Thanks to you both! I just pushed the branch of the same name to the official repo. Unfortunately, it looks like the buildbot queue has stalled on this build:
http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4528 Once it gets through the queue, though, my branch should have a build available. On Sun, Sep 25, 2016 at 11:41 AM Gene Heskett <[email protected]> wrote: > On Sunday 25 September 2016 08:47:13 Robert Ellenberg wrote: > > > Hi All, > > > > Recently, I found a way to reduce the acceleration / velocity jitter > > during threading and tapping motions (see this issue on github > > <https://github.com/LinuxCNC/linuxcnc/issues/164> and the > > feature/spindle-sync-jiiter-fix branch > > <https://github.com/robEllenberg/linuxcnc-mirror/tree/feature/spindle- > >sync-jitter-fix>). The effect looks significant on the HALscope, but > > I'm not sure how much of a difference it would make on a real machine. > > Is anyone interested in doing a before and after test on real > > hardware? > > > I am in the middle of converting a Sheldon 11x36, and finding I need to > make everything but most of the bolts. Otherwise I'd be up for it in a > heartbeat. > > > Thanks! > > Rob > > I hadn't seen the github posts, but will look. And the 1st link is > exactly what I've been fighting with. Enough Pgain to be usefull and its > hammering the headstock gears. Fiddling with the opto spacings helps, as > does a brghtness pot to each opto's led. But there is no true null > point, ever. > > What I've now done is a bit of filtering on the two, soon to be 3 > machines that have a spindle encoder, but it does so at the cost of 2 to > 4 edges worth of delay of the encoder output fed back to the PID. I am > running the position data thru a 4 stage fifo I've constructed out of > hal modules, with the feedback to the PID being the unity gain sum of > those 4 fifo buckets. Timing and sequencing being controlled by the > addf order such that its output is available without any additional > delays caused by miss-ordering in the servo-thread period. At low > spindle speeds, the update of the fifo is done only if an edge has been > detected, fifo update triggered by an unequal from a comp module, > looking at the last data and the current data. It works well, reducing > that noise by /4, allowing more Pgain, and hasn't seemed to have > affected rigid tapping operations unless thats why a peck cycle cuts a > looser tap. I have noted that using cheap taps that I have to back out, > clean and anoint about every turn. But I have blamed the majority of > those problems on the fact that cheap taps aren't straight. Cheap, > straight and sharp, pick any 2. :( > > Thank you very much for taking a look at this Robert. > > Are you running a buildbot, the output of which is available to apt-get > etc? That would make it much easier tested here. > > 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-developers mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-developers > ------------------------------------------------------------------------------ _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
