DPRG
DPRG List  



[DPRG] Why high resolution encoders are desirable

Subject: [DPRG] Why high resolution encoders are desirable
From: David Peterson robologist at yahoo.com
Date: Fri Aug 24 15:21:36 CDT 2001

I may be getting about 517 ticks per inch with the above
encoders, would it be a good idea to try and scale this,
perhaps reduce the precision a bit? That comes out to a 19
bit number for traveling 48 feet or so, might be better to
stick to word sizes, though it may work anyway, have no real
idea at this point.
David P
http://www.geocities.com/robodave2000/
----- Original Message -----
>From: "Clay Timmons" <ctimmons at cadence.com>
To: <dprglist at dprg.org>
Sent: Friday, August 24, 2001 12:57 PM
Subject: Re: [DPRG] Why high resolution encoders are
desirable


>
>
> Dan Creagan wrote:
>
> > While I agree with your assessment, you have used the
extremes to show
> > your point. 1 tic per inch would be 9.4 tics per
revolution on a 3 inch
> > wheel.  Not likely a choice of patterns.
> >
> > With a 240 tics per revolution wheel, the wheel movement
is 4/100s of an
> > inch (less, actually) per tic.  So with a modest
encoder, you can have
> > enough accuracy that you *can* make those turns.  The
encoder tics won't
> > be the problem. Gear lash and wheel slip will now
overcome any accuracy
> > you gained by having an incredibly accurate encoder.
When I first
> > started this, I was using a 90 clicks per turn encoder
and it was
> > definitely getting accumulated error; however, running
on carpet, slopes
> > or running fast was always more detrimental to accuracy.
Much more!  I
> > have movies of the 90 clicks job doing perfect squares
on smooth
> > surface.  The errors tend to cancel one another for
short runs like
> > that.  They do accumulate over time.
> >
>
> My encoders work out to 40 tics per inch while David
Anderson has about 80 tics
> per inch.   4/100s would be 25 tics per inch.     These
are about the
> ballpark of accuracy for hobby robots.
>
> I can make turns to within +/- 0.5 degree.   A good way to
check your
> dead recokoning accuracy is to run the UBMARK test.   I
don't
> have the link handy but look for Johann Borenstien?  at
the University
> of Michigan.
>
> Later,
>
> -Clay Timmons-
>
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at dprg.org
> http://nimon.ncc.com/mailman/listinfo/dprglist


_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com


More information about the DPRG mailing list