No announcement yet.

Torque measurement of rotating shaft

  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    Way back in the late '60s when I was working for Avco-Lycoming, we measured the torque on the engines under test. A water brake was mounted to a ring supported by four short beams. Each beam had strain gages bonded to two sides, and their values changed as the beams flexed from the torque absorbed by the brake. The beams and their mounting rings were machined from one piece of steel tubing. The systems were very durable and reasonably accurate.

    You could probably substitute a dial indicator for the strain gages for a less complicated system.
    Any products mentioned in my posts have been endorsed by their manufacturer.


    • #32
      Did someone mention this already? If measuring the twist is what is required that is easily indicated by triggering a strobe at one end of the shaft and aiming the strobe at a scale stuck on the other end of the shaft.


      • #33
        Originally posted by darryl
        I wonder how much twist a suitable shaft can withstand and still return to its original 'alignment'- for what it's worth, I just had a thought, so I'll contribute it-

        Have the piece of shaft, the drive couplings on both ends, and two pieces of reasonably close fitting tubes placed end to end between the couplings. Clamp the ends of the tubes to the shaft at the ends where the couplings are. While at rest, scribe a mark across both tubes where they meet. If you twist the shaft, those marks will go out of alignment. How far out is determined by how much you twist the shaft.

        You can calibrate that by putting a known force on the end of a known length of bar that's attached to one end of the shaft. Scratch the mating marks across for each torque value you apply. In this case, you might want a range up to maybe 1200 ft/lbs if you'll be routinely measuring roughly 800 ft/lb.

        In operation, you read it using a strobe light to freeze the markings, or you could trigger the strobe from the rotating shaft.
        Well described! Such a mechanical arrangement is often used to measure torque. However, rather than the strobe and index marks, (not a bad idea for a manual setup) a pair of gears are fitted on the ends of each tube in place of the index marks. A pair of gear tooth sensors are placed adjacent to the gears and the phase difference of the sensor outputs is measured with a circuit such as Evan suggested, or more simply with an op-amp circuit (RPM needs to be factored in either case) feeding an analog meter movement - perhaps more appropriate in the Peterbuilt application that Guido has in mind.
        Weston Bye - Author, The Mechatronist column, Digital Machinist magazine
        ~Practitioner of the Electromechanical Arts~


        • #34
          Originally posted by Evan
          Yes it was. It's called a propellor. That is exactly how some dynomometers work.
          OK. the shaft is turning a propellor... NOW what's the torque?

          Didn't think that helped much.......

          Aside from that being entirely encompassed and made totally *obvious* by my prior statement about ship power measurement.... you are just transferring the calibration problem to a different point on the shaft.....

          if you REALLY want a measurement, as I and a couple others have pointed out, you may get an absolute measurement by measuring the reaction at the engine mounts..... At that point you don't need any shaft dingamusses.... but it may be a lot less convenient in certain cases.

          BTW, according to my shipbuilding text, it was not unknown to calibrate the shaft, or a section of it, at the works, so that later measurements could be accurately made...

          A force and lever are a means of absolute calibration. Any known force and known lever will do if it provides a measurable deflection in same conditions as in use.

          Keep eye on ball.
          Hashim Khan


          • #35
            OK. the shaft is turning a propellor... NOW what's the torque?
            You sure are going to great lengths to try and cover up your mistake. Reminds me of a cat trying to scratch dirt over some *hit on a rug.

            The torque is whatever the drive system is rated to produce. I am sure somebody knows that. The propellor is the load. Duh.
            Free software for calculating bolt circles and similar: Click Here


            • #36
              Originally posted by Evan
              You can measure the torque on the output end of the shaft with a dyno. That is all you need to know to calibrate the system. If you don't understand that I can't make it any simpler. That doesn't give you information about the properties of the shaft since it doesn't even tell you the amount of deflection that takes place. You don't need to know that.

              The only information that you obtain is a phase difference between input and output. That is enough.
              If you have a dyno, what is the point in dicking around with reference marks on a shaft. Read the torque off of the dyno.

              If the OP had a dyno, he wouldn't have needed to start this thread.

              I am puzzled, isn't the phase difference a function of the deflection? In that case using a dyno to determine what phase difference corresponds to what amount of torque IS telling you something about the shaft...



