Steve Stallings wrote: > Many systems used by LinuxCNC and Mach3 users push the timing much > harder to try to get higher speeds. Jitter of 25% is not uncommon > in some software only schemes. This can result in a torque loss > of about 87.5% which is enough to produce results like those Jon > cited in his Sherline experiments. > Yes, this is mostly what I was trying to get at, perhaps not so gracefully! Unless you have characterized the jitter VERY carefully, and truly KNOW the bounds, then you can't trust a 5-minute run of the latency test to be representative. And, for software step gen, you really need to be conservative. > The motor in the NIST test case was run at a speed that allowed the > motor to produce 50% of its holding torque. This is a reasonable value > for industrial use. I only wish we could get most users of LinuxCNC > and Mach3 to accept such assumptions when setting up their machines. > And, this is WHY I really think SW step generation is not a great idea. RT jitter + frequency granularity adds up to pretty rough step timing in many cases. There are larger machines with no microstepping where you can get away with it. But, smaller machines with geared-down steppers and microstepping need much higher step rates, and then it becomes a problem. I'd much rather do it "right".
Jon ------------------------------------------------------------------------------ LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial Remotely access PCs and mobile devices and provide instant support Improve your efficiency, and focus on delivering more value-add services Discover what IT Professionals Know. Rescue delivers http://p.sf.net/sfu/logmein_12329d2d _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
