Re: [Emc-users] Hobbing
using the hobbing setup should allow the spindle to drive a linear axis instead of rotary axis then MPG tapping is possible On Sat, Apr 3, 2010 at 12:26 PM, Stephen Wille Padnos wrote: > Stuart Stevenson wrote: > > driving the spindle with an MPG would allow EMC2 to tap a hole in manual > > mode - interesting no? > > > Would be, if EMC2 actually geared to the spindle ;) > > You wouldn't be able to do the "1/2 turn forward/1/4 turn back" method > though. > > (which would have been much much easier doing 1/2" taps into steel, like > I did in the military) > > - Steve > > > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > > -- dos centavos -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
Stuart Stevenson wrote: > driving the spindle with an MPG would allow EMC2 to tap a hole in manual > mode - interesting no? > Would be, if EMC2 actually geared to the spindle ;) You wouldn't be able to do the "1/2 turn forward/1/4 turn back" method though. (which would have been much much easier doing 1/2" taps into steel, like I did in the military) - Steve -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
driving the spindle with an MPG would allow EMC2 to tap a hole in manual mode - interesting no? -- dos centavos -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
Jon Elson wrote: > Frank Tkalcevic wrote: > >> >> >> >> Nope, you are exactly right. On both accounts. Turning the spindle by hand >> will make Z move until it gets to the Z specified in G33/33.1 then stop. >> G33.1 will move back, but only when it is within the G33.1 bounds. >> >> > Are you sure? When the tap reaches the commanded depth, the command is > given to reverse the spindle. > EMC has no way to know how long that takes, so it has to continue moving > Z forward until the spindle can slow to a stop and begin reversing. > If the Z feed actually stopped immediately when the depth was reached, > it would break the tap. At least this applies to G33.1 > The behavior should be different between G33 and G33.1 On the G33, it > is assumed you have made provisions for the proper exit of the tool from > the work at the end of the single-point threading pass. On a G33.1, the > tap is buried in the workpiece, and Z MUST stay in sync absolutely until > the tap is fully backed out of the work, no matter how much the spindle > overshoots when it is commanded to reverse. > Rigid tapping is a little bit of a special case. Motion is synched to the forward-moving spindle until the reversal is detected, at which time motion is synched to the reversed spindle. I don't think that Z will track if you manually wiggle the spindle back and forth, in either direction. - Steve -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
Frank Tkalcevic wrote: > > > > Nope, you are exactly right. On both accounts. Turning the spindle by hand > will make Z move until it gets to the Z specified in G33/33.1 then stop. > G33.1 will move back, but only when it is within the G33.1 bounds. > Are you sure? When the tap reaches the commanded depth, the command is given to reverse the spindle. EMC has no way to know how long that takes, so it has to continue moving Z forward until the spindle can slow to a stop and begin reversing. If the Z feed actually stopped immediately when the depth was reached, it would break the tap. At least this applies to G33.1 The behavior should be different between G33 and G33.1 On the G33, it is assumed you have made provisions for the proper exit of the tool from the work at the end of the single-point threading pass. On a G33.1, the tap is buried in the workpiece, and Z MUST stay in sync absolutely until the tap is fully backed out of the work, no matter how much the spindle overshoots when it is commanded to reverse. Jon -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
> Anyway, on the basis that a picture is worth a thousand words, here > are approximatelt 2000 pictures. > http://www.youtube.com/watch?v=ZhICrb0Tbn4 > > -- > atp > > The video is great, so is the result. But the captions from youtube are hilarious ;) sounds like it has a bit of a problem with your accent. Regards, Alex -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing: really cool
Hi Andy: Just wanted to say thanks for showing ur video of hobbing. It inspires the rest of us to get off our a-- and do something. did u make the indexing fixture your self?? Really cool: Bill On Sat, 3 Apr 2010 01:20:34 +0100 Andy Pugh wrote: > On 3 April 2010 01:07, Frank Tkalcevic wrote: > >> G33 sounds more complicated. Won't G33 have problems requiring the spindle >> turning under power, being "at-speed", and then synching with index pulse? > > The spindle definitely does not need to be under power or at-speed. I > am not sure about the index pulse, but if it does then you only have > to turn it far enough to get an index then align the clock. > > I think G33.1 will wind back out (it is the tapping code) but I am not > sure it will reverse more than once. It's not hard to find out though. > > -- > atp > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
