Actually ran this test series yesterday and analyzed the data today. It is becoming more apparent that acceleration is a factor we need to take into account. We also will need intensify adjustments based on how well the last adjustment responded (i.e. if we call for 300°/s² and get 150, we need to double whatever we did in the last frame).
It appears the transducers are more responsive than the 250ms latency I have seen published. I am bumping up the master timer to 125ms to see if that fixes some of our weird occilations.
There may be an fundamental design problem with the actuator attachment points. In the design I thought that since the center of rotation of the double-cardan main joint was 5 inches below the bottom sheathing that the centers of rotation of the pitch/roll actuators needed to also be at 5 inches. What I am seeing now is pressure on the linear bearings (especially roll actuators) as they try to “straighten” with the now translated extended rotation point.
Is this the reason other 3DOF designs use 3 actuators instead of 4? I thought it was to save coin and that the calculations for 4 actuators is actually easier since you have 2 dedicated to pitch and 2 dedicated to roll.
Yes, I know I didn’t document any findings from day 3 and that is because it was a complete regression. We had some luck today and added a new feature – data recording!
We are definitely getting somewhere… with the goal to put weight on it soon. Now I just have to make sense of the motion tests 5-26-2012 data. I will need to add a separate “weight” slider for roll (lateral) than pitch (longitudinal) because the roll actuators are much more responsive. I also need to figure up a minimum “overvoltage” to break static friction and get the platform moving on command. The aggression should probably be divided by 10 so that we can select a more fine tuned value.
On another note, testing these old Planar Toshiba LTM10C942 (640×480 10″) LCD monitors that run at 75Hz with the AV-620 works great (they don’t advertise the 640×480 because the Windows 7 UI doesn’t show it as an option but if you drill into Advanced Settings -> List All Modes you will see it in there.
Still waiting on the replacement transducer (don’t get me started) I thought I would catalog the parts I have left over from the Audrey and start layout out the consoles.
I disassembled and reassembled the faulty transducer. Nothing visibly wrong but it must be the circuit board is fried. They say the NEMA case can be installed at any angle but what I found is that the hole in the conduit pipe threading where the electrical pigtail comes out is not watertight and leads directly to the circuit board. I don’t think that caused the failure in my case but it is worth point out for others that find this article.
I also tried switching to the second set of in/out ports thinking maybe one side is completely clogged – no dice. So I ordered a less expensive replacement (although the 900X is still readily available, I opted to try the 500X series). The mechanism is electromagnetic instead of a piezoelectric. It also wastes less air and has more known variables like the relief CFM rate. I hope I don’t like it too much and want to replace the remaining three 900X units!
I left off on Saturday with the two roll transducers working sporadically after centering the calibration screws. I bought an air pressure gauge but my old one is just as accurate so I think I will take the new one back to Home Depot.
This evening I booted up and behold – the right roll transducer was working good! I did a field calibration on it but it wasn’t far off. I couldn’t get the left one to work. It is sporadic. If I apply signal voltage it will start pushing air but if I turn off the signal it will not stop. Span calibration seemed to have no effect. Zero calibration was sporadic (kept getting a pressure signal with no signal voltage).
I ordered a 500-EH replacement from the Valve Store Online. Good prices, reasonable shipping rates, and they even take PayPal. The 500-EH only has a 12ma at +10V lead and ground (no supply voltage loop). Also, the air consumption is next to nothing. The mechanism is different than the other ones but it will let me go up to 145 PSI if I wanted to. Depending on how much I love this new unit I will look forward to replacing the old ones when the time comes.
Finally, the control program. It isn’t smart enough. I basically need the autopilot kind of trend analysis and trim logic for the platform. If it is off balance it never fully recovers to level… ever. So I guess I will work on that while the new transducer is on it’s way. EDD is 5/7.