Power Draw Bar on NM-200 (leading to ATC)

[kvom]

FWIW, 14mm 12-point socket is a good fit on the drawbar.

[webgeek]

Hi all; got a bit of an update with some of those previously promised images. Here are all the shots of the current design as it stands. The motor mount was the only one that caused any problems as it required a ton of finicky measuring. This design uses this impact wrench. I like how this works as it lets you remove all the plastic. It also lets you remove the nylon spacer that holds the bearings, gear race, etc. It really makes everything sturdier by making it all a cohesive aluminum unit. This thing is now bullet proof.

The final image shows how it all comes together. What's next is the changes needed to add in the solenoid mounts, limit switches and electronics. Please let me know if you have any questions, comments, or ideas. Thanks!

-Mike

[mvcalypso]

Hi guys,

A friend asked a question that got me to wondering...

What's needed for a power draw bar is to turn the draw bar relative to the spindle. The typical approach is to lock the spindle mechanically, then to use a mechanism to turn the draw bar.

A manual tool change involves using two wrenches - one on the lower part of the spindle where the flats are (to prevent the spindle from turning) and one on the top of the draw bar.

A problem on the NM-200 (and the similar Smithy 1240) is that there is not an easy way to lock the spindle. Making a mechanism to lock the spindle using the wrench flats via an arm etc puts the mechanism where it is exposed to coolant and could be in the way when not being used.

So this may be a dumb idea, but.... can we reverse the normal approach?

Suppose one used the top of the draw bar to lock the draw bar from rotating (a mechanism just like what is being discussed except that all it has to do is lower a non-rotating socket over the top of the draw bar to lock it). Now we need to turn the spindle relative to the locked draw bar... there is a big motor already sitting there, attached to a VFD that can do that.

My thought is with an index sensor on the spindle (which some may already have for feeding RPm back into mach) we can count slow turns of the spindle. The idea is that X rotations of the spindle is enough to loosen the Tormach tooling collet and the reverse X turns tightens it back up.

OK, I can see some challenges off the top of my head:
1) Might need to have a senor capable of seeing partial turns in case a full spindle turn is too little resolution.
2) A torque wrench applies a fixed torque to the draw bar; It seems that some equivalent can be calculated from the draw bar pitch and # of spindle rotations (maybe - it would be turns after slack was taken out of the draw bar and the collet starts to tighten - that may not be as easy to detect).
3) There would be some software involved at the mach side to control the spindle - but that doesn't seem too hard.

I'm thinking that we have a computer already there, connected to a motor that turns the spindle, can we leverage that to make the mechanical portion simpler?

Has anyone already gone down this path and can tell me why it's doomed to fail? Or could this be a viable approach?

Dave

[JoeBean]

The min spindle speed is around EDIT that should be 600rpm, so assuming around 8 thread engagement you get 8/600 *60 = 0.8 seconds to full disengagement. Might not be a problem, but it's pretty fast.

What I actually wanted to comment on was the mount for the impact wrench. You might want to consider adding linear bearings/bushings to ride on the rod. I recently went through this exact issue with a mini drillpress design and the sled will be subject to binding with any off axis loads, which from what I can see you'd have with the weight of the drill being unevenly distributed. Even pushing it straight from above will, because of the imprecise pressures your hand will exert, cause occasional binds. Otherwise you need the thickness of the "guide" part of the sled to be in the area of 3x the dia of the rod to mitigate binding. There's a standard somewhere in Machinery's Handbook IIRC. But linear bushings would be the best bet, especially considering that any wear on the sled will exaggerate the problem.

[webgeek]

mvcalypso - that's a clever way to look at the problem and if we had a servo-driven spindle it'd be a superb solution. I'd worry, like JoeBean, about being able to control the spindle accurately enough. At low RPM, it's pretty inconsistent.

JoeBean - ha, I'm kicking myself as the reason it's a 3/4" rod was because the holes in the top plate were 3/4" and the holes in the bottom plate were 1/2" originally. Last night when I finished putting the assembly together in CAD I noticed the discrepancy but forgot that it was because I intended to put bushings in the top. Instead I just resized the lower holes to match the bigger ones in my sleepiness. Good catch, I'll get that updated ASAP.

As for it being an unbalanced load, that's not really the case. The motor and nose sits directly above the draw bar and there is no handle or case on it so it's balanced. I did seriously consider using a third riser bar and I think I'm going to consider adding it back in as it's going to be useful for strength/balance in general and it costs very little to implement while still making it far more secure.

New CAD files coming in the next couple of hours. Thanks!

-Mike

[mcphill]

Any interest in sharing your 3D files? Looks like SolidWorks?

[JoeBean]

UHMW machined into a bushing helps binding, but doesn't eliminate it IME. Teflon might be better. But the problem, really, is that once you skew the sled relative to the 2 posts you are, in effect, jamming it between them. So it's not like normal binding or seizure of a single shaft.

Thinking about it I wonder if the actual thickness of the sled needed to avoid binding shouldn't be expressed as a percentage of the ratio of the distance between posts and the post diameter... Hmmm...

[webgeek]

Any interest in sharing your 3D files? Looks like SolidWorks?

Yup, it's SolidWorks. Not sure I'm going to share the raw files themselves as I'd like to sell this complete. I'll have to think about it a bit. I will provide some free schematics for sure though.

Thinking about it I wonder if the actual thickness of the sled needed to avoid binding shouldn't be expressed as a percentage of the ratio of the distance between posts and the post diameter... Hmmm...

I'm going to stick with the KISS principle here and keep it simple. I went ahead and added a third post for stability and made sure the design properly uses nylon bushings. The longest bushings available for 1/2" diameter shafting is only 1" long.

OK, so new pictures galore. The new assembly pictures show the solenoid attached as well as the three shafts vs. the original two. It also shows the bushings, shaft collars (to retain the springs) and the like. It does not show the nose of the impact driver or the motor of the impact driver though it shows the motor mount. The motor mount/gear housing was upside-down in the previous assembly pictures.

The most controversial part of the design is the pulley on the solenoid. Rather than use a series of levers, I'd like to just use some braided wire and the single pulley. This will double the distance the solenoid can pull to 2" which will work great. It will half the force but I think that's ok too as it's strong enough already.

So here is the latest parts breakdown.

3 x shaft collars - McMaster #6435K14 - $1.79
1 x pulley - McMaster #3434T12 - $9.12
3 x nylon bushings - McMaster #6389K449 - $7.52 (for 5)
1 x impact driver - $20
1 x 12v 20amp power supply - $33
1 x solenoid - $25
3 x springs - ~7 for multiple
Assorted electronic components (atmega, mosfets, etc) - <$40
2 x 1/2" x 5" x 6" aluminum = $6
1 x round bar for motor mount = $5
3 x 6" linear shafting = $15

Target price of 200-250 in parts/materials is still looking really doable

As always, I'd love any feedback or suggestions. Thanks!

-Mike

[mcphill]

Looking good! If you need a beta tester when the time comes, I would be interested... FYI, I will be using it on a Mikini 1610L rather than the Novakon, but the design should not be too hard to adapt over...

[webgeek]

Does the Mikini 1610L use a captive draw bar?

[mcphill]

I believe it does.

[webgeek]

If you have the dimensions of the bolt holes used to attach the base plate than I can mock it up in CAD and make sure it all comes together. Thanks!

-Mike