At 04:45 PM 4/2/2010, you wrote: >In the end I went for a fast rotary axis based on a 3/4" >straight-shank ER32 collet chuck from eBay (about £15) with a 6:1 belt >drive to a NEMA 23 stepper. This is held between taper-roller bearings >in a housing that bolts to the table. >The hob is mounted in the milling spindle, and the spindle is set over >at the correct angle to suit the lead angle of the hob. >In the HAL file there is a direct link from the spindle position to >the rotary axis position, with a scaling factor to suit the number of >teeth to be cut. This keeps the two axes in permanent synchronisation. >The number of teeth comes from a spinbox in a PyVCP panel. >(I need to find a way to zero the encoder and steppgen when the number >of teeth is changed, otherwise it tries to re-synch when I change the >tooth count). > >Anyway, on the basis that a picture is worth a thousand words, here >are approximatelt 2000 pictures. >http://www.youtube.com/watch?v=ZhICrb0Tbn4 > >-- >atp Andy, That's pretty slick! It worked out great. Ain't it great when it all comes together? Mark PS: Anybody ever tell ya you have an accent? =8^Þ -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
> -Original Message- > From: Stephen Wille Padnos [mailto:spad...@sover.net] > Sent: Saturday, 3 April 2010 11:54 AM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Hobbing > > Andy Pugh wrote: > > On 3 April 2010 01:07, Frank > Tkalcevic wrote: > >> G33 sounds more complicated. Won't G33 have problems > requiring the > >> spindle turning under power, being "at-speed", and then > synching with index pulse? > >> > > The spindle definitely does not need to be under power or > at-speed. I > > am not sure about the index pulse, but if it does then you > only have > > to turn it far enough to get an index then align the clock. > > > > I think G33.1 will wind back out (it is the tapping code) > but I am not > > sure it will reverse more than once. It's not hard to find > out though. > > > I don't think so. The rigid tapping code expects the spindle > to only go forward while it's tapping, and to only go > backward when it's retracting the tap. The way > spindle-synched motion works at the moment, motion goes > forward as long as the spindle moves forward, motion stops if > the spindle stops or reverses, and forward motion will > continue when the spindle goes past its previous furthest > excursion. So if you stop the spindle and turn it 1/4 turn > backwards, there will be no axis motion until you have moved > it past the point where you stopped it - 1/4 turn forward. > If you wiggle it back and forth, without ever passing that > mark, there will be no axis motion. > > This is as I understand it, but I could be wrong. Nope, you are exactly right. On both accounts. Turning the spindle by hand will make Z move until it gets to the Z specified in G33/33.1 then stop. G33.1 will move back, but only when it is within the G33.1 bounds. I had a quick try at modifying my .hal file, and it worked, but I left a lot of other things in there that made the axis fault if I turned it too far. I should be able to get that working, which will be fine for what I'm trying to do. Thanks, Frank -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
Andy Pugh wrote: > On 3 April 2010 01:07, Frank Tkalcevic wrote: >> G33 sounds more complicated. Won't G33 have problems requiring the spindle >> turning under power, being "at-speed", and then synching with index pulse? >> > The spindle definitely does not need to be under power or at-speed. I > am not sure about the index pulse, but if it does then you only have > to turn it far enough to get an index then align the clock. > > I think G33.1 will wind back out (it is the tapping code) but I am not > sure it will reverse more than once. It's not hard to find out though. > I don't think so. The rigid tapping code expects the spindle to only go forward while it's tapping, and to only go backward when it's retracting the tap. The way spindle-synched motion works at the moment, motion goes forward as long as the spindle moves forward, motion stops if the spindle stops or reverses, and forward motion will continue when the spindle goes past its previous furthest excursion. So if you stop the spindle and turn it 1/4 turn backwards, there will be no axis motion until you have moved it past the point where you stopped it - 1/4 turn forward. If you wiggle it back and forth, without ever passing that mark, there will be no axis motion. This is as I understand it, but I could be wrong. - Steve -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 3 April 2010 01:07, Frank Tkalcevic wrote: > G33 sounds more complicated. Won't G33 have problems requiring the spindle > turning under power, being "at-speed", and then synching with index pulse? The spindle definitely does not need to be under power or at-speed. I am not sure about the index pulse, but if it does then you only have to turn it far enough to get an index then align the clock. I think G33.1 will wind back out (it is the tapping code) but I am not sure it will reverse more than once. It's not hard to find out though. -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
