Update 01.09.14

 

Update 01.09.14

 

Even though there's been a long time since last update doesn't mean that nothing has happened! Before I traveled home for the summer vacation, I milled and drilled the tailcone so that the brackets that hold the shroud could be installed. There was a lot of work making everything fit together, but it worked out well in the end. While I was home this summer, I soldered the electronics for the throttle, and after I returned to Trondheim in mid August I've been working with mounting the motor in the tailcone and making a bracket that will hold the controller. After this was done, I was able to test the assembled tailcone in a lab to see how much power it draws at no external load. It turns out that there is, as suspected, a lot of eddy current loss in my magnetic coupling. In addition to all this, I have been ordering all the parts for the battery pack. I ended up ordering 16 pcs 10Ah LiFePO4 cells that I will wire in series, yielding a capacity of around 480 Wh.

In order to drill the holes in the brackets that hold the shroud, I had turn the head of the mill 90 degrees. It was a little difficult to set up the brackets perpendicular to the drill axis, but it turned out well.

 

 

Some creative workholding was needed to set up the tailcone for milling the mating surface where the shroud brackets will be bolted. I messed up a little on the first pass because I set the height inaccurately. Fortunately, no great damage was done to the part. Once the mating surfaces had been milled, holes were drilled and tapped in the tailcone. Next, the rest of the holes in the brackets were drilled and tapped.

 

 

 

Everything assembled. This was a real milestone in the project, very rewarding seeing the final product take shape. I even looks like a real DPV!

 

The angle between the tailcone and brackets was slightly off, so I had to make some angled shims from some 10 mm PVC plate I found in the workshop. With these in place, everything fit together nicely.

 

 

Testing out the grip. Feels good.

 

This was all done before I went home for the summer. When I came back to Trondheim, I started right away making the motor mount and controller bracket. I didn't take any pictures of making the bracket, but it wasn't that interesting anyway. A rectangular piece of 1,5mm aluminium plate was cut and hammered into shape.

 

 

An angled drill was used to drill holes in the controller bracket and tailcone. This was a nice tool for this job.

 

 

This is how I will mount the controller in the DPV. It is placed so that it will provide some countertorque to the propeller when the DPV is running.

 

 

Next up was installing all the electronics in the throttle. Here you can see the reed sensor to the left. The hall effect sensor is glued to the round metal plate in order to increase the magnetic flux that passes through it, enabling the full range of the sensor to be output. With the plate installed, the hall effect sensor now outputs the sufficient range that the controller requires.

Once all the electronics were installed, I was ready to run the motor with the propeller to see how it performed under no load. The lab I was using didn't have power supplies powerful enough to deliver 48v at more than 3A, so I had to use a 31.4V 10A power supply instead. This means that I didn't get to run the motor at the same voltage I will be using when I'm finished, but I did get some numbers with respect to the inefficiency of the magnetic coupling.

 

 

This is an overview of the data I gathered during testing. The graph to the upper left is the most interesting - it shows the eddy current losses with respect to the speed of the motor. As you can see, the losses are very high. At 750 RPM, almost 160 W is lost in the coupling.

If the measured results are extrapolated, it is revealed that over 250 W will be lost at 1000 RPM which is the rated speed of the motor. This is over half of the rated power of the motor, and is totally unacceptable. Obviously something had to be done to reduce or eliminate these losses. The easiest way would be to reduce the thickness of the barrier, but this would still result in some losses. The best way in terms of losses would be to remove the aluminium barrier and replace it with something non-magnetic, for example an engineering plastic of some grade. This would require a lot of complex machining, as well as the difficulty in joining the two parts. Since plastics have a very low modulus of elasticity, it is hard to ensure a watertight mate between the two parts. I did some further simulations to see how the barrier wall thickness would affect the eddy current losses.

 

 

It is seen that the eddy current losses are greatly reduced by reducing the wall thickness. Note that the simulations are done with a constant air gap of 1.5 mm. This means that the magnets will be placed closer together, enabling a higher torque to be transferred. I don't need more torque, so by increasing the air gap I will get even less losses. By using a wall thickness of 1 mm and air gap of 2.25 mm, as well as compensating for the difference I got between the original FE-model and the measured data, I get:

 

 

Although losing 100 W sucks, I chose to go for this alternative instead of the complicated solution of machining a non-magnetic barrier.

 

 

Finally, I have ordered all the parts for the battery pack. I have decided to build a 48V 10Ah pack from LiFePO4-cells. The cells I will be using are called Heter 42110. I also decided not to use a BMS (Battery Management System), instead I have invested in a balancing charger (iCharger 208b+) and some cell monitors that will make a loud beep when the cell voltages dip under a pre-programmed value. I am just able to cram 16 cells into the body of the DPV, which is just enough to provide 52.8V of rated voltage. I will need to build a holder for the batteries, and for this I plan to 3D-print some plates that the batteries will be sandwiched between.

 

This is a first draft on how the battery pack will look. Last, here is a picture of how an explorer edition of my DPV would look. Thank you for reading!