library wrote: Well, just as an update I decided to go with a VFD and a 1.5HP motor. I'm pretty sure I can make it fit because I'm adding a spacer between the column and head of the mill to increase the y axis travel and this will give me the few inches I need. I'm also changing my power drawbar plans to use a hydraulic ram instead of an air cylinder so it can be narrower. With those two parts in place I'm pretty sure I can make it work. I would love it if large servo type motors were cheap and easy but honestly VFD's and AC induction motors are the cheapest easiest way to go.
I found a screaming deal on a 2HP Emerson Commander SK VFD (the exact same one Tormach uses) and I wanted to go with a 2HP motor but they are too heavy. Also, I decided to be realistic and 2HP is probably excessive compared to the rigidity of the mill. The motors I'm looking at have 1000:1 constant torque so they will provide full torque throughout the entire low range. I'm planning on mounting a cheap rotary encoder on the back of the motor.
So let's clarify the encoder part. I'll more than likely be running the motor with a 1:2 pulley. BigJohnT indicates that I need the encoder to match the spindle speed, not the motor speed. I did some more reading and it appears what I really need is the index pulse to match the spindle speed, the A/B quadrature output from the motor encoder can apparently be scaled. I could probably put an index only pulse on the spindle, that seems like an easy electronic part to make.
I am using one on Jons pwm-servo amps to drive the stock 1 hp motor, but using a 126 volt supply good for around 20 amps surge. So its likely I'm peaking at around 2 hp.
I tried to make an all optical encoder disk but quantization noise from the number of slots available destroyed low speed, hammering the motor so bad it sounded like the spindle bearings were destroying the ball cages or had square balls.
So I bought a 1000 line Omron encoder, and made a short shaft extension to drive it off the top of the motor. Then I put a notch in the shift knob rim, and added a couple switches so LCNC knows what gear its in. Then I added some counter stuff to my hal so I could determine the gear ratios. I and also using the index pulse from the optical encoder, which gives LCNC a one full turn reference. A mux4 tied to the gear knob tally's selects the encoder scale, so the tach is still correct. I took advantage of the lack of a gear tally to feed an offset into the mux4's unused inputs so the motor if on, is getting about a 30 rpm at the motor signal, so when the gear is between home positions, the gear change is smooth as it meshes into each gear, so i don't have grab the spindle by hand and turn to mesh the gears.
Counting a A pulse for 100 turns of the spindle gave me the two "scale" figures. I use one in the .ini file, and scale it by the ratio when in the other gear. The numbers are just over 7000 and just over 14000. Quantization noise is gone, and the Pgain is raised from 1.5 to 20 and its still dead stable. So speed control is extremely stiff and I am unaware for how hard its working until I hear the chirp from the motors iron as Jons amplifier goes into current limit at about 17 amps. The motor is rated at 9.7 amps, but I've not hurt it in better than 2 years. I do have a limit3 in the reversal circuit to slow that so the z drive can keep up, but I can still do a turnaround at the bottom of the hole in 400 milliseconds even at 3000 revs. Now, if the rest of the machine was truly square, I'd have a real mill. But I lack the gages to determine where its off.
Best mod ever for one of those machines.