> > > Which pins to you connect together? I want to be able to > do this to > > keep a dial indicator in the groove of a ballscrew when I try to > > center it in the lathe. When I turn the chuck, I want the > feedscrew to move. > > It was (hm2_7i43.0.).encoder.00.position => mult2.in > mult2.out => (hm2_7i43.0.).stepgen.3.position-cmd I think. > > But cradek's way is much better. > > G33 does nothing until the spindle moves, so is probably ideal. > I'll have a play with that. I like the way you turned the spindle by hand and the synchronised one moved too. That's what I want to do to center the ballscrew in the 4-jaw chuck - even if it means having a second hal config just for that. G33 sounds more complicated. Won't G33 have problems requiring the spindle turning under power, being "at-speed", and then synching with index pulse? -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 2 April 2010 23:19, Frank Tkalcevic wrote: > Which pins to you connect together? I want to be able to do this to keep a > dial indicator in the groove of a ballscrew when I try to center it in the > lathe. When I turn the chuck, I want the feedscrew to move. It was (hm2_7i43.0.).encoder.00.position => mult2.in mult2.out => (hm2_7i43.0.).stepgen.3.position-cmd I think. But cradek's way is much better. G33 does nothing until the spindle moves, so is probably ideal. -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On Sat, Apr 03, 2010 at 09:19:35AM +1100, Frank Tkalcevic wrote: > > Which pins to you connect together? I want to be able to do this to keep a > dial indicator in the groove of a ballscrew when I try to center it in the > lathe. When I turn the chuck, I want the feedscrew to move. You could just use G33 or even better G33.1 to do this, with no hal changes needed. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
That is really nicely done. > In the HAL file there is a direct link from the spindle > position to the rotary axis position, with a scaling factor > to suit the number of teeth to be cut. This keeps the two > axes in permanent synchronisation. Which pins to you connect together? I want to be able to do this to keep a dial indicator in the groove of a ballscrew when I try to center it in the lathe. When I turn the chuck, I want the feedscrew to move. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
In the end I went for a fast rotary axis based on a 3/4" straight-shank ER32 collet chuck from eBay (about £15) with a 6:1 belt drive to a NEMA 23 stepper. This is held between taper-roller bearings in a housing that bolts to the table. The hob is mounted in the milling spindle, and the spindle is set over at the correct angle to suit the lead angle of the hob. In the HAL file there is a direct link from the spindle position to the rotary axis position, with a scaling factor to suit the number of teeth to be cut. This keeps the two axes in permanent synchronisation. The number of teeth comes from a spinbox in a PyVCP panel. (I need to find a way to zero the encoder and steppgen when the number of teeth is changed, otherwise it tries to re-synch when I change the tooth count). Anyway, on the basis that a picture is worth a thousand words, here are approximatelt 2000 pictures. http://www.youtube.com/watch?v=ZhICrb0Tbn4 -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 7 March 2010 09:29, Erik Christiansen wrote: > The hob must always cut at the gear helix angle, so that alone > determines the hob helix positioning. The hob axis then ends up where > the hob helix puts it. There does not appear to be any options there. > (According to the grindings of my mental gears, anyway.) Indeed not, I checked this with my dad. The textbook explanation of a helical gear is that it is a stack of conventional gears side by side with a phase difference, to that the tooth meshing and un-meshing is evened out. However if you try to cut it like that then the hob will cut excess material from the adjacent "conceptual gears" and you end up with very thin teeth. For this reason the hob always runs at right angles to the tooth flanks. > They insist on feeding "parallel to the blank's axis of rotation", and > cite a supporting reference. I can't see that working, though, except > for free-wheel hobbing. If you look at that picture again: http://school.mech.uwa.edu.au/~dwright/DANotes/gears/photos/BrownHobbing.jpeg You can see that the direction of travel of the cutter is parallel to the blank axis. (it is the slide covered by the bellows). In a conventional hobbing machine there is a differential gear arrangement such that the gear rotational position is a ratio of the hob rotational position, plus a proportion of the cutter axial position. (This would be very easy to do in EMC, as I think you have already mentioned). I have seen a pretty good diagram of the gear train that does this, but I can't find it now. Anyway, I have concluded that I don't have the axis separation I need to make the gears I want to make using the lathe spindle as the cutter axis, and I could not see an easy, rigid, way to adjust the blank-to-hob distance, so I am now back to Plan A, which involves making my milling head tilt-able (which will also mean that I need to add an encoder on the milling head, which was always the long-term plan anyway) -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 7 March 2010 11:41, Dave Caroline wrote: > > see diagram at top of page http://www.collection.archivist.info/hobbing.html Ah, yes that is where I saw that diagram. -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
