Subject:
|
Re: more on the rotation sensor
|
Newsgroups:
|
lugnet.robotics.rcx.legos
|
Date:
|
Sat, 29 Jan 2000 17:32:18 GMT
|
Viewed:
|
2313 times
|
| |
|
|
Ben (ben@ben.com) wrote:
> I did some more experiments last night. My conclusion is that the majority • of
> apparent 2-state transitions never actually occur (that is: the sensor does • not
> actually travel through 2 states between two consecutive sensor readings).
>
> The order of the states (as defined by legOS, in order of increasing raw • value)
> is: 2, 1, 3, 0. 0 is unambiguous: the sensor always returns 1023 under • all
> conditions. The other three are large ranges of values from the sensor. • These
> ranges (with my patch) are contiguous, because the battery level and length • of
> connecting wire skew the return values up and down.
>
> In order to move 2 states you either have to have a jump from 0->2 or 2->0, • or
> a 1->3 or 3->1. The 2/0 transitions never happen accidentally because they • are
> so far apart. What I think is happening is that in some cases the observed
> sensor value for a 1-state or a 3-state is landing right in the middle and
> being misinterpreted. When I captured the raw sensor values at the instant • the
> glitch occurred, they were almost always right on the upper edge of the 1 • range
> or the lower edge of the 3 range. These readings don't show up if you move • the
> rotation sensor slowly through all of the values (at the same battery • strength
> and with the same connecting wires).
>
> So the question is what to do about these transient readings. I still need • to
> gather more data to understand what's going on...
>
> --Ben
I think I have an idea what is going on.
Consider the sensor rotating forwards. If typical readings for states 0 to 3
are
3, 1, 0, and 2 (ordered as in your message) the input goes through the
series:
state : 0 -> 1 -> 2 -> 3 -> 0 ->...
input : 3 -> 1 -> 0 ->* 2 -> 3 ->...
As the input is an analog signal, if you read it while it is changing from
state
2 to 3 and the input is changing from 0 to 2 (marked with a * above) you may
get a reading corresponding to a 1. This will be interpreted as a move
backwards to state 1. The next reading will correspond to state 3 (which is
where the sensor really is) giving an apparent 2-state transition.
The thing to do about transient readings is to debounce them, that is to
take
successive readings until for example you get two the same. You can only do
this if you can read faster than the hardware changes state. I'm afraid I
don't
know enough about legos or RCX timings to suggest a scheme, but you should
be able to find one that works with a bit of experimentation.
Hope this helps
Philip Brown
|
|
Message has 1 Reply:
 | | Re: more on the rotation sensor
|
| (...) value) (...) The order is numerical, either 0->1->2->3->0... or 0->3->2->1->0... That's why legOS has them enumerated that way. The Mindstorms fw numbers them differently, which is why I mentioned it. (...) I don't think your specific example (...) (25 years ago, 29-Jan-00, to lugnet.robotics.rcx.legos)
|
3 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
