John Kasunich wrote: > Jon Elson wrote: > >>Jon Elson wrote: >> >>>>Gentlemen, >>>> >>>>> The screenshots are on imagebin.org under stustev. They are >>>>>representative of the previously presented scenario of homing >>>>>attempts. aligned, .030, and .060. >> >>I do homing with both velocities in the same direction, and I >>might have seen it glitch once, but it might have been a sticky >>indicator. I then changed it so it did the final search for the >>index pulse in the opposite direction, and the home position is >>set to +0.1, just like Stuart's file, and I got very erratic >>home positions. > > > To simplify troubleshooting, make HOME and HOME_OFFSET the same. > That will tell EMC to go exactly to the index pulse position after > homing. Eliminating the post-homing move will make it easier to > see what is going on. > Will do - but I think it may work fine that way. > Then confirm once again that you still have the problem. > > Next, figure out and mark exactly where the index pulses are on your > machine. Turn the screw slowly by hand while observing the index > pulse signal either thru HAL or better yet with a real scope or meter. > When you get an index pulse, mark the table. Keep turning and mark it > again at the next one. I assume on a Bridgeport that the marks will > be 0.200 apart, but that might not be true if you have belts or such > between motor and screw, with something other than a 1:1 ratio. > it is .200" between marks. I am using a dial indicator to detect the exact position. > Once you have marked the table positions for several index pulses > in the vicinity of your home switch, try homing and see where it > stops compared to the marks. > I was seeing it stopping in discrete quanta of .020", ie. 0.000, 0.020, 0.040" etc. > > I would be shocked and astonished if EMC is doing anything besides > homing to the position that it reads on the falling edge of > "index-enable". In your shoes, I would attempt to arrange for a > couple of LEDs, one driven (thru a digital output) from index-enable, > and another driven directly by the index pulse. Hold both LEDs close > to the marks on the table, and home with a very slow latch velocity. > LATCH_VEL = O.01 would mean 20 seconds to travel 0.200", so it would > be easy to see if index-enable is switching off between index marks. > I would think HALscope would be better at this. I did not notice anything anomalous while watching the halscope displays of encoder-position, encoder-index-enable and index. The picture looked about the same whether it homed to 0 or to one of those alternate positions. > > > Jeff has already explained that. Using three arguments for a setp > command has always been wrong, but HAL used to silently ignore the > error (and drop the trailing argument). Now it calls your attention > to the error so you can fix it. Note that HAL error messages are > usually preceded by HAL:<linenumber>, so it's not that hard to trace > them down. The ini file substitution means that mistakes might not > be in the HAL file itself. > Yup, I just never thought about where and how the values came from the ini file. I looked at the ppmc_motion.hal file and didn't see anything wrong THERE! > >>Anyway, I can confirm that at least when homing in the plus >>direction, and then searching for the index pulse in the - >>direction, this error does show up with the late April version >>of the driver. I am hoping that the latest driver will fix this >>problem. > > > WHAT!? Late April? Why the hell are you reporting a "bug" when > you aren't using the latest version of the driver? > Ahh, because the only machine I have here with PPMC hardware is my Bridgeport, and I don't keep it totally up to date due to possible surprises. I TRIED to update it before reporting anything, but ran into the problem above. > I rewrote the indexing code BECAUSE IT WAS BUGGY! and committed > my changes on May 1st. OK, I don't know exactly what rev Stuart Stevenson is at. Those changes were backported to the 2.1 > branch and released as version 2.1.5. You fixed an index polarity > issue on May 7, that was backported and released as 2.1.6. Since > then you made additional changes in June at the CNC workshop. Those > changes were backported, and will be released as part of 2.1.7. > > http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/hal/drivers/hal_ppmc.c?graph=1 > > OF COURSE there are bugs in the April version! > > To be very blunt: reporting a "bug" when you are using an old version > of the driver is going to get you ignored or flamed. Users will get > some slack, but Jon, you should know better, you are the one who made > the June fixes. Test the LATEST driver. If you find problems there, > report them and I'll try to help. I don't want to hear about bugs that > either you or I may have already fixed. Sorry - it wouldn't have happened if I hadn't hit the problem with the setp command. I got on irc and couldn't raise anybody, either, although there were a number of member listed as being on the channel. Only then did I send a message. > > On a more positive note: the upcoming 2.1.7 contains indexing fixes > for the Vital systems Motenc card, the Mesa 5i20, and the PPMC card. > The motenc was tested at the CNC workshop, and I've tested the 5i20 at > home. Let's try to get the PPMC fixed and tested, so that 2.1.7 can be > the final 2.1 release. There's nothing more annoying than doing a > "final" release and then having someone commit a significant bugfix > a day or two later. I'm hoping it will be. I don't believe there are any more gremlins in the PPMC driver. I will try to debug the homing with the latest release now and make sure all the possible combinations of homing sequences work repeatably.
Jon ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