              • #37
                Originally posted by Guido
                Been brainstorming this problem and a solution for years. Should have been more detailed with the problem of wanting to measure/display HP and torque on the dashboard of a Peterbuilt moving down the highway at speed. Idea popped up when we saw how turboprop engines display torque in real time. A nonlinear pressure guage is mounted on the panel, with pressure from engine compressor stage number so-and-so measured and displayed. The pressure guage displays in ft-lb of torque.

                There's gotta be a market for a device to be added to an over the highway diesel truck tractor, which would reliably display torque/HP. Instant operator visual of motor performance.

                Everyone's thinking and input is appreciated.

                The most accessable place on the Peterbuilt would be the driveshaft between the transmission and the differential. Anyplace else would require considerable reengineering of the transmission and/or engine. Torque reaction motor mounts would have to contend with gravity and inertia factors. The driveshaft location suggests a bolt-on aftermarket solution.

                Some engineering would be necessary to size components properly and choose the appropriate materials. The torque element would have to be sized to twist a little bit, but not yield. Perhaps some torque limiting positive stops. A balancing act of engineering - I've seen stock drive shafts twisted in two.
                Weston Bye - Author, The Mechatronist column, Digital Machinist magazine
                ~Practitioner of the Electromechanical Arts~


                • #38
                  In that case using a dyno to determine what phase difference corresponds to what amount of torque IS telling you something about the shaft...
                  The dyno is for calibration only. You don't leave it attached to the system. Tell us then what knowing the phase difference tells you about the shaft? It doesn't tell you how much it twists since that hasn't been measured and we don't need to know that. All we need to know is what degree of twist corresponds to the established torque limit.

                  Rpm differences can be automatically factored into the phase difference measurement by just using a tachometer output from the engine as the phase clock. There is no need to develop any sort of compensation table or other complication.
                  Free software for calculating bolt circles and similar: Click Here


                  • #39
                    Originally posted by vincemulhollon

                    Easy way to remember the HP equation, BTW, is to just memorize 5280 aka feet in a mile. Its close enough.

                    And I thought the "easy way" was to count the number of feet and divide by 4.........Of course, that assumes each horse has 4 feet.........

                    I thought this thread needed a little humor........



                    • #40
                      Measuring the shaft rotational deflection might work, but you should keep in mind that there will be a lot of vibrational and harmonical related components in that measured output. Those likely would have be well filtered to get a meaningful measurement, I don't think just a simple digital display of the difference in rotation would be useful, it would just be flying all over the place.

                      Its possible that if the overall speed of the shaft is pretty constant that a simple weighted or sliding averaging method would work.

                      I've worked on developing instrumentation for dyno's, and handling measurement noise and vibrational components in the measurements is a significant issue.

                      If I had a choice and its practical to implement I'd go for the load cell/beam approach, I think there would be less measurement noise to deal with there versus measuring the shaft deflection, particularly if the shaft has a reasonable safety factor.

                      If you do decide to try the shaft deflection method you may not have to use encoders to do that. If there is anything with a reasonable radius on either end of the shaft (or if you can add "measurement" wheels) just glue magnets on the outside radius of the wheels and set up proximity switches to catch the magnets going by. Then you just need to time the difference between the switch transitions. This could be implemented pretty cheaply using a microcontroller with an internal timer/counter.

                      I still think this output would need to be filtered or at least averaged to get a useful result.

                      Paul T.
                      Last edited by PaulT; 04-14-2010, 01:56 PM.


                      • #41
                        The problem has been solved for many years. See for many different designs to fit nearly any mechanical configuration.



                        • #42
                          Drive shaft torque gauge.

                          Old P&W "C" 12x36 lathe, Covel SG, DoAll T&C Grinder, Rottler SG9M, HP7A, F69ATC


                          • #43
                            In the case of the truck there will be some movement in the suspension in response to torque and that may be a more convenient place to mount a strain or movement sensor.


                            • #44
                              Which means you need to know something about the shaft, ie its torsional stiffness.


                              Originally posted by Evan
                              Nope. All you need to know is how much twist is produced by a particular amount of torque on the output.


                              • #45
                                No you don't. You only need to calibrate whatever phase difference occurs with a certain applied load. You don't even need to know what material the shaft is made from. It could be metal or wood or fiberglass or day old french bread. Why is this so difficult to understand?

                                X load produces Y twist. Period.
                                Free software for calculating bolt circles and similar: Click Here