Log in

No account? Create an account

Previous Entry | Next Entry

Encoder 360

After some feedback from some folks over in tugrik's journal, I modified my encoder-wheel generating code to make one with exactly 360 positions. Here's my backtracer's rendering of the new toolpath.

Encoder Wheel 360


( 14 comments — Leave a comment )
Feb. 9th, 2007 11:57 pm (UTC)
I don't think there's any real reason it needs to have 360 positions. The commercial HP shaft encoders I used on my robot were 512 position.

Cool to make it your self tho, and neat to see the tool path.
Feb. 10th, 2007 12:00 am (UTC)
Yeah, given my code handling is all floating point anyways, I don't really see a reason not to use 512 positions. I just thought it'd be interesting to see the modified toolpath.
Feb. 10th, 2007 12:49 am (UTC)
I just thought it cool that it could be done in something other than base 2.

I'm gonna have to go play with some code now and see if I could
figure out a sequence that would support 0 to 9 cycling.
( 360.ddddd for fixed point )

Yeah, I know it's anti-floating point (almost anti-geek) in a very COBOL way.

But it's still cool.

- Krin ^ _ ^
Feb. 10th, 2007 01:21 am (UTC)
Re: S999V9999
0 : 0010
1 : 0110
2 : 0111
3 : 0101
4 : 0100
5 : 1100
6 : 1101
7 : 1111
8 : 1110
9 : 1010
- wrap -
0 : 0010


Now, why is that neat and why not just do it in the CPU?
(retorical question, Revar! Put away the 2x4. Please, bat-bat?)

For those that really get into hardware, you can set up logic gates
directly on the switches and drive a LED display with full numbers
directly off the encoding wheel. No computer interface needed.
(I seem to recall some really BIG ones sitting around somewhere, eh?)

For the ultra geek reference, you can model XOR gates using pressure
valves and do it all with STEAM!



- Krin ^ _ ^
Feb. 10th, 2007 01:57 am (UTC)
Re: S999V9999
Actually, it's pretty easy to generate the Gray codes for any arbitrary _even_ number of steps. In fact, to do the 360 position encoder wheel, I just genericized the generator code. I can make wheels for just about any arbitrary even value. I suspect, though, that some will make wheels that are susceptible to breaking, with small connecting tabs. In particular, numbers only a small amount larger than a power of 2 should make for small tabs.
Feb. 10th, 2007 03:17 am (UTC)
Re: S999V9999
Yeah, I relized after drawing that up that if you center around the middle of the range then any even multiple would work.

- Krin
Feb. 10th, 2007 02:11 am (UTC)
Re: S999V9999
It's still in base 2; just not counting to a power of 2. For bases other than 2, you'd need to use coloured filters instead of holes };>.

Feb. 10th, 2007 03:16 am (UTC)
Re: S999V9999

Up, Down, Ground.

If you wanted enough values, you could use relative amplitude.

We get enough fidelity, we would have to use some ridged but easily
maluable material.... like say... vinyl?


- Krin
Feb. 10th, 2007 08:02 am (UTC)
Oooh. Binary made physical. A Cantor dust drawing itself around the wheel.

Feb. 13th, 2007 05:02 am (UTC)
Dear One, it looks like a beautiful piece of art--and I'm sure glad you know what to use it for. Some one of these times you're going to have to explain it to me--in very broad terms.
Love, Mom
Mar. 31st, 2007 06:43 am (UTC)
Whazzis for? I mean it looks cool... but I don't get it.
Apr. 6th, 2007 06:02 pm (UTC)
Basically, with that hole pattern, and nine light sensors to watch them, the rotational angle/position of the disk is known exactly to an accuracy of 1 degree.
Apr. 9th, 2007 08:44 am (UTC)
It's a clock! But why are the holes lined the way they are, instead of in a standard binary sequence? For balance?
Apr. 9th, 2007 09:01 am (UTC)
The holes follow the pattern of "Gray Codes" which have the nice property that for any transition only one bit will ever change at a time. That prevents problems with sensing two-bit changes where one bit's sensor may be very slightly misaligned, reacting later than the other, causing false results.
( 14 comments — Leave a comment )