For a helical gear the blank rotates and its no longer a simple 40:1 for a 40 tooth helical gear, a mechanical hobbing machine has a differencial gear to add or subtract the rotation with cut. see diagram at top of page http://www.collection.archivist.info/hobbing.html I have done the maths in gcode to make a helical, it was milled but the principle is the same http://www.youtube.com/user/davethearchivist?feature=mhw4#p/a/u/1/HAtziCsUj5Q I tilt the B axis(tooth helix angle) and rotate the A axis(pitch of helix) Dave Caroline -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On Sat, Mar 06, 2010 at 12:31:04PM +, Andy Pugh wrote: > On 6 March 2010 07:58, Erik Christiansen wrote: > > From the photographs, that makes > > the feed perpendicular to the hob axis, which seems to me to ignore the > > helix angle. How that creates a spur gear with proper gaps, is not clear > > to me. > > I don't think it does. I was intending to set the vertical slide at an > angle and use coordinated motion in X and Z so that the blank moves > along it's true axis. Ah, yes, of course. (Maybe if I have "coordinated motion" tattooed inside my eyelids, I won't keep forgetting that option.) Yes, if the gear helix doesn't match the hob helix angle, then the vertical slide needs to be rotated. > > On the other hand, preferring the quiet running of helical gears to the > > whine of spur gears, I have run the above setup in my mind, with (a > > virtual) EMC advancing the phase of the gear blank as it is fed across > > the rotating hob. If the rate of phase advance matches the helix angle, > > then the blank should come out the other side as a helical gear, I > > believe. (And both blank and feed are perpendicular to the hob axis. > > What could be simpler?) > > I am not sure. It depends in if a helical gear is conceptually a gear > with the teeth rotated on the surface, or a stack of infinitely thin > gears with a pitch difference between each. > In this picture: > http://school.mech.uwa.edu.au/~dwright/DANotes/gears/photos/BrownHobbing.jpeg > The hob axis seems to be tilted to match the gear helix angle, rather > than the hob helix angle. The hob must always cut at the gear helix angle, so that alone determines the hob helix positioning. The hob axis then ends up where the hob helix puts it. There does not appear to be any options there. (According to the grindings of my mental gears, anyway.) Looking for some support on this, I found a simpler way to hob helical gears: Just crank the hob over some more: http://en.wikipedia.org/wiki/Hobbing They insist on feeding "parallel to the blank's axis of rotation", and cite a supporting reference. I can't see that working, though, except for free-wheel hobbing. > I think that in either case you ideally want to match the hob helix > angle to cut a true gear form. For the feed angle, yes. But if you use coordinate motion, you've solved that without effort. If my changing the gearblank's phase, while feeding it across the hob, is equivalent to your coordinate motion to simulate skewed feed, then it should work as well. Since a helical gear meshed with a worm must either move with your coordinate motion, or rotate, to pass its teeth across the worm, I think they're equivalent. (OK, your method is probably easier in gcode, so I'd go with it too.) > I did discuss this with my dad (50 years a gearbox machinist then > designer then service manager) but he seemed unable to grasp that the > feeds and drives are trivial with CNC, but the axis geometry is less > so) What he knows is easy for him. What we haven't done yet is hard for us. Erik -- If you understand what you're doing, you're not learning anything. - A. L. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
Hi Dave, I haven't checked the wheel for accuracy but I would think that this method of generation actually is less prone to errors than methods which cut each tooth separately. Once the teeth are established as to number and you infeed the wheel onto the tap, there are always at least 3 cutting teeth (threads) in contact with the wheel at any one time and this should even out the effects of any minor errors in the tap pitch. I also usually feed the worm along the tap a little each way as I cut which will again reduce errors by using different parts of the tap. I don't know whether there would be any benefit in reducing cyclic errors by producing only wheels with an odd number of teeth like 91 instead of 90 - I did work out what the error would be if I could only make a 91 tooth one and set EMC2 up to suit it and, from what I remember, the error was infinitessimal. The great beauty of this method is that you can make a wheel from scratch in just a few minutes. I simply set up a post with the same diameter as the bore of the blank on my miller's bed ( actually made as a tube sleeve over a bolt ), slipped a spacer onto it ( a second wheel blank) and then the wheel blank I wanted to cut and capped it off with a nut and washer to stop the blank lifting during cutting. The cutter was a normal 10mm x 1mm tap and, with the blank initially set to centre height of the cutter, I wound the blank into the cutter so as to take a fairly hefty cut as I found that this was the easiest way to establish a set of even teeth ( i.e. not 90 1/2!). Then I just left the thing running merrily along as I slowly wound the blank further in until I thought that the teeth in the wheel were deep enough - i.e. not quite full depth on the tap so that I didn't get 'bottoming' when using the finished worm. Once this stage was reached I decided that I wanted the teeth to be 'flat' across the wheel so that I could remove it from the rotary without having to disturb the worm ( don't know why but it just seemed like a good idea at the time...) and so I then wound the knee of the machine up and down slowly as the work was still spinning away. On other wheels I have made (which I made on the lathe) I didn't do this and so the worm and wheel make contact for maybe 1/3 of the worm's diameter - this may be better - I don't know. Anyway, the whole cutting job and subsequent cleaning up took