Just had to tell all of you about this $59.95 three axis gimbal deal on EBay:
This seemed like a great deal and looks to be identical to DYS Gimbal on Hobbyking which is $200.00.
I got one and quality appears excellent though I haven't tried to use it yet (I will probably get another.)
I'm planning on converting to a Storm 32 for a laser rangefinder scanner.
Apparently DSLRPros has moved on and is wholesaling them out.
I get nothing from anybody for this, but if you want a real 3 axis gimbal, this is the best deal you'll ever see.
Heres a link to the manual for the DYS: http://www.getfpv.com/media/wysiwyg/documentation/3AxisbrushlessGimbalManual.pdf
Heres a link to the Basecam (Alexmos) Home page: http://www.basecamelectronics.com/
And a link to download the appropriate 8bit Alexmos Gui: http://www.basecamelectronics.com/files/SimpleBGC_GUI_2_40b7.zip
And a link to download the Alexmos manual PDF file: http://www.basecamelectronics.com/files/v10/SimpleBGC_manual_2_4_eng.pdf
(This does include the Alexmos 3rd axis expansion board.
Best regards to you all,
Gary
Comments
Hi LD,
That makes perfect sense.
I really like your analogy to a petrol engine misfiring, if the PWM isn't updated at the right moment or an edge timing is off, that is exactly what it would be like.
I made my own optical encoders years ago with a decode based on straight flip flop logic and there was always uncertainty during reversing and max speed was set by the characteristics of the TTL logic.
One of the problems with our current Ardu, Android and Linux everything methodology is that everybody is solving hardware problems with software.
As you point out that can be really sub-optimal and at least requires a really strict RTOs to make it work on a heavily time dependent system.
I can easily believe a dedicated FPGA could greatly improve things.
Of course for stabilization positional accuracy isn't so critical because it is a closed feedback loop with the IMU and I would suppose it is still reasonable to produce good positional accuracy based on offsetting the IMU coordinates at least within the limitations of the Gyros on the IMU.
However, I still think the brushless gimbal motors themselves are really a kludge, basically a under designed stepper that is being fooled by the analogue equivalent of microstepping into behaving roughly like an AC or DC servo motor, using it's dedicated IMU for maintaining absolute position.
These are really just rewound normal hobby grade brushless motors and they make much better brushless motors than they do gimbal control motors.
I favor top down design and this is really one of the most blatant examples of simple adaptive bottom up design I have ever seen.
The fact that they managed to make them work has caused them to propagate everywhere, but the basic concept is heavily flawed and the control motors themselves really need to be seriously rethought.
The method of using them closed loop with the gyros, seems excellent, but even there as you point out, you really need a robust, fast and precise feedback loop to make it work properly and imprecise software timing certainly isn't it.
One of the problems we have in this business is that some guy at home manages to cobble something together that sort of works and immediately everybody promotes it and copies it.
Not to in any way detract from AlexMos, but that should have been the start, not the end.
As for Chris and Euan,
I do think you can get this to work OK with a GoPro, probably more easily in 2 axes than 3 and at least somewhat due to the increasing software timing problems dealing with 3 axes versus 2.
I would strongly suggest, aside from a lot of finicky "tuning" of PIDs and ramped up current during actual moving, that you do as good a job of anti-vibration mounting of the gimbal as possible.
The Storm32 uses a faster more robust controller and ostensibly provides more robust control but I haven't had a chance to hook up mine and try it yet.
From what LD has said, it does seem like it might help however and I will give it a try when I get a chance.
Best Regards,
Gary
ok...so inherent brushless limitations aside, who has actually managed to get this gimbal working with a go pro, to an even acceptable level?
+1 Gary. From our analysis it seems the biggest limitation of the Alexmos clones is that they are changing the PWM values (the equivalent of setting a position) under software control. In a constantly rotating or accelerating system the timing of the PWM updates is erratic and leads to less than optimum waveform generation. Even a very small timing error, the equivalent of one or two software instructions when entering the interrupt routine, is enough to disrupt the phase relationship of the magnetic fields, dramatically reducing the torque. It's like a petrol engine that is misfiring.
We were able to create a more precise timing system using FPGA hardware and nearly doubled the available torque. Surprisingly, this brings the torque and speed performance of the gimbal motor to the same or better than a stepper motor of the same weight and current consumption. This doesn't make the absolute positional accuracy any better as this is a function of the number of poles, but getting more than 2000 steps per revolution is quite possible.
Hi Euan,
A few Observations.
My laser scanner is working OK in stabilize, but settings would not be ideal for a GoPro because my laser rangefinder is much lighter.
The X/Y scan rate is not high enough to be useful with these gimbal motors, they simply can't provide sufficient torque to achieve a decent scan rate.
I think the gimbal motors themselves are a bit anemic for any use.
You can turn up the motor current, but then they get steppy or twitchy.
I think it takes quite a bit of trial and error adjusting values to get the smoothest possible operation,(preferably mostly while you have the gimbal hooked up to your PC while moving the copter around to visually judge the smoothness of the compensation).
This gimbal has made me aware what a half ass mechanism our brushless gimbal motor based gimbals really are.
A gimbal motor is basically a wide step stepper motor using the analog equivalent of microstepping to maintain intermediate (between pole) positions.
The result of this is a compensation mechanism that can barely work for stabilization because it is a real time feedback loop with a gyro for holding it steady, but which is not well suited at all for fine and precise pointing.
And the gimbal motor design is intrinsically low torque and worse, variable torque between poles.
Bigger gimbal motors with more poles work better, but an honest 200 step stepper motor with good microstepping should work a lot better (albeit probably using more power and possibly weighing more).
And AC brushless servo motors seem like they would be the ideal, although they are seriously power hungry.
Basically the current brushless gimbal motor is a travesty of motor design.
Best Regards,
Gary
Has anyone got a good tune yet? Anyone want to share their "almost there" settings?
Im looking how to connect pitch and yaw to my RX so I can trim/tilt when needed.
What do I connect and where, and are there any setting I need to edit for this?
Thanks in advance :)
Well that was a short lived fix, it seems as though when I cycle the power to the gimbal it goes back to its crazyness. I'm going to see if I can get that replacement.
Thank you, that seemed to help stabilize her. Uggh that was a huge pain.
The sensor axis TOP and RIGHT don't look right to me. Could be wrong though.
Going off the MPU's axes, I think RIGHT should be +Y (facing the camera) and TOP should be +X.
-
1
-
2
-
3
-
4
-
5
of 7 Next