• aditya mehrotra.

dyno update: yes! a dyno!!!! [updates]

So currently we have two mini projects going on as the semester wraps up. There's the hoppy leg we talked about last time (the 1DOF leg we will use to test the control system for chip before actually building it). And the second is a new addition! All of these will be open source on Github.



We found a dyno in the solar car shop and we're going to use this to develop the current/torque controller for the motor. We will command a current and see what kinds of torque we can produce, the repeatability, the accuracy to the theoretical Kt value and etc. The torque sensor we found is a Interface T-8 made in Germany rated to 20Nm. The shafts are 18mm and the Neo shafts are 8mm so we need to make an adapter for this. We went looking for the dyno when trying to determine how to best test the efficiency of the motor on solar car.

The above image was drawn for the solar car efficiency test but it's also a good way to explain dyno construction. Essentially we have two motors, one we're testing and the other one is fighting the first motor. The torque sensor sits in the middle and measures the torque between the two motors. These setups are used to measure torque-speed curves and so many other things. Between the motors and the torque sensor are misalignment couplers that allow for a slight amount of alignment imprecision. This is very important to not damage the sensor, the motors, or anything else in the system.


For us we're going to use it for two things:

  • Static locked-rotor tests to see if command current properly produces torque at the output shaft - and for controller tuning.

  • The same thing but not static, while the motor is spinning at certain RPM to see if the system behaves as we expect.

  • Test the system at 24V as opposed to 12V!!! See what changes and what torques we can produce.

We will be developing an IMPEDANCE CONTROLLER for the motors of the form Kp*Pe+Kd*Ve+T_ff. So that's PD w/ a feed-forward torque just like is on most legged robots with proprioception. We will leave the force-feedback development to hoppy leg.


OK enough preface, today we wanted to get the whole alignment couplers and etc setup. We couldn't find an 8mm to 18mm shaft misalignment coupler but we found and 8mm to 14mm on amazon and thought we'd put it on the late: https://www.amazon.com/gp/product/B078VRTFQJ/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1



The simple solution would be to take the diameter of the 14mm side to 18mm on a late. We used a boring bar for this. We did this only on one side because we kinda messed up one coupler and made the diameter a little tooo big...



For the other side, we decided to make an adapter between the too big diameter and the 18mm-18mm shaft coupler that was already on the torque sensor. The sensor works by measuring shaft deflection and a sensor determines the amount of torque we are applying. Adding other shafts may slightly skew this measurement. Since we care about shaft deflection we would prefer steel (but we only had Aluminium so we used that). But the other technique is to make this shaft coupling as short as possible as angular deflection increases as we go down the length of the shaft, a longer shaft will see more rotational deflection.

(https://www.linearmotiontips.com/mechanical-properties-of-materials-shafts-in-torsion/)


So making it as short as possible is the move here, essentially we left no material in-between the two misalignment couplers. We'll do some tests with this and see if it becomes a problem we'll do some math to account for it or we will make a new part. For now we think it'll be OK - we mostly want to start by seeing how the torque constant changes as we increase voltage and how accurate the torque constant is with the armature input current and if we can do feedback around this properly (or if we really need to do a clarke/park transform and measure the phase currents).



Here's the system parts turned down and then assembled together to produce the finished middle section of our dynamometer. We're calling it "Dino" of course. The next steps are to build the main frame and the mounts for the motors (we'll just order some from Andymark who has cheap ones for the NEO motors). Then put this whole thing together, we need to figure out how to mount the torque sensor itself to the frame, but that'll be a later thing. We'll also test the torque sensor later when we're not in shop.


#updates #omgrobots #chip3 #dino_the_dyno

3 views0 comments
© copyright 2019 | aditya mehrotra