Jeff Epler wrote: > John, > > When using stepconf, the BASE_PERIOD is chosen for you. If the required > step rate is low, stepconf chooses a high BASE_PERIOD, capped at > 100000ns. In the case of Alan Condit's machine, the maximum required > step rate is 3500Hz which is one step per 285us or so. > > Do you see any reason why this large BASE_PERIOD would be a problem? >
Yes, jitter. Base period doesn't just determine the maximum step rate. It also determines the granularity of step rates. With a period of 100uS, the available step rates between 1250 and 3500 Hz are: 3,333 (3 periods) 2,500 (4 periods) 2,000 1,667 1,429 (7 periods) 1,250 (8 periods) If the period was 29uS, the available step rates would be much closer together: 3,448 (10 periods) 3,135 (11 periods) 2,873 (12 periods) 2,652 2,463 2,298 2,155 2,028 1,916 1,815 1,724 1,642 1,567 1,499 1,437 1,379 1,326 (26 periods) 1,277 (27 periods) When the machine needs to move at a rate between the ones listed, stepgen alternates between the two nearest ones. The closer together they are, the less radical the jumps, and thus the less jitter. In my opinion if you are running software stepping, you should run your base thread as fast as possible, regardless of your maximum step rate. For drives that require long step pulses, doublestep is NOT a good thing. In that scenario, it wastes an enormous amount of CPU time busywaiting that would be better spent running a faster thread. The real need for doublestep is when the step length is a few microseconds or less, since it is impractical to run a thread fast enough to generate such short pulses. This is a personal opinion only, but IMO standard stepping should be the default. The main advantage of doublestep is that it provides a higher maximum step rate in many (but not all). Alan doesn't need that higher rate. (And even if he did, with his long step pulses, he would actually get a higher rate using regular stepping and a faster base period.) I recommend http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TweakingSoftwareStepGeneration as the reference for calculating base periods and such. I admit it would be helpful if that page and the associated spreadsheet were updated to account for doublestep. Regards, John Kasunich ------------------------------------------------------------------------- SF.Net email is sponsored by: Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://sourceforge.net/services/buy/index.php _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
