the problem exists if the size of the one step is not evenly divisble into the smallest commanded step size. making smaller steps still cant fill .001 units or .01 units if they are some odd number. you cant make a dime out of 3cent pieces!. Ther's leftover motion that needs to be taken care of when this happens ( when step size is not 'nice' )
and the software has to accumulate the errs till it hits some threshold, then the software has to insert a step to get the dance back on track. Another basic cnc control problem is what happens when you move X&Y at same time? the distance moved is sqrt(2) greater than you may expect, yet the time allotted might be same ( wrong!) unless you take measures to slow down. ditto for 3D moves. I'm sure Julian is thinking about those problems also Julian is asking where these things are handles in the code (I have faith but no proof ). It does exists else we'd jerk a whole bunch with speed changes and accel lurches. TomP TjTr33 On 08/18/2015 09:56 PM, Gene Heskett wrote: > On Tuesday 18 August 2015 21:53:48 Jullian wrote: > >> Hi, >> In the pulse generating mechnism, where's the code processing step >> losing? >> >> >> for example, if 1mm refering one pulse, then 3.3mm will generate 3 or >> 4 pulse? and if 3 pulse, we lose 0.3 mm, if 4 pulse, we do more 0.7mm. >> >> how do we compensate this, where's the code about solving this >> problem? >> >> Best regards! > > We generally don't, Jullian. What we do is use microstepping drivers, > and/or mechanical geardowns so that 1 actual step is usually less than a > micron. When software generating the steps, this is a balancing act > between the maximum step rate the computer can generate, and even on a > good computer this absolute limit is nominally 40 Khz. > > You have the microstepping in the motors driver set for too big a > divisor if you can hear the motor making fairly large speed changes as > the speeds go up, so you have to reduce the microstepping divisor, and > halve that axis's "scale" setting so it doesn't have to issue so many > pulses to move an inch or mm. Its doiing well when as you slide the > feed override up and down, the tone from the motor is like sliding the > steel on a steel guitar. This makes each step a larger move. > > OTOH, more than 5 steps to move .001 would be overkill, and quite likely > beyond the ability of the machine mechanicals to control anyway. > > The use of an auxillary card, such as one of the Mesa 5i25 cards, moves > that speed limit up to the speed of the opto-isolators in the better > drivers, (or in the breakout board if it also has isolators) the one I > measured w/o a breakout board could just barely get past 300Khz before > the motor stalled. That was a hair over 3000 rpms! But the stepper has > little torque at that speed, I could grab the 1/4" shaft and stall it > with my fingers, and that was a 425oz motor running on 48 volts. > >> ---------------------------------------------------------------------- >> -------- _______________________________________________ >> Emc-developers mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/emc-developers > > > Cheers, Gene Heskett > ------------------------------------------------------------------------------ _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