only about 5 minutes per wheel. The other thing I did was to mount the wheel on an eccentric in the rotary table so that I can easily adjust backlash if necessary to compensate for wear - that hasn't been necessary yet.. Why not make one and try out this way - it costs nothing except a bit of scrap material and maybe half an hour Best wishes, Ian __ Ian W. Wright Sheffield UK Dave Caroline wrote: > Looks nice Ian but I would love to measure the wheel for errors in > tooth to tooth distance. We had trouble directly attributable to the > worm and wheel in a bought dividing head, we made some 144 tooth > wheels for a project at the BHI and they were rightly rejected, for > normal clock work the error did not show or was withing tolerance but > for a high count the error becomes pronounced. > > The easy check I use to test dividing now is get your digital > calipers, measure over n teeth, zero caliper , rinse repeat but only > zero if less than a previous measurement. you then see a large > percentage change in places around the wheel if you have a dividing > error its due to worm/wheel form error giving a sawtooth error to the > dividing. > > Dave Caroline > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > No virus found in this incoming message. > Checked by AVG - www.avg.com > Version: 9.0.733 / Virus Database: 271.1.1/2725 - Release Date: 03/05/10 > 19:34:00 > > -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 6 March 2010 07:58, Erik Christiansen wrote: > In MEW 75, another author used a couple of CMOS chips for the > programmable divider between the spindle encoder and stepper driver, to > select the number of teeth. This is very easy to set up in HAL (and quite amusing too, watching the rotary axis turn as you turn the chuck to tighten it, for example) > That setup has the hob running on a mandrel between lathe headstock and > tailstock. The stepper-driven spindle holding the gear blank is mounted > on a vertical slide on the cross-slide. That is probably the setup I will try first, mainly as it needs the least extra hardware. > From the photographs, that makes > the feed perpendicular to the hob axis, which seems to me to ignore the > helix angle. How that creates a spur gear with proper gaps, is not clear > to me. I don't think it does. I was intending to set the vertical slide at an angle and use coordinated motion in X and Z so that the blank moves along it's true axis. > On the other hand, preferring the quiet running of helical gears to the > whine of spur gears, I have run the above setup in my mind, with (a > virtual) EMC advancing the phase of the gear blank as it is fed across > the rotating hob. If the rate of phase advance matches the helix angle, > then the blank should come out the other side as a helical gear, I > believe. (And both blank and feed are perpendicular to the hob axis. > What could be simpler?) I am not sure. It depends in if a helical gear is conceptually a gear with the teeth rotated on the surface, or a stack of infinitely thin gears with a pitch difference between each. In this picture: http://school.mech.uwa.edu.au/~dwright/DANotes/gears/photos/BrownHobbing.jpeg The hob axis seems to be tilted to match the gear helix angle, rather than the hob helix angle. I think that in either case you ideally want to match the hob helix angle to cut a true gear form. I did discuss this with my dad (50 years a gearbox machinist then designer then service manager) but he seemed unable to grasp that the feeds and drives are trivial with CNC, but the axis geometry is less so) -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
Looks nice Ian but I would love to measure the wheel for errors in tooth to tooth distance. We had trouble directly attributable to the worm and wheel in a bought dividing head, we made some 144 tooth wheels for a project at the BHI and they were rightly rejected, for normal clock work the error did not show or was withing tolerance but for a high count the error becomes pronounced. The easy check I use to test dividing now is get your digital calipers, measure over n teeth, zero caliper , rinse repeat but only zero if less than a previous measurement. you then see a large percentage change in places around the wheel if you have a dividing error its due to worm/wheel form error giving a sawtooth error to the dividing. Dave Caroline -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
>>I also have a photograph around here somewhere, of a wormwheel being free-wheel hobbed with a tap held in the lathe chuck.>>> Been there - done that very successfully several times.. The biggest problem is in getting the wheel blank to be the right size before cutting and in taking an appropriate sized bite out of it on the first run round with the tap to establish the right number of teeth - its a bit hit-and-miss. Having said that, I have made a number of wormwheels which have worked very well including the one on my current 4th axis. In this case, I wanted one of 90T and the first attempt produced one with 92T - I could have used that one with no problems but I decided to have another go and the second one gave me 90T with no problems. Its made of brass and, since I wanted to be able to dismantle the rotary without too much effort, I decided I needed straight teeth, not concave as are made by a single pass of the tap, and so I cut the wheel on my little horizontal miller and, once the teeth were formed and to depth, gently tracked the table up and down to cut the teeth to the edge of the brass blank. The worm is just a length of screwed rod the same size as the tap - lathe cut on decent quality and finish stainless steel You can see the wormwheel and worm after a couple of years regular use at http://www.watchman.talktalk.net/temp/Rotary%20table/rot1.jpg Ian W. Wright Sheffield UK -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
