If I could figure out how to install my surpluse $800 encoder I would but the way the machine is designed it's not easy.
Faving done quadrature encoder work back in the late 80's with a Z80 processor and then later on with separate chips that are quadrature detectors the reality is the edge is the only important part and the two phases are only important for direction detection. As far as speed goes only one edge is needed. In fact because of the non-symmetrical wave form trying to determine speed between both rising and falling A and B edges is an exercise in futility. Now positioning is something else of course. If the distance from edge to edge is critical then non-symmetry is not a good idea. One alternative for mounting in my mill is to shift the custom dual pulley downwards so there is room at the top for a toothed belt to an encoder. But why even bother? If I put toothed pulley in place of the V belt then I can use the AC Servo encoder with the appropriate ratio to determine spindle RPM. For now, I'm working on bringing these pulses into LCNC so I can close the loop since although the AC servo will use step/dir the V belt pulleys are not exactly the same size. The 1055Hz did not match what the Bergerda AC servo readout showed for RPM. I doubt the Tek TDS3032 scope is wrong nor the servo. But the pulleys... John > -----Original Message----- > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > Sent: July-03-21 7:18 PM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Mounting spindle sensors. > > Yes, and an easier way to verify is to think the 60 RPM = 1 Rev. per > second. So you need 60 slots per rev to give 60 Hz at 1 rev per second. > > But why bother making the math easy for a computer? Use however many slots > gives the best data for the PID controller. That said, 60 is enough unless > you are going really slow. > > Thinking about the non-symetric square wave. I doubt it is possible to > make it perfect and I don't think it needs to be. Say the slots > are parallel cuts. Then they are only 50% at one exact diameter but the IR > light beam overs a spot, not a point. So half the beam is being > interrupted with a less then 50% duty cycle and the other have more then > 50%. > > You would need to make wedge shape slots to be perfect 50% over the entire > sensor aperture > > Worse is that the Schmidt trigger is asymmetric. The on-point is 1.5 times > higher than the off-point. That 1.5 factor is relatively large > > But does it matter if the controller only looks at leading edges? OK, you > have some fraction of a degree error with a direction change. It might > matter if the motor was driving a lead screw. > > It makes the wave square I think the solution is to add a radial adjustment > > All that said, I cheated and just bought a $15 600 line optical encoder, an > Omron clone. > > > On Fri, Jul 2, 2021 at 8:56 PM John Dammeyer <jo...@autoartisans.com> wrote: > > > Oh and if my math is correct, the 60 slots/tabs when reported as frequency > > by the scope is the spindle RPM. > > > > 1055 pulses/second divided by 60 pulses/rev = 17.5833333 rev/sec. (pulses > > cancel out). > > > > And 17.5833333 rev/sec multiplied by 60 sec/minute = 1055 revs/minute > > (seconds cancel out) > > > > Here's the test setup with the protoboard that took me all afternoon to > > wire up. When I'm ready I have a TTL to RS422 driver on it in case > > electrical noise is an issue. At the moment it's not populated. > > > > > -----Original Message----- > > > From: John Dammeyer [mailto:jo...@autoartisans.com] > > > Sent: July-02-21 8:30 PM > > > To: 'Enhanced Machine Controller (EMC)' > > > Subject: Re: [Emc-users] Mounting spindle sensors. > > > > > > > I am prepared to state categorically that it must be a coincidence as > > > > "math" was not a word in Douglas Adams's lexicon. As a speaker of > > > > British English he would have always used "maths" as the shorter > > > > version of "mathematics" > > > > > > > Hi Andy, > > > That's the same comments I've heard. However, nothing is to say that > > someone didn't tell him to use that number and may have even > > > explained that to him. The chance has to be very very very small that > > someone would pick a number that by luck that worked out to > > > spell MATH which is required for an understanding of the universe. > > > > > > In either case MATH is also required to understand what the scope is > > showing and why the waveforms aren't symmetrical. > > > > > > A couple of caveats here so don't judge me too harshly. > > > 1. The drawing of the encoder disk was done with the idea of using a 3mm > > tool to cut 3.32mm slots which would result in 3.32 mm > > > tabs at the expected diameter. > > > 2. Using a 4mm tool to cut 3.32 mm slots was a bit of a mistake. Ever > > so slightly. Well. OK. A big mistake. > > > > > > However it was a good exercise, using LinuxCNC and a number of different > > G-Code files for the hub, the perimeter, the mounting holes > > > and the slots. > > > > > > And I'll have to draw up a new larger disk and do this all over again > > because although the sensor bodies do a great job of being clear of > > > the lower pulley the actual sensor mounting tabs stick up too high. > > Can't get the belt on the lower pulley without removing a sensor > > > bracket mounting screw and pivoting it out of the way. > > > > > > Finally, and that would only show up in a video, the ever so slightly > > larger bore of the self-made cast aluminium motor pulley gives > > > that pulley an ever so slight wobble which through the V belt is > > transferred to the spindle pulley (also self-made casting with inner > > > cone to fit on cone spindle drive). > > > > > > It shows up as jitter on the scope. > > > > > > Didn't really care at the time (wasn't going to do it over) since the > > intention is to change to toothed belts anyway. Once the full CNC > > > system is operational with coolant, mister and splash shields, power > > draw bar. And I haven't even started on ball screws yet. > > > > > > I'll post the schematic of the interface board to the PhotoLogic OPB990 > > series slotted optical switches in another posting. > > > > > > John Dammeyer > > > > > > > > > > > > > > > > _______________________________________________ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > > -- > > Chris Albertson > Redondo Beach, California > > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
