Post by thread:Code For Bourns ACE-128 Encoders

Happy Xmas All, Just for once, not a problem mail! I've recently been using Bourns' new ACE-128 rotary encoders. (For those of you who haven't meet them, info is on my web page shown below) These give out a type of 'grey-code' to give the position of the encoder. Bourns say to just use a 256 byte look-up table, but as that can be upto 1/2 a PIC's memory, I've writen decoder routines that are shorter (at the expense of speed). In case anybody else out there is using these beasties, I stuck 4 of the routines on my web page. For further info, or to see the routines, the web page is:- http://ourworld.compuserve.com/homepages/steve_lawther/bourns.htm If anybody uses any of the routines, and / or improves on the code size, for the speed specified, could they let me know - I've spent the best part of a week trying to optimise them, but I isn't perfect! Hope this helps somebody, Steve Lawther PS - I'm not in any way linked with Bourns, and the ACE-128 is not a solution to the world problems - but it has its uses! *** Opinions are my own, not my employers, nor clients *** *** For More PIC stuff, see *** ourworld.compuserve.com/homepages/steve_lawther/ucindex.htm

> Just for once, not a problem mail! > I've recently been using Bourns' new ACE-128 rotary encoders. > (For those of you who haven't meet them, info is on my web page > shown below) > These give out a type of 'grey-code' to give the position of the > encoder. Bourns say to just use a 256 byte look-up table, but as > that can be upto 1/2 a PIC's memory, I've writen decoder routines > that are shorter (at the expense of speed). In case anybody > else out there is using these beasties, I stuck 4 of the routines > on my web page. > > If anybody uses any of the routines, and / or improves on the > code size, for the speed specified, could they let me know - I've > spent the best part of a week trying to optimise them, but I isn't > perfect! Hmm... I've looked at your web page and I must say that encoder sure does have a wierd coding sequence, doesn't it.... :-) Methinks I'll have to brainparse that one a little bit...

> Hmm... I've looked at your web page and I must say that encoder sure does > have a wierd coding sequence, doesn't it.... :-) Methinks I'll have to > brainparse that one a little bit... It's probably using Gray code to encode the shaft angle of the encoder. Gray codes have the property that only one bit changes in going from one state to the next. This prevents errors, since you don't have clock new samples to make sure you've got a stable value; you need only look for a valude which differs from the previous one. You don't get this same effect from a binary-encoded ripple counter. There's a simple algorithm using XOR operations which can be used to translate between gray code and "normal" binary encoded values. Actually, it's a pretty simple circuit as well using XOR gates. I'd refer you to "The Art of Electronics" by Horowitz and Hill. ISBN 0-521-23151-5, but I know there's a new edition subsequent to this one. louie

On Sat, 21 Dec 1996, Louis A. Mamakos wrote: > > Hmm... I've looked at your web page and I must say that encoder sure does > > have a wierd coding sequence, doesn't it.... :-) Methinks I'll have to > > brainparse that one a little bit... Just take a walk on the wild sides of the 8-cube. No need to see all of the sides, just turn once on every corner. Voila, the turning of the shaft is in a roundtrip of the cube. No, I am not on drugs and yes, I have had a good night's sleep. The hallucinatory stuff is also known as mathematics. A Gray code can be found by tracing a Hamiltonian cycles on the graph of the n-cube. For you folks exclamating "now he's doing it again!": an n-cube is the n-dimensional brother of 2-d square and 3-d cube, a graph is a network consisting of points and lines (just like a schematic, but without the parts), and a Hamiltonian cycle is a path over the lines of a graph, traversing every point exactly once. Not every graph has such a cycle, but every n-cube graph does. In practice, you label all the points of an n-cube graph with their respective "coordinates", like in: 100 101 *-------------* |\ /| | \110 111/ | | *-------* | | | | | | | | | | | | | | *-------* | | /010 011\ | |/ \| *-------------* 000 001 Now, if you go point to point, only one coordinate changes at a time, making for very unambiguous state transitions, as opposed to a straight binary scheme. A Hamiltonian cycle would be: 000 -> 100 -> 110 -> 010 -> 011 -> 111 -> 101 -> 001 (-> 000 ) Note that this method is very easyly extended to higher values of n (if you can't figure it, notice how this one is already an extension of n=2.) > There's a simple algorithm using XOR operations which can be used to > translate between gray code and "normal" binary encoded values. Actually, > it's a pretty simple circuit as well using XOR gates. I'd refer you to > "The Art of Electronics" by Horowitz and Hill. ISBN 0-521-23151-5, but > I know there's a new edition subsequent to this one. Well, why not post the algorithm, it's an interesting thing to put in a pic. BTW, I got mine from "Graphs, an introductory approach" by Wilson and Watkins, ISBN 0-471-51340-7. For all you engineers thinking math is only integration and differential equations, this is a definite gem. If you hate math, get a copy of "Alice in wonderland". Also makes a great Christmas gift to newborns :-) Joost

Louis A. Mamakos <spam_OUTPICLISTTakeThisOuTMITVMA.MIT.EDU> wrote: > [The Bourns encoder is] probably using Gray code to encode the > shaft angle of the encoder. > > [long explanation of Gray codes deleted] Louie: The Bourns encoder does NOT use Gray codes. If you'd bothered to look at the specs, you'd have seen that they use what appears to be a completely ad-hoc encoding scheme, instead. -Andy === Andrew Warren - .....fastfwdKILLspam@spam@ix.netcom.com === === Fast Forward Engineering - Vista, California === === === === Custodian of the PICLIST Fund -- For more info, see: === === http://www.geocities.com/SiliconValley/2499/fund.html ===

J.P.D. Kooij <PICLISTKILLspamMITVMA.MIT.EDU> wrote: > [a long description of Hamilton cycles on n-dimensional cubes and > their relationship to Gray codes, including the following lines] > > A Hamiltonian cycle would be: > 000 -> 100 -> 110 -> 010 -> 011 -> 111 -> 101 -> 001 (-> 000 ) Joost: While every n-bit Gray code describes a Hamilton cycle over an n-dimensional cube, the converse is nut true; every Hamilton cycle over that cube does NOT describe a Gray code. While many Hamilton cycles over your example 3-dimensional cube are possible, only ONE of them is the 3-bit Gray code (and the example you showed, by the way, isn't it). -Andy === Andrew Warren - .....fastfwdKILLspam.....ix.netcom.com === === Fast Forward Engineering - Vista, California === === === === Custodian of the PICLIST Fund -- For more info, see: === === http://www.geocities.com/SiliconValley/2499/fund.html ===