Some do not realize a hob is by definition a helical cutter for generating a form. A single or multiple slot cutter is a milling cutter. Where the user needs some quality and accuracy then the ME method is a bit of a bodge but can give reasonable results relating to the quality of the cutter. One can with rack form generate involute form by running a cut shifting the cutter and rotating the blank a small amount and recutting this is how a gear shaper/shaver works, they also use a gear form cutter to shape internal teeth this way. Dave Caroline -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On Fri, Mar 05, 2010 at 01:23:46PM +, Andy Pugh wrote: > I am thinking of making a faster rotary axis using an ER32 collet > holder I have on a 3/4" ground shaft and some taper roller bearings. I > would drive that with a spare stepper I have, at about 10:1 ratio. (or > one of the little servos) > I can't believe that there are very large rotating forces on a gear > during hobbing, I think it is probably largely balanced. The proof of that seems to be in the success of free-wheel hobbing. In MEW 78 [1], a gashed blank is fed into a hob, without any form of synchronisation. i.e. The forces must be "restorative", not disruptive, in order to spontaneously push a gashed blank into synchronisation. The photograph of the finished gear is not large enough to show if it is more than "useable", which is all that he claims. I also have a photograph around here somewhere, of a wormwheel being free-wheel hobbed with a tap held in the lathe chuck. In MEW 75, another author used a couple of CMOS chips for the programmable divider between the spindle encoder and stepper driver, to select the number of teeth. He describes no problems with cutting forces, other than cutting to the tooth depth in three passes in harder steel. Otherwise he just sets the depth, and cuts the teeth in one pass, in an ungashed blank. That setup has the hob running on a mandrel between lathe headstock and tailstock. The stepper-driven spindle holding the gear blank is mounted on a vertical slide on the cross-slide. From the photographs, that makes the feed perpendicular to the hob axis, which seems to me to ignore the helix angle. How that creates a spur gear with proper gaps, is not clear to me. On the other hand, preferring the quiet running of helical gears to the whine of spur gears, I have run the above setup in my mind, with (a virtual) EMC advancing the phase of the gear blank as it is fed across the rotating hob. If the rate of phase advance matches the helix angle, then the blank should come out the other side as a helical gear, I believe. (And both blank and feed are perpendicular to the hob axis. What could be simpler?) Checking that with another thought experiment, we run the helical gear on a rotating matching worm. As the gear is slid back and forth on its axis, its rotation advances and retards in accord with the helix angle. Hmmm ... where can I dig up physical examples quickest, to try it out? Incidentally, if choosing to generate the tooth profile, using a straight (no helix) hob, then that slow process can be accelerated by making a three or four tooth hob. It cuts the prior iteration on the previous tooth and the next iteration on the next tooth, speeding up the process. (Or providing a cleaner tooth form for a given number of iterations.) There was an article on that in MEW 107. Hopefully some of that is useful, Andy. Erik [1] www.model-engineer.co.uk says it is putting 130 back issues on line, but will be charging a subscription. -- Wisdom is one of the few things that looks bigger the further away it is. - Terry Pratchett, _Witches Abroad_ -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 05/03/2010 18:10, Kirk Wallace wrote: > > The problems with trying to replicate hobbing with CNC rotary axes are, > the rotary encoders would need to be very high resolution, feedback and > command control very tight, and data throughput high enough to keep up. > I think it has been stated before that current systems aren't up to the > task. I think it would be great to have someone try, but I think the > odds are against any success. > iv cut gears with just a standard involute gear cuter it works grate, but hobbing it would or should make it alot quicker process. i guess also comes down to how acerate you wish to make these gears etc also. i think if you can overcome your speed problem on rotating the blank like you have said you should be on the way to getting some where close. let the spindle do as it likes (just like in rigid tapping) make the blank follow the spindle nice and close in motion. (we do this with the Z axes in rigid tapping no?) i know some servo drives have the ability to take an exsternal encoder or other source and follow it with out the need of anything else so maybe something to look at too? if you can keep them locked in sync at all times every pass should line up so doing a rerun over the gear should work fine. will be grate to see how you get on if you take this further. good luck, rob -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 5 March 2010 18:10, Kirk Wallace wrote: > The problems with trying to replicate hobbing with CNC rotary axes are, > the rotary encoders would need to be very high resolution, feedback and > command control very tight, and data throughput high enough to keep up. > I think it has been stated before that current systems aren't up to the > task. This message was cast into the void some time ago. I reposted it today. I am now well on the way to finishing my fast rotary axis. I guess I will soon know if you are right... -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hobbing
