWM signals and feeding it to a timer pin on the ATMega to measure the pulse width.

                  The data sheet seems to say that using the PWM you get a bit more resolution than what is available in the analog signal but it's not clear to me why.

                  The part I'm using (ADXL202) is no longer available (at least I don't see it at Mouser or Digikey) but I'm sure there are newer chips that would work for you.

                  Also check out http://SparkFun.com as they have several with eval/breakout boards. (they also gave gyros and compasses if you need them).

                  > How much time is needed between readings to "see" the 0.2 deg change?

                  With the added low-pass filters to reduce the response frequency and increase resolution the settling time is still well under sub-1/4 second which works for my application.

                  Cheers,

                  Chuck Hackett
                  "Good judgment comes from experience, experience comes from bad judgment"
                  7.5" gauge Union Pacific Northern (4-8-4) 844 http://www.whitetrout.net/Chuck




                  [Non-text portions of this message have been removed]
                Your message has been successfully submitted and would be delivered to recipients shortly.