On Wednesday 17 July 2019 21:59:53 Gene Heskett wrote: > On Wednesday 17 July 2019 16:51:30 John Dammeyer wrote: > > There are a number of videos out there changing up the G0704 gear > > driven spindle to belt drive. Some use toothed belts, others a > > double set of pulleys. Most of the videos explain that they will > > soon but the slotted RPM sensor disk back on but for most of them > > they've not done that yet. > > > > For LinuxCNC what sort of encoder resolution is required on the > > spindle to be able to do tapping. I know I can do it with one pulse > > per rev with my Electronic Lead Screw to cut nice threads but I also > > don't slow down the spindle at the end of a tap into a blind hole > > while tracking spindle speed. > > > > My new 1.8kW AC Servo motor with a 2500 pulse per rev encoder can > > echo the encoder pulses out of the drive so if I used a 1:1 toothed > > belt I could use that. But if I want stepped pulleys to give me > > extra torque at low speeds then I'm looking at V belts so I'll need > > an encoder on the spindle. > > > > The spindle pulley on my G3616 type mill has a tapered cone. That > > also suggests I want to stay with V belts and just mount a slotted > > disk onto it rather than try to make a similar cone that holds > > toothed GT3 belts. There's room. But how many slots? > > > > Thanks > > John Dammeyer > > I have a G0704 John, and have not managed to damage those plastic > gears yet, but I found a 67 slot wheel replacing the original drilled > tach wheel, with slot interruptors had way too much quantization noise > to be used as spindle speed feedback. An amount to gain in a pid to > hold the average speed close enough for rigid tapping, injected so > much noise into the servo that it sounded like the spindle bearings > had square balls slowly destroying their cages. The noise was > actually the servo amp slamming the tooth clearances back and forth. > > Some background: I took the OEM control box off, and built a power > supply makeing a well filtered 126 volts dc, good for 20 amps surge. > And I took the output of the spindle PID to a pwmgen module in a > 5i25, and that PWM signal to one of Jons (Pico Systems)PWM-Servo amps > which is heavy enough to control a PSU with that sort of muscle. The > motor has a nameplate voltage of 0 at 9.7 amps for its one hp rating. typu correction motor has a nameplate voltage of 90 at 9.7 amps for its one hp rating.
> So I've power enough to reverse the spindle from 2900 revs fwd to 2900 > revs backwards in about 400 miliseconds, and back again. I have Jons > amp set to current limit at about 17 amps and you can hear it chirp as > it does this. > > But that homemade encoder was just too coarse, so without really > figureing it out ahead of time, I bought on fleabay, an Omron encoder > with a 1000 line quadrature A/B/Z output. Once it was in hand. I > drilled and tapped the top of the motor shaft and planted a 5mm > extension about 5/8" long and made a mount to put the encoder on top > of the motor so it spins at motor rpms. > > Somewhere along the line I discovered the encoder was a low voltage > differential output, so the A/B wires were interrupted by a pair of > rs-485 interfaces, about $2 each on fleabay. They turned out to be > bi-dir, so I had to jumper them for one way, and they gave me some > purty 5 volt rail to rail signals to feed a 5i25's encoder inputs. The > Z still comes from the old optical because it needs to mark zero > degrees on the spindle. Then I put a notch in the speed knob that > fitted the rollers of 2 small micro switches, and made a sandwich out > of a pair of half circles cut from .030 alu sheet with the > microswitches between them and arranged to see that notch, sending a > pair of "I'm in gear so and so" back to the 5i25. Those face a mux4 > which is used to both slow the motor to a crawl when neither switch is > closed, and scale the SCALE from the ini file for feedback to the pid > and video tach. I temporarily added some more hal to make me a scale > reference, so the SCALE is scaled to the correct values regardless of > which gear its in, and rigid tapping is correct in either gear. > > That gives me an effective scale of a bit over 7000/rev in high gear, > and something over 14000 in low gear. Quantization noise has vanished! > I won't claim its silent, but I have to either look, or listen real > carefully to make sure the spindle is turning. The 2 unused inputs to > the mux4 are fed about a 25 motor rpm offset voltage by a setp in the > hal file, and the speed response is so instant that I can reach up and > change gears while running wide open since by the time the knob has > been turned 3 degrees, the motor has slowed to a crawl, the flat faced > gears change silently and are completely engaged by the time the other > switch closes. 200 millisecs later the motor is back to top speed and > the spindle is turning at either 1500, or 3000 revs. All with zero > possibility of damaging the gears. And I can run a Pgain of about 20, > even as high as 40, but thats overkill, in the spindle PID. Worst > effect is that until Jon's servo amp regulates, making the motors > magnetics chirp, I haven't a clue how hard the motor is actually > working. The spindle doesn't slow as the load of tapping a hole comes > in. Makes me wish I could find a 15 horse motor as that would remove > the need to peck tap the bigger holes :) > > If interested John, I can pm you a tarball of that whole config dir so > you can borrow what you want. Theres also some alignment stuff that > still needs fine tuning and which is waiting on me finding a camera > that doesn't die about the time I get a good view. Teeny cameras with > long lenses have seemed to disappear from the fleabay listings over > the last 2 years. Darnit. > > Cheers John, Gene Heskett 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) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/gene> _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
