Here's an interesting experiment. Mount one of these so it tracks the circumference of a rotary table -- say a 12 inch diameter one. Setup a camera or laser or other means to determine when the table has rotated a full revolution.
Then rotate the table a full revolution and see how many counts you get. Do this again. Does the number of counts repeat? How large is the error. If so, we've just made a rotary table with a resolution of a few tenths (1/8200 of an inch) at the circumference. That's .00116 degrees. or 4.2 seconds of arc. Ken On Sat, Aug 8, 2015 at 6:37 AM, Julian WIngert <julian.wing...@web.de> wrote: > Reading the datasheet: > > http://datasheet.octopart.com/ADNS-9800-Avago-datasheet-10666463.pdf > > Page 13: Path deaviation > > It seems that a homogenous surface puts down the error to ~1.25 / 1600 > inch / inch with an optimum @~0.8 /1600 inch / inch with photopaper and > a sensing distance of 2.25mm. > All at a relatively low speed (6ips) > > Combined with a rough syncing (every 2mm) - using a frame capture of the > sensor - you shall be able to get pretty high resolution with a minimum > of effort. > The sensor itself works up to 150ips and 30g, what the resolution and > deviation there may be... I dont know. > > It is - not recommended for new designs, but I am sure there is > equivalent stuff out there. > > best regards > julian > > > On Sat, 08 Aug 2015, Julian WIngert wrote: > >>> Am 08.08.2015 um 04:00 schrieb EBo: > >>>> On Aug 7 2015 5:32 PM, Peter C. Wallace wrote: > >>>>> On Fri, 7 Aug 2015, EBo wrote: > >>>>> > >>>>>> Date: Fri, 07 Aug 2015 17:13:25 -0600 > >>>>>> From: EBo <e...@sandien.com> > >>>>>> Reply-To: EMC developers <emc-developers@lists.sourceforge.net> > >>>>>> To: emc-developers@lists.sourceforge.net > >>>>>> Subject: Re: [Emc-developers] research on optical encoders > >>>>>> > >>>>>> On Aug 7 2015 4:16 PM, andy pugh wrote: > >>>>>>> On 7 August 2015 at 12:23, EBo <e...@sandien.com> wrote: > >>>>>>>> Possibly, but I cannot tell from the information Renishaw > >>>>>>>> published > >>>>>>>> in > >>>>>>>> that brochure. > >>>>>>> I think that the target is a barcode. The head can see enough > >>>>>>> barcode > >>>>>>> to tell exactly where it is on the code sequence to within one > >> bar, > >>>>>>> then looks at the absolute position of the bars in the viewing > >> area > >>>>>>> to work out the rest of the bits of data. > >>>>>> I think it is following on the same idea roughly. Looking at the > >>>>>> renshaw they claim it can give you 1nm (1e-9m) or 3.9e-8 inches > >>>>>> precision. I have no idea how they are pulling that off besides > >>>>>> laser > >>>>>> interferometry and ring counting. Can you suggest another method > >>>>>> that > >>>>>> would work? > >>>>> AFAIK they dont use a laser, just a bright LED thats pulsed to > >> take a > >>>>> snapshot > >>>>> of the barcode, probably with a rather high resolution linear > >> sensor > >>>>> array (or > >>>>> multiple arrays with pixel interleaving) > >>>>> > >>>>> Quite high-sub pixel interpolation should possible with such a > >> setup > >>>>> because > >>>>> of all the duplicated edges > >>>> agreed with the laser/LED. I would have to study sub pixel > >>>> interpolation to see how much additional interpolation you could > >> get. > >>> You can get really fine results on a theoretical perfect black/white > >>> change and the imaging sensor mounted 45deg. of an almost unlimited > >>> degree of subsampling. > >>> Even with a cheap camera and optics there should be no problem to > >>> resolve down to the uM scale. Problem is the speed of such a > >> construct. > >>> Even with high performance camera systems you have a delay that makes > >> it > >>> imho unusable in realtime positioning. > >>> > >>> If you are able to interface the sensor with an FPGA doing the > >> realtime > >>> analysis - well then you have what renishaw probably has build... > >>> > >>> What should "relatively" easy to be doable is to add such a slow > >> scale > >>> to recalibrate the machine position regularly. > >>> > >>> My first idea was to use a laser mouse sensor, which is easily > >>> interfaceable even with MESA cards - there are ones with SPI > >> interface - > >>> but my application is the calibration of my astronomic mount - which > >>> hardly moves more than 1RPD (Rounds per DAY)... > >>> > >> maybe a somewhat naive question - but how would you deal with vibration > >> > >> of the cameras/sensors ? gut feeling - if you try to deduce 10E-9 m > >> then > >> even just environment noise would become a problem or is there some way > >> to eliminate that in practice ? > >> > >> thx! > >> hofrat > >> > >> > ------------------------------------------------------------------------------ > >> _______________________________________________ > >> Emc-developers mailing list > >> Emc-developers@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/emc-developers > > > -- > Mit freundlichen Grüßen aus Pinneberg > Julian Wingert > > Hellenkamp 8 > 25421 Pinneberg > > Phone: 0170/4516094 > Mail: julian.wing...@web.de > > USt-IdNr.: DE272503212 > > > > ------------------------------------------------------------------------------ > _______________________________________________ > Emc-developers mailing list > Emc-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-developers > -- Kenneth Lerman 55 Main Street Newtown, CT 06470 ------------------------------------------------------------------------------ _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