On 03/05/2010 07:23 AM, Andy Pugh wrote: > (This is one of potentially several reposts of questions that never > made it to the list due to operator error) > > I am unsatisfied with the results of my attempts at gear-milling. I > think this stems partly from me not knowing which of the three cutters > I have is for what tooth count, or what the addendum and dedendum is > meant to be for each cutter. > I have decided that as a set of milling cutters comprises about 6 > cutters at £10-£20 each, a hobbing cutter which will make any size of > gear seems like a good plan. > > Hobbing, as you almost certainly know, involves rotating a hob and the > work on not-quite-right-angles axes in a fixed ratio. Some parts of > that are very easy with an EMC-controlled CNC machine (mine is one of > these one of these, convertulated) > http://www.amadeal.co.uk/acatalog/CX23A-750_Multi-Purpose_Lathe_Milling_Machine.html > I have an encoder on the lathe spindle and it is a simple > matter of connecting that to the rotary table in HAL to keep them > geared together permanently at any > arbitrary ratio. (there is even an encoder_ratio module for this sort > of thing, but that isn't exactly right for what I want) > > However, the stepper-driven rotary table tops out at 5rpm. I can swap > the motor for a servo, and that gets me 17rpm. The lathe and milling > spindles don't really like doing less than 200rpm. I want to cut > 12-tooth pinions, and that really isn't enough overhead for the rotary > axis to catch up and synch. > It is also a major change to the control box wiring and the software > setup to swap to servo motor. However here are advantages, and I have all the > parts. > Holding the rotary axis at the correct angle to a hob held in the > milling head is difficult. (though obviously doable) > clamping it to the table at an angle to the lathe spindle is easy, but > then I can only make gears of one diameter. > It would be nice to be able to rotate the milling head. There is > presently no facilty to do that, but there is a joint between two > castings where the facility could be added. > > I am thinking of making a faster rotary axis using an ER32 collet > holder I have on a 3/4" ground shaft and some taper roller bearings. I > would drive that with a spare stepper I have, at about 10:1 ratio. (or > one of the little servos) > I can't believe that there are very large rotating forces on a gear > during hobbing, I think it is probably largely balanced. > > If I do make this rotary axis, then I can either mount it at an angle > to the milling spindle, or at an angle to the lathe spindle. In the > latter case I would need a vertical slide, but I do have a spare > compound slide that the CNC machine no longer uses. In either case I > would need a compound feed on the two translational axes, or the > offset angle will mean that the gear moves down the axis of the hob, > giving a very slight second-order helix angle (if I am visualising it > right) > Modifying the milling head to tilt avoids this problem. > > The lathe spindle already has an encoder, the milling spindle is still > waiting. > > Ideally I would make a hobbing head to clamp to the lathe saddle to > cut gears held in the lathe spindle, with a nice powerful servo motor > and encoder just like a real gear hobbing machine. But if I had a > suitable servo motor like that it would already be my lathe headstock > motor. > > So, rambling over, does anyone have any ideas for ways of arranging > the shafts and slides that I have not thought of yet? (one idea has > just occurred to me, I could mount a flexible coupling in the milling > or lathes spindle. mount a secondary pair of bearings elsewhere, and > put the hob on that to get the required angle). > I think my favourite so far is a combination of new, faster rotary > spindle holding the gear mounted on a vertical slide on the lathe > saddle, with the hob in the lathe spindle. > > -- > atp > > Here is a stepper based 4th axis that may work the way you're describing: http://www.cnczone.com/forums/showthread.php?t=100160 It is 4 to 1 ratio, and set up for microstepping. It should be able to keep up with your spindle. I'm assuming that yow want to use a spiral hob?. If so, put some shims under the front of the axis, to match the helix angle of the hob. The better way, would be to create a straight hob, and just index the axis for each tooth. The idea is the same, it's just slower. -- - Ne M'oubliez ---Family Motto Hope for the best, plan for the worst ---Personal Motto (\__/) (='.'=) This is Bunny. Copy and paste bunny into your (")_(") signature to help him gain world domination. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See
Re: [Emc-users] Hobbing
