It's back end looks similar to the lathe in the video you linked too. Nice thing about these lathes is the tapered gibs and hardened bed for a machine made in China it has a few bells and whistles. I scrapped all the screw cutting stuff because it was a let down , no back gear or tumbler reverse. I'll open it up again and see what sort of reduction drive I can squeeze in.
I just tried it out and successfully produced a hexagon in 5/8" brass bar. I realised that the lathe already has the reduction belt drive and bottom speed is 30rpm with useful torque . I think I might look at a larger reduction by fitting different pulleys to get down to 15rpm. 12 sided polygon yields a following error in X at 30rpm . I'd like 12 tooth cutters.
As it is I think I can produce a 6 tooth cutter with form relief , this is far better than fly cutting. I will look at a pre gashed blank to see how cutting goes but a solid blank may be OK and gash after.
This really simplifies a lot of mechanical complexity.
I think that I will probably try using lincurve and cunning scaling to generate an asymmetrical form, but this is looking like it can work.
I just found a 40:1 worm drive reduction gearbox in the workshop (2 actually and they could be coupled) , I'll next look at a way of coupling it to the lathe headstock spindle. I'll probably drive it with an auxiliary motor.
It's all very promising.
Peter_Cassar wrote: The hobb idea is probably the better way to go for involute gears as only the single point tool is needed ground to the pressure angle. I was thinking about threading with this method to get a lobbed thread, you're thinking ahead. With clock (cycloidal) wheels , they can be hobbed but you need a different hobb for each module and tooth to pinion leaf ratio. Most of the time we are using single wheel cutters per module which is an approximation.
Now you have made your CNC relieving lathe, making a gear hobbing machine is relatively trivial...
Looks awesome Peter. Great to see the external offsets in action.
It would be very cool if you wrote a component that calculated the involute gear form now.
There is some background and a link to some code on git hub here:
Good Luck. I'm sure Andy will help you. Sadly I don't think my maths is strong enough.