On Wednesday 12 August 2015 01:31:07 John Dammeyer wrote: > > I am one who is running a small lathe with LinuxCNC, which it does > > far better than I can. > > > > This subject has come up in the past, and I don't recall anyone > > saying "no it can't be done." > > > > Given a big enough motor, I see no huge show stopper in substituting > > a stepgen for the pwmgen module. But the high speed performance as > > a spindle motor might not be universally usable. > > Cheers, Gene Heskett > > Thanks Gene, > In this case I'm considering installing it on a Unimat Lathe. The 650 > oz-in size 34 motor appears to be large enough compared to the small > DC brush motor currently attached. I'd have to step up the RPM in > order to get the turning speed currently available. In either case, > the question is whether or not LinuxCNC can even create stepping > pulses for a spindle or if only PWM is available. > John
Warning, generalized ramblings of an old fart follow. This is a case where a hardware stepgen might be needed as software only has a limit in output step frequency on the x86 driven pc of 40Khz that is largely removed by the use of something like a mesa 5i25 card. Software has, when near its limits, rather large steps in speed cause by the speed limit of the base-thread that drives it on the X86 platform. I did some fooling around with a 425 oz nem 23 a few months back, driving a 2M542 driver with a function generator. The results were basically a speed limit detector of the opto-isolation used in that particular driver, but by playing with the microstep divisor settings I was able to get above 3000 rpms but no usable torque. The speed limit of the opto's seemed to have been something just above 350 kilohertz as long as the duty cycle was near 50%. I was also using a 48 volt power supply, so that was more voltage than is normally used with a 2M542 driver as its rated at 50 volts & 4.2 amp max. Usable torque was pretty much gone by 2000 rpm. Software step generation suffers from latency which causes less than a steady frequency, and this detracts from the usable torque because the motors speed is being asked to vary as much as 20% in a single revolution. On the x86 platform, the next slower software step frequency is nominally 20 Khz, but thats such a huge percentage change that neither is likely to be a usable step frequency for software generation. Because of that, the practical limit is lower, probably under 5Khz for stall free operation. Thats about 600 Hz as you hear it from the motor when using a /8 diviser. Hardware (FPGA) generation raises that "bar" quite a ways. /16 to as much as /64 is usable then. But a /64 explores the speed limits of the opto's in the drivers, limiting the top speeds. Generally speaking about steppers, the top speed of a given motor will go up as a function of the applied voltage as its limited by the inductance of the motors windings. You get less current flow because of the inductance, and the only way to alleviate that is more voltage, or find a motor with lower inductance windings. Since even here, the operative word is TANSTAAFL, that also implies it will need more current to achieve the same torque. I only have one nema 34 motor, on the Z axis of my GO704 mill. Using a 5i25 card, and /16 as the microstep divisor, it has a huge resonance at one relatively low speed, but can happily run at 3x that speed while lifting the head of the machine. How much of that resonance is the rather filligree mounting of this particular conversion kit I haven't determined. I would love to have been able to install some dampers, but the motors supplied are single ended shaft. The damper can be a huge help, absorbing much of the resonance between the mass of the armature and the magnetic springs of its operation, and adding some home-made dampers to my micro-mill rather easily took the rapids moves from 7 or 8 ipm (on the 20 TPI OEM screws) to as high as 34 ipm, however that mills head sled onfit the post is so poor that in everyday usage its rapid limit is more like 14. But that was still a very real improvement. That mill now has ball screws in its XY table, but the Z drive is still the 10 tpi acme screw in front of the post, a modification I made within months of its purchase when I found that no amount of torque applied to the Z handwheel could generate more than 5 lbs of downforce on a drill bit because it was all used up in binding the short wheelbase head on the post. With an extended wheelbase by adding 1/2" thick alu bars carrying some roller skate bearings riding the post, and a 425 oz driving the nut, I can now put just a hair over 150 lbs on the tip of a drill bit. That _will_ drill the hole. And that Z drives accuracy is spotty but can for a jobs duration, be adjusted for less than 0.002" effective backlash. The rotateing acme nut is actually 2 nuts that allow the backlash to be adjusted out. I'd probably put a 16mmx5 screw in there, if I thought the working accuracy might be improved, but that slender post isn't rigid enough to warrant the effort. And I'll never make the mistake of bragging about the tables v-ways. At least the ultra cheap GO704 has easily adjusted tapered gibs. Both machines would benefit from some means of lubing the ways with a 1 shot system. Rambling mode off, time to go make a pot of coffee and get the day started. Thats the nice thing about being long retired, no schedule to meet. The only boss to please is me. :) Here is a thought. The pwmgen in a 5i25 can also do PDM, which might open up the possibility of still using it to drive a stepper. Not something I have tried however. 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-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users