Yet another thought experiment ...

[arvidj]

Lets assume you have a small knee mill that you want to convert to CNC. Lets also assume you have made the decision to CNC the knee rather than the spindle for the Z axis. Based on these assumptions you end up with something like this ...


  diagram1.jpg (Size: 13.17 KB / Downloads: 166)

Note that the ball screw does not run. The servo turns the ball nut on the top of the pedestal causing the knee to move up and down.

A process check at this time suggests this is a reasonable path to go down, but there is a challenge based on the weight of the knee, which would include the table and anything on it. This could easily go over 400 pounds with 600 or more being realistic numbers.

A possible solution is an "assist" being added to offset some of the weight. Though you can not eliminate the mass you can at least offset some of the effects of gravity.

If you chose to use an "air assist" rather than a "counter weight assist" you gain flexibility in the ability to adjust the amount of assist via an air pressure regulator and do not add any mass as a counter weight solution would.


  diagram2.jpg (Size: 16.91 KB / Downloads: 160)

It should be noted that the air cylinder is single ended as there is only a need to push the piston up and never a need to push the piston down.

Another process check at this time suggests you are still on a reasonable path, but now the challenge is to make it all fit under the knee.

Looking at diagram 2 you notice that there are two shafts going to the bottom of the knee, they rod from the cylinder and ball screw. It then occurs to you that (a) neither of the turn, (b) both of them are held firmly at the top of their container ... i.e. the bushing in the cylinder holds the rod firmly in the center of the cylinder and the ball nut holds the ball screw firmly at the top of the pedestal ... and (c ) they both go up and down.

So what would happen if you replaces the pedestal with the cylinder and replace the cylinder rod with the ball screw. You use nor more space that was originally used by the manual screw and pedestal.


  diagram3.jpg (Size: 13.72 KB / Downloads: 159)

There are no issues with sealing the ball screw at the top of the cylinder as it is a single ended cylinder and the rod side is never under pressure. In fact it is the exact opposite as it has to be open to the atmosphere for the cylinder to work in any configuration.

There are not issues with the ball screw and the piston as the ball screw does not rotate, the same as the existing rod does not rotate.

There are no issues with the ball screw moving laterally at the top of the cylinder causing some issue with piston alignment as the ball nut holds it firmly in place laterally.

Which brings you to the end of your thought experiment with a very uneasy feeling that you have missed something really important. The solution in diagram 3 seems to almost be a "free lunch" with the assist and no space penalty. Yes, there are some implementation details but they do not seem to be formidable.

So what was missed? What intuitively obvious gotcha was not thought of?

Arvid

[doubleboost]

I have seen gas struts used to reduce the load on the screw (tail gate struts)
John

[arvidj]

(11-13-2012, 04:35 PM)doubleboost Wrote: I have seen gas struts used to reduce the load on the screw (tail gate struts)
John

John,

Thanks for the response. I've also see gas struts used successfully.

I dismissed them as (a) not being adjustable and (b) the weight involved seemed to be outside of the gas strut ... even multiple gas struts ... envelope.

I have also seen successful air cylinder implementations similar to diagram 2.

I guess my apprehension is centered around using the ball screw as both the "screw" and the "rod".

It seems to me that in diagram 2 the load is shared between the cylinder rod and the ball screw ... i.e. if the total table weight is 600 lbs and the cylinder is able to provide 400 lbs of force then the ball screw and ball nut will only see 200 lbs of downward force from the table to the nut and zero force below the nut.

And again, in my mind, in diagram 3 the ball nut only sees 200 lbs of downward force because the lower half of the screw is being pushed up by 400 pounds from the piston, but the part of the ball screw above the ball nut see the entire 600 lbs of force and part below the ball nut sees 400 pounds from the piston in the "anti-gravity" direction.

But now that I write it down the ball screw is not seeing any more force above the screw than it would have if there was no assist at all so why worry about it.

