G