I made some useful progress on this. Although Fusion appears to allow different values to be specified, the post simply generates a G84 tapping macro command, which doesn't pass the retract rate. The alternative, closely related tapping / boring canned cycles all use either the rapid retract or the same as the infeed.

The tapmatic would work but the head has to soak up a fair bit of axial movement as the spindle withdraws at the wrong retract rate. May be fine in most cases but for longer holes and coarser pitches it could get a bit iffy. Other issue for my machine is that the reaction arm collar mounts on the threaded nose of my quill that holds the main bearing in place. During the retract, the reaction arm tries to unscrew it, which wouldn't be ideal if it succeeded.

Given these and the fact that I also had a tension compression head and a fair selection of collets, at this point I changed my focus from the tapmatic method. Although I got it working I've now put it back in the cupboard and may bring it out later for use with the Bridgeport.

The tension compression approach is simple enough to implement, the main difference to the tapmatic being that you have to stop the spindle and reverse it, ideally in synch with the infeed and retract moves. The std G84 works fine as long as you don't try to run it too fast. The main difference between this and rigid tapping is that you don't need a spindle encoder.

My tension compression head seems to have around 8-10mm axial movement either side of the unloaded position. But if you are running slow accel and decl times in the VFD, there can be a fair amount of overshoot, so that the tool is still going forward while the spindle has started retracting. I have a braking resistor in my Yaskawa VFD but even so, if I set the decl time to under 1 second, I can't run much more than 400-500rpm without overvolting the VFD. You don't want that to happen with the tap at the bottom of the hole.

Given the need to sharpen up the start and stop times for the spindle (increasing the risk of overvoltage), it's a good idea to tie the VFD error signal into the controller, so that the machine will stop if you cause an overvoltage and the spindle stops. It took a bit of fiddling to optimise the accel / decel rates against the max speed I could run. If you are going to be using a larger tap, there will be less chance of the VFD overvolting as the spindle will be running slower and there will be more torque required to drive the tap. So I did my tests on a thin aluminium section with a small (M4) tap.

Not the most exciting video you will ever see but here's what I got:


I'll be doing some more challenging tapping into aluminium plate shortly and hopefully they will go OK now...