Yes, my head hurts.
Arvid

[doubleboost]

http://www.sgs-engineering.com/gas-strut.../gsv10-400#

[f350ca]

Neat idea, can't see why it wouldn't work. On my shop press I use air on the bottom side of the hydraulic cylinder to return it. Holds air for weeks at a time.
You may want to look at suspending the cylinder from the top to avoid misalignment problems. As drawn you have three fixed points down the shaft.

[arvidj]

(11-13-2012, 06:42 PM)doubleboost Wrote: http://www.sgs-engineering.com/gas-strut.../gsv10-400#

Thanks for the link. The good news is that they are adjustable to allow gas out. The bad news is that once the nitrogen is out it can not be put back in.

My though with the cylinder is that I have a regulator on it. When the ball screw moves the table up the regulator will provide additional air at the same pressure to fill the volume evacuated as the piston moves up. When the ball screw moves the table down the regulator will bleed the pressure off ... it is a .2% precision regulator ... and allow the piston to move down in the cylinder the appropriate amount.

The additional advantage of the regulator is that I can fine tune the assist based on the real world weights ... like I have no clue as to what the knee and table really weight now, how much they will weight after they have been CNC's with servo's and such, I can add a little pressure when I put the 90 lb rotary table on it or one or two vices or who knows what that I actually might want to machine and lower the pressure when I am milling air.

But again I admit to not knowing exactly what I am doing.

Arvid

[arvidj]

(11-13-2012, 07:13 PM)f350ca Wrote: You may want to look at suspending the cylinder from the top to avoid misalignment problems. As drawn you have three fixed points down the shaft.

Thanks for the feedback and a very good point about the "three fixed points".

The intent was to allow the top of the leadscrew where it meets the underside of the table to float laterally a small amount. But I have not figured out how to make that happen while retaining the 'Z' accuracy.

[4R8]

I believe it's an extremely innovative idea.

I wouldn't worry about three points of fixing, build slight adjust-ability into at least one point. Easiest imo would be the top. Simply lower the knee right down prior to torquing the mount bolts which reside in slightly oversize holes, your could do the same easily at the base.

If you are worried, you could fix the top and bottom. Then build/fix the servo mount onto the end cap of the cylinder thus making it free to move if required while having the cylinder mounted to the more critical points (Knee and base)

Most commercial AND converted cnc mills have three points along the long axis with no ill side effects.

[TomG]

I'm surprised no one has mentioned that a 400 pound knee is a lot of mass to move when doing rapid moves in Z like peck drilling, even if it's perfectly counterbalanced. It's going to take a pretty healthy stepper or servo to get much acceleration. Why do you want to move the knee instead of the quill?

Tom

[arvidj]

(11-14-2012, 02:57 PM)TomG Wrote: I'm surprised no one has mentioned that a 400 pound knee is a lot of mass to move when doing rapid moves in Z like peck drilling, even if it's perfectly counterbalanced. It's going to take a pretty healthy stepper or servo to get much acceleration. Why do you want to move the knee instead of the quill?

Tom

Excellent question and I hope the answers are at least reasonable.

The first is based on how the quill is supported on this mill. It is an import and the quill is not supported along 100% of its length. There is a "collar" at the top of the housing and a "collar" near the bottom of the housing with much of the middle removed to allow the rack and pinion mechanism for the quill to fit. Therefore I think locking the quill down and moving the knee will provide a more accurate system.

This is not my mill head but it is an example of what you usually find when you take one apart ...


  hm52head.jpg (Size: 38.55 KB / Downloads: 74)

The second ... and probably bigger reason ... is that it is a horizontal\vertical mill and CNC'd knee offers some interesting possibilities. If a chuck is installed in the horizontal spindle then CNC'd lathe possibilities exist. Material in the chuck and the tooling attached to the table. An example of this might be ...

















The difference between diagram 2 and diagram 3 would be the difference between his implementation and the implementation that I envision.

Does that seem to make sense or is it still the wrong path to be taking?

Arvid