On Tue, Nov 12, 2013 at 1:28 PM, sam sokolik <sa...@empirescreen.com> wrote:
> this is the splash screen running.. Z is the only axis that doesn't > obey the 30in/sec^2 limit > http://imagebin.org/276777 This is good news, and similar to what I saw. > The spiral - goes over the limit. > http://imagebin.org/276778 > Here is a halscope of one of the overshoots - it is a ways into the > spiral.. > http://imagebin.org/276780 Cool, this is good data. A few things come to mind based on this: - It looks like the little spikes in acceleration are similar to my results. The pattern (spike up then down) suggests a jump in position, which means a trajectory discontinuity. I'm working on some debug output to inspect this more closely. - There are some overshoots on X acc. that don't seem to be from a position jump alone. I think this may be to improper fall-back to parabolic blends. I'll run this example myself and check the logs. Out of curiosity, which config are you using to run these? Is this your original config with max acceleration of 30 on all axes? Also, what are you using to calculate the "acc-ok" signal? I'd like to add that to my hardware config. > > This is at 1khz.. 10khz - you seem to get about the same overshoot - > about 43 - 44 in/sec^2 on the spiral > > and now - setting the feedrate override to 120%... :) (it had to be done) > http://imagebin.org/276779 > > > sam The feed override is a bit disappointing, but I'm not really surprised. I haven't spent much time yet getting feed overrides > 120% to work. I need to read the max feed override from the user config from within TP, which means a bit of hunting in the code. Thanks again for the testing. A second set of eyes on this really helps catch the little stuff! Rob > On 11/12/2013 7:50 AM, sam sokolik wrote: > > Very cool!! > > > > This motion is much better than the previous release. > > > > I am not seen the odd start motion that I was seeing before - Nice > > work! (starting a ways into the spiral instead of the start) > > > > couple of things. (initial runs) > > > > The Gcode program doesn't finish. > > http://imagebin.org/276743 > > the motion stops - but the vel shows 116ipm and the linuxcnc still > > thinks it is running the program. (the play button is pushed in) > > You can hit the stop button and re-run the program and is seems ok. but > > it still pauses at the end. > > > > Acceleration constraints. This seems better to me also. It gets better > > with a higher servo thread also (but the peak velocity doesn't seem to > > increase - I had though you had said that would happen...) > > 1khz > > http://imagebin.org/276740 - almost 40in/sec^2 (should not exceed 30) > > 5khz > > http://imagebin.org/276741 - 32in/sec^2 > > 10khz > > http://imagebin.org/276742 - didn't actually look - but it looks like > > the spikes are less > > > > Now - I don't know if it gets worse (as you can see I don't have as much > > sample length at the higher base periods. ) I can also check for peak > > constraint violation also.. > > > > I think though that you are headed in the right direction! (that peaks > > at about 330ipm vs around 120ipm with current blending) > > > > Awesome work! > > sam > > > > > > > > > > On 11/12/2013 1:32 AM, Robert Ellenberg wrote: > >> The public folders aren't shared unfortunately, but I can create a > private > >> folder shared between testers, with a public link if need be. > >> > >> The first batch of test logs and screenshots is uploaded as well, which > >> uncovered a few hints to the remaining acceleration issues. The > >> "blend_tests" subfolder of nc_files has the rest of the test code. I > also > >> did a hardware run on my spectralite: > >> > >> https://www.youtube.com/watch?v=PAVuByKH_Fw > >> > >> This was slightly slower than the initial demo at about 40 seconds, > >> probably due to better adherence to the acceleration constraints. > >> > >> -Rob > >> > >> > >> On Mon, Nov 11, 2013 at 4:46 PM, TJoseph Powderly <tjt...@gmail.com> > wrote: > >> > >>> On 11/11/2013 02:41 PM, Robert Ellenberg wrote: > >>>> I've been doing some rework to fix the issues we found. > >>> ...snippety snip snip snip > >>> > I'm going to run all of my test > >>>> code again tonight and save logs / scope screenshots to my public > folder > >>>> here: > >>>> > >>>> https://www.dropbox.com/sh/m801unl2kh7if6x/f59HALU8nh > >>>> > >>> works for me :) > >>> > >>> i couldnt upload a 'works4me.txt' > >>> but no surprises > >>> > >>> thx for all the work! > >>> TomP tjtr33 > >>> > >>> > >>> > ------------------------------------------------------------------------------ > >>> November Webinars for C, C++, Fortran Developers > >>> Accelerate application performance with scalable programming models. > >>> Explore > >>> techniques for threading, error checking, porting, and tuning. Get the > most > >>> from the latest Intel processors and coprocessors. See abstracts and > >>> register > >>> > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk > >>> _______________________________________________ > >>> Emc-developers mailing list > >>> Emc-developers@lists.sourceforge.net > >>> https://lists.sourceforge.net/lists/listinfo/emc-developers > >>> > >> > ------------------------------------------------------------------------------ > >> November Webinars for C, C++, Fortran Developers > >> Accelerate application performance with scalable programming models. > Explore > >> techniques for threading, error checking, porting, and tuning. Get the > most > >> from the latest Intel processors and coprocessors. See abstracts and > register > >> > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk > >> _______________________________________________ > >> Emc-developers mailing list > >> Emc-developers@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/emc-developers > >> > >> > > > > > ------------------------------------------------------------------------------ > > November Webinars for C, C++, Fortran Developers > > Accelerate application performance with scalable programming models. > Explore > > techniques for threading, error checking, porting, and tuning. Get the > most > > from the latest Intel processors and coprocessors. See abstracts and > register > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk > > _______________________________________________ > > Emc-developers mailing list > > Emc-developers@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-developers > > > > > > > > ------------------------------------------------------------------------------ > November Webinars for C, C++, Fortran Developers > Accelerate application performance with scalable programming models. > Explore > techniques for threading, error checking, porting, and tuning. Get the most > from the latest Intel processors and coprocessors. See abstracts and > register > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk > _______________________________________________ > Emc-developers mailing list > Emc-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-developers > ------------------------------------------------------------------------------ DreamFactory - Open Source REST & JSON Services for HTML5 & Native Apps OAuth, Users, Roles, SQL, NoSQL, BLOB Storage and External API Access Free app hosting. Or install the open source package on any LAMP server. Sign up and see examples for AngularJS, jQuery, Sencha Touch and Native! http://pubads.g.doubleclick.net/gampad/clk?id=63469471&iu=/4140/ostg.clktrk _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
