Direct Drive Motor Testing, Testing, 1,2,3...
12/21/2007, by Dan Gray.
Click on the images for a larger view.
This motor was designed and made by Dave Rowe. He shipped me the
rotor and stator, and so I put the motor together with more plywood,
and a large ball bearing for testing. I put a Gurley 320,000 tick
per rev encoder on it for preliminary testing (I stole the encoder from
my 14", I sure hope it stays rainy for a while! The final motor
will have 12,000,000 ticks per revolution or so.
The idea is to completely eliminate all the gear train, which will eliminate backlash, periodic error, gear train flex, etc.
First, it would be worthwhile to see the wiki page that Dave Rowe
created about this motor. The motor in the pictures is the same
one here. Dave Rowe shipped it and I received it on
The Alt/Az project wiki, page about Direct Drive Motors
In this test, and in the video, I'm running the motor at about 30 deg's per second!
Notice the white strapping tape on the left side of the rotor disk.
There is a pulley and weight connected to the other end.
So, the motor encoder is used for two purposes.
1. It's used for the PID control of the motor, to keep the motor angle constant.
2. It's used to find the motor electrical angle, so the pwm can be applied just right to the 3 windings.
is my workbench. Marion and I share it, and she hates it when I
do projects like this! Sure are a lot of wires, eh?
Here's the 3 digital meters. I have one lead in series with each
winding, so I needed 3 meters to figure the torque constant.
The two boards on the left are both standard SiTech servo controllers.
I modified the one on the right to control the synchronous ac
motor. Notice the extra H-bridge hanging from the right hand side
(behind the ribbon cable).
connecting the motor as a Y connected motor, we are using an H bridge
driver for each winding. I found out that this works much better
than a Y connected motor. This is because, so far, we're not
using current feedback to do the PWM (yet).
You can see the 8.6 lb weight hanging from the pulley here.
Final torque test is 6.3 average nM per amp. I also have too much gap, so Dave thinks we can increase this by 20%.
I stated 9.1 first, but I found a mistake in my spreadsheet.
Left to do:
1. Test with large mass
2. Test using 10 bit PWM instead of 8
3. Test using the 12,000,000 ticks per rev encoder
4. Design and make a new circuit board that will control this motor!!!
5. Create and test additional firmware for the new circuit board.
Here are two videos that show the test setup and I'm saying a few words too. Notice the dog food!
Movie No. 1, (20 megs)
Movie No. 2, (16 megs)
The following spreadsheet has been corrected, and gives the 6.3 nM per
amp. I initially made a mistake while moving columns around.
Here is a (corrected) spreadsheet with the torque test results
Here's the first movie I made yesterday. When this movie was
shot, one of the phases was connected backwards (unknowingly), and it
worked reasonably well like this, but once Dave and I figured that out,
it works much better. We were only able to get about 3.5 nM per amp
when connected wrong.
These are all the same movie, just different formats.