On Fri, 2010-03-05 at 13:23 +, Andy Pugh wrote: ... snip > I am unsatisfied with the results of my attempts at gear-milling. I > think this stems partly from me not knowing which of the three cutters > I have is for what tooth count, or what the addendum and dedendum is > meant to be for each cutter. > I have decided that as a set of milling cutters comprises about 6 > cutters at £10-£20 each, a hobbing cutter which will make any size of > gear seems like a good plan. I haven't cut any gears yet, but my plan is to use the rack form gear cutter to cut all gear sizes for a particular pitch. https://www.travers.com/Default.asp (Click on Keyword Search box, click Guest, search for "involute", their search and descriptions are pretty weak) or http://www1.mscdirect.com/cgi/nnsrhm (search "involute", or "invoute 14-1/2" for 14 1/2 degree pressure angle) (Other links welcomed, Number 1 cutters seem to be the rack form) The hobbing motion can be replicated, except you only cut one section of tooth width, one tooth at a time (actually the slot between two teeth). The gear blank is mounted to a rotary axis. The gear cutter is mounted in the spindle. The coordinated motion of the rotary and cross feed will mimic the motion of a rack tooth engaging the gear as it rotates for the entry an exit of one tooth slot. The process would need to be repeated to cut the width of the gear, then repeated for each slot. The problems with trying to replicate hobbing with CNC rotary axes are, the rotary encoders would need to be very high resolution, feedback and command control very tight, and data throughput high enough to keep up. I think it has been stated before that current systems aren't up to the task. I think it would be great to have someone try, but I think the odds are against any success. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Hobbing
(This is one of potentially several reposts of questions that never made it to the list due to operator error) I am unsatisfied with the results of my attempts at gear-milling. I think this stems partly from me not knowing which of the three cutters I have is for what tooth count, or what the addendum and dedendum is meant to be for each cutter. I have decided that as a set of milling cutters comprises about 6 cutters at £10-£20 each, a hobbing cutter which will make any size of gear seems like a good plan. Hobbing, as you almost certainly know, involves rotating a hob and the work on not-quite-right-angles axes in a fixed ratio. Some parts of that are very easy with an EMC-controlled CNC machine (mine is one of these one of these, convertulated) http://www.amadeal.co.uk/acatalog/CX23A-750_Multi-Purpose_Lathe_Milling_Machine.html I have an encoder on the lathe spindle and it is a simple matter of connecting that to the rotary table in HAL to keep them geared together permanently at any arbitrary ratio. (there is even an encoder_ratio module for this sort of thing, but that isn't exactly right for what I want) However, the stepper-driven rotary table tops out at 5rpm. I can swap the motor for a servo, and that gets me 17rpm. The lathe and milling spindles don't really like doing less than 200rpm. I want to cut 12-tooth pinions, and that really isn't enough overhead for the rotary axis to catch up and synch. It is also a major change to the control box wiring and the software setup to swap to servo motor. However here are advantages, and I have all the parts. Holding the rotary axis at the correct angle to a hob held in the milling head is difficult. (though obviously doable) clamping it to the table at an angle to the lathe spindle is easy, but then I can only make gears of one diameter. It would be nice to be able to rotate the milling head. There is presently no facilty to do that, but there is a joint between two castings where the facility could be added. I am thinking of making a faster rotary axis using an ER32 collet holder I have on a 3/4" ground shaft and some taper roller bearings. I would drive that with a spare stepper I have, at about 10:1 ratio. (or one of the little servos) I can't believe that there are very large rotating forces on a gear during hobbing, I think it is probably largely balanced. If I do make this rotary axis, then I can either mount it at an angle to the milling spindle, or at an angle to the lathe spindle. In the latter case I would need a vertical slide, but I do have a spare compound slide that the CNC machine no longer uses. In either case I would need a compound feed on the two translational axes, or the offset angle will mean that the gear moves down the axis of the hob, giving a very slight second-order helix angle (if I am visualising it right) Modifying the milling head to tilt avoids this problem. The lathe spindle already has an encoder, the milling spindle is still waiting. Ideally I would make a hobbing head to clamp to the lathe saddle to cut gears held in the lathe spindle, with a nice powerful servo motor and encoder just like a real gear hobbing machine. But if I had a suitable servo motor like that it would already be my lathe headstock motor. So, rambling over, does anyone have any ideas for ways of arranging the shafts and slides that I have not thought of yet? (one idea has just occurred to me, I could mount a flexible coupling in the milling or lathes spindle. mount a secondary pair of bearings elsewhere, and put the hob on that to get the required angle). I think my favourite so far is a combination of new, faster rotary spindle holding the gear mounted on a vertical slide on the lathe saddle, with the hob in the lathe spindle. -- atp -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users