Gear Generation on a Shaper - With CNC

[alsinaj]

Patclem's thread once again mentions a method for generating gears on a shaper.

I'm intrigued by the possibility of making a CNC set-up to replace the wrapped-wire mechanism for coordinating table movement with work rotation.
That would simplify indexing, eliminate the need to make discs with the precise pitch diameter (taking into account the diameter of the wire), to adjust the height and tension of the wire, etc.

I have an 8" Lewis shaper disassembled, waiting for me to fix all the mistakes the students who originally built it made. It might be perfect for the job, if I knew how to do it.

I'm no CNC hand, but my simpleminded approach would be to put one stepper motor on the work-holding spindle (like Sir John's CNC indexer), and another stepper on the shaper table leadscrew. A sensor would detect when the cutter was retracted from the work, and rotate both steppers the appropriate amount for the next stroke.

So here are a few questions for you CNC hands out there:
1) Any reason this approach wouldn't work? Any gotchas awaiting the unwary experimenter?
2) Will it be necessary to mechanically lock the spindle during the cutting stroke? If so, how?
3) Does Mach3 have any facilities that could be used to do this job?
4) Could helical gears be produced this way?

Looking forward to your thoughts. JCA

[dp]

One thing to be careful of is the shaper table may not move on every stroke. The pawl mechanisms can have dead spots that just barely miss ticking over. My old Whipp shaper had a pretty wide dead spot, in fact. Sometimes the advance cam would miss every time. So if your CNC doesn't both advance the work and rotate it correctly in relation to the stroke it won't produce satisfactory results.

[alsinaj]

dp,
Good point. However, the pawl mechanism will be gone -- replaced with a stepper motor triggered by a crank position sensor and controlled by a table feed direction switch.

[The Artful Bodger]

1) Any reason this approach wouldn't work? Any gotchas awaiting the unwary experimenter?

I believe it would be perfectly doable but I expect there would be gotchas.

2) Will it be necessary to mechanically lock the spindle during the cutting stroke? If so, how?

Yes, but locking would be possible. For example, put a taper on the end of the spindle to engage with a securely mounted socket. Arrange it so that the ram coming forward at the start of the stroke pushed the taper into the socket, maybe with a strong spring. Arrange some sort of striker to hit the end of the spindle at the end of the cutting stroke to disengage the taper.

3) Does Mach3 have any facilities that could be used to do this job?

Sorry, I dont know anything about CNC software.

4) Could helical gears be produced this way?

I guess nothing is impossible and presumably if the above mentioned socket could be caused to rotate during the cutting stroke that would do the trick.

One of the challenges I have seen when considering how to cut gears on a shaper is that as each tooth is formed there is a point where the direction of backlash in the horizontal feed reverses, the tight wire system of course has no backlash and avoids that issue.

[alsinaj]

Artful,
I think you answered the question about helical gears: the spindle can't be locked and unlocked at the same time.
If it's turning, it has to be unlocked; if it's unlocked, it's going to have backlash. QED. Him no work.
For the locking mechanism, how about a band brake actuated by an air cylinder? The electronics would disengage the brake when the steppers need to turn.

[dp]

I wonder if an electric clutch like used on automotive air conditioner compressors would work as a brake.

[The Artful Bodger]

There is more to this subject than the spindle brake.

If the spindle was turned by a worm there would be no need for a separate brake but the cutting sequence would be a bit more complicated and might involve cutting half of each tooth with the spindle rotating in one direction then reversing to cut the other half of each tooth, this would would ensure the cutting forces were working against the worm and would avoid backlash problems. That would be a bit tedious if the spindle was being rotated manually but fortunately computers are more tolerant in that respect!

[John Stevenson]

No reason why not.
I did a fair bit of background work on this a few years ago, even bought a 14" shaper as a donor machine than realised my brain fart about having a shaper taking up floor space probably never to get used and resold it.
In theory it would be able to do helicals but only low angle ones and how good would remain to be seen.
The work holding head would have to be worm driven to stop any wind back and backlash could be pretty well eliminated by decent assembly.

Mach is able to control this with a simple limit switch that only allows the rotary head to move when the tool is not in the cut.

It will only cut one tooth at a time starting off the blank, working thu and ending off the blank where it then returns to it's start position and indexes round for another tooth. Unless you are talking big gears it would be able to cut at full depth of the tooth i.e. one pass because you would only be taking a shaving cut off one side at at time.

It would be slow but able to work unattended and it's saving grace would be the cutter would be just a Vee shaped cutter with straigh sides equal to the PA of the gear, identical to a screwcutting tool for an Acme thread.
In fact that would work well for 14.5 degree PA gears.

Basically you are building a CNC controlled Sunderland gear shaper that only uses a single tool.

[EVguru]

There doesn't really need to be any CNC involved.

If you fitted a glass or magnetic scale to the traverse, you could then feed the pulses into simple divider box, just like John did for gear hobbing.

[TGTool]

There doesn't really need to be any CNC involved.

If you fitted a glass or magnetic scale to the traverse, you could then feed the pulses into simple divider box, just like John did for gear hobbing.

Which also means he could use only one stepper and also not have to worry about the feed pawl missing a pickup now and then. The rotation of the blank responds exactly to how much feed occurs, including none at all.

[The Artful Bodger]

TGT, the rotation of the blank depends on both the feed and the PCD.

[TGTool]

QUOTE=The Artful Bodger;886744]TGT, the rotation of the blank depends on both the feed and the PCD.[/QUOTE]

Right. But if the feed is left as mechanical with either an encoder or a linear electronic scale, the electronics just reads feed amount (if any) and calculates the appropriate steps to send to the blank holder. Same as John Stevenson's setup where he's reading rotation of the spindle to synchronize the workpiece. Of course the size of the workpiece has to be entered to calculate the relationship between feed and rotation.