Subject:
|
Re: Fun with MindStorms Rotational Sensors
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Thu, 3 Jun 1999 09:53:18 GMT
|
Original-From:
|
S. Crawshaw <(sc10003@eng.)Spamcake(cam.ac.uk)>
|
Viewed:
|
1117 times
|
| |
|
|
On Thu, 3 Jun 1999, Ben Williamson wrote:
> I spent a while trying to build a PID controller, working from vague
> memories of an undergrad control systems subject. Feel free to jump in
> and correct me on this stuff, it's not my field and I hated the subject.
It _is_ my field (I do control system research), but I won't go into great
depth - I don't think this is the right forum!
I've tried this too - and it sort of worked (using just proportional
feedback I can control a shaft's position _fairly_ smoothly, with little
or no overshoot or oscillation). But it was pretty slow, and I didn't
even do a full PID controller.
Basically I reckon you're suffering from one or more of the following:
1 Sample time: because of the way the firmware works, you have no way of
knowing what the time between samples is (ie the time between "previous"
and "current"). Hence you can't possibly use this for speed measurements,
or at least, not accurately. For this reason I would expect any
controller using the differences to be unreliable. (Although if the device
being controlled is fairly benign, then it possibly doesn't matter)
2 Resolution: the rotation sensor, good though it is (low friction
especially) has only 16 values per turn. So to get 1 degree accuracy in
position measurements, it needs gearing up by 23 times... which adds
friction, delay and backlash - all bad in a PID controlled system.
I'm sure these problems could be overcome - although the sample time
problem probably means moving to LegOS or raw H8/300 programming!
BTW, does anyone know why Lego have discontinued the old gearing down
units? (These were sealed bricks with 1 to 20 (I think) gearing inside,
which had much less friction and backlash than a normal gear train)
Combined with the rotation sensor, this would give almost 1 degree
resolution, which would be enough for almost any robot project.
I'd better stop there - don't want to waste your phone bills! If anyone
really wants to talk more about control, email me off-list?
Stuart
--
Did you check the web site first?: http://www.crynwr.com/lego-robotics
|
|
Message has 1 Reply:
 | | Re: Fun with MindStorms Rotational Sensors
|
| I had two of these gearing down blocks, one of each type. They must both have stripped gears in side now because they both slip when under strain. The 2x4x3 square gearing down block must have used a worm gear because you could not use it to gear (...) (25 years ago, 3-Jun-99, to lugnet.robotics)
|
Message is in Reply To:
| | Re: Fun with MindStorms Rotational Sensors
|
| (...) Agreed, I went and bought one too. :) I'd like to get really smooth speed and position control of a motor using this thing for feedback. Motors with two speeds (go and stop) get boring after a while, all the really cool robots I've seen glide (...) (25 years ago, 3-Jun-99, to lugnet.robotics)
|
5 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
