On Saturday 13 October 2018 10:01:53 theman whosoldtheworld wrote: > So you're trying to make a laser reset ... a special type point to > point .... that is a kind of scanner to linearize your axis > particularly worn by age. Are we talking about a lathe right?
Right. > another > thing .... the CDS sensor how is it? .... the glass above it, the > laser glass ... is clean? I think its clear epoxy dipped, and clean, came from a 4 dollar yard stake light I demolished to get, so you have to know its the cheapest crap on the planet. :) > ... your ohms vary according to the external > light, which is consistant a 5000 lumen led thing on an overhead frame above the lathes bed and 10" to the rear. > the cleaning of the surfaces and the projected shadows ... I > fear that the speed of data acquisition and its variability. ... for > such small measures (0.025 mm) is your real enemy .... I usually know > that we try to use triangulation to be less immune to environmental > disturbances. So that in fact the use of a CDS sensor not divided into > pixels it has the usual serpentine pattern top conductor, so its not exactly a bulk device. > and a laser that shoots in the center in a perpendicular way is > probably the least efficient configuration for the lowest amount of > input data and for their low quality. Agreed > If you look at all the commercial zeroers use triangulation or laser > with a very small spot (50um) ... actually now that I update my > knowelge (with googling) for the greater the laser with micro spot ... > this is because there are sensors with very small pixels and with many > few points in the matrix (precisely 8-16-32). I would much prefer to have a device like that, but this is semi-rural WV, at least 100 miles away from the nearest ipv6 address yet. > then there is the problem of time ... if you used the red laser (a > laser line not a dot) Yes, exactly. Its a line of dots actually. I don't have a green laser, are they point sources? > and a very poor webcam (to have less pixels to > filter No, a DVM in ohms mode, which likely has a dual slope A/D in it with conversion time of many milliseconds is the display device. Some of the reading variations I see at low rpms are I believe related to the rotational alignment, or miss-alignment of the lasers line with the serpentine conductor pattern in the CDS cell. Perhaps a diffuser on the CDS cell would reduce that effect by distributing the light more evenly over the face of the cell? A plastic bb sized ball made of white translucent materiel stuck to the face of the CDS cell? That will probably need my finding a a few hundred more millivolts to fire the laser with. > .... the ideal would be 8-16 pixels for calculation time) you > would still have a lot of calculation to do so that a useful answer > would get you with a minimum of 35ms of delay .... these are laws of > physics and the laws of poor quality of computing processors in 2018 > .... Maybe in 10 years it will be better. > > I think for your purpose the better idea was using a very precise > LVDT and put the measure of your axis in a table for make a good > backslash linearization kinematics file .... LVDT? If I have to buy it online, I might as well buy the best device for the job. That table is the x-val-## inputs to a lincurve, 00 to 15. So theres 16 points where I can set the correction spot on, with the correction at other distances being interpolated between those inputs. The y-val-## numbers are the amount of correction needed at that same x-val-## distance. > Il giorno sab 13 ott 2018 alle ore 14:26 Gene Heskett > <[email protected]> > > ha scritto: > > On Saturday 13 October 2018 04:49:40 theman whosoldtheworld wrote: > > > I'm sorry but my English is rubbish ... you're trying to measure > > > and / or chase the ripple of a sheet metal for a laser system > > > (measuring or cutting .... for example, like those used to cut > > > gold ?) > > > > > > bkt > > > > Not quite that. > > > > I have a laser cartridge made to look like a 38 caliber pistol > > cartridge, with its back battery cap removed and about 22" od .375" > > brass tubing threaded at 50 tpi screwed into that, with a teflon > > plug in both ends, and a brass screw in each teflon plug, with a > > wire between the screws. I've made a pair of spring contacts, one to > > ride the OD of the tubing sticking out of the rear of the spindle, > > and one to ride the screw head on the end. + terminal of a 1.7 volt > > wall wart to the tubing, - to the screw. So the laser is running on > > a stable supply but dimmer than if it was running on 3 volts and 15 > > USD worth of batteries. It is mounted in an 3/8" er20 collet in a > > TTS toolholder, which in turn is in the 8"4jaw and the TTS holder > > had been adjusted in the 4 jaw four about .0003" runout. > > > > When the laser is powered up along with the spindle, it eccentricity > > at the tip of a dead center wobbles about 20 thou around the tip of > > a dead center, and the tailstock has been adjusted to center that > > wobble. > > > > With nothing in the beams path, a small CDS cell stuck in a gob of > > modeling clay has been mounted in the tailstock barrel, and the clay > > adjusted so the center of the spinning beam is on the center of the > > CDS cell. With the laser off, the CDS is subject to the room > > lighting and has a R value of about 2.7k. With the laser on, its > > about 250 ohms. > > > > A sheet metal piece to serve as knife edge is attached to the > > toolpost so it can be adjusted with the X jog dial, which I can jog > > the x in .0001" steps to block around half the beam, and the spindle > > is run at 600 revs, so the lag of the CDS cell tends to average the > > resistance reading. > > > > But it may be that the rev rate is too well synchronized to the > > DVM's A/D rate, something I'll check today by running the spindle at > > 625 revs to see if I can see the same reading wobble I can see at > > 500 revs, its quite a bit of reading wobble, about 100 ohms pear to > > peak at 500. I can go faster, but will need to loosen the front > > spindle cap to control the heat as 1/2 hour at 600 gets it up to > > smelling the hot 5w-20 oil. Its a bronze bearing and the spindle is > > slightly tweaked from a fall-over that also destroyed the compound. > > Its been replaced with a carved cast iron block the same height as > > the compound, which of course isn't needed with the computer driving > > the x at any angle to the y. > > > > I've very gently honed the high spots around the bed dings so they > > are a thou or less. This leaves bed wear, 70 years of it, mostly > > near the spindle, therefore I am using a point a few thou from > > touching the tailstock, which is carrying 3 each, pieces of 3/8" SS > > tubing up the right edge of the cross feed to get all the wireing > > from a cable chain at the rear of the carriage, to all the stuff on > > a new front apron replacing the old drive. The Z drive nut, the X > > motor and the jog dials and their mode switchs are on this 1/2" > > thick alu plate, bolted to the front of the carriage face. > > > > The intention is to measure the offset from a touched off to zero X > > at the tailstock, measureing the offset needed to maintain the same > > ohms from the CDS cell as the carriage is driven, carrying the > > "knife edge" from the tailstock back to the side of the carriage > > just ticking the chuck. Find those offsets, list then on paper, > > then edit into the points of a lincurve module, which will then > > autocorrect well enough to turn a cylinder within a thou even if its > > 4" in diameter and 2 feet long as long as its supported on both > > ends. > > > > But everytime I move to the calibrated point at the tailstock, I > > have to re-touchoff by as much a .001" so theres still a need for a > > better means of averaging the CDS cell to get a stable R out of it. > > Thats todays project, finding the lack of repeatability. Because > > there is a bit over .001" backlash, I always get the 400 ohm setting > > by pulling out, and driving it back in, which should reduce any > > error from x drive backlash. > > > > Its possible the laser itself isn't well centered in its holder, and > > further adjustments of the 4 jaw are needed to further reduce its > > eccentricity at the target. With the bent spindle and refaced > > backing plates, I might even have to resort to grinding the backing > > plates, I refaced them only enough to cut 100% of its face, so there > > could be some axial miss-alignment yet. And thats just for > > starters. :-) > > > > And that is what I am trying to do, calibrating a lincurve to make a > > worn bed fairly straight again. While at the same time, I am effectively aligning the bed to the head. Thanks bkt. -- Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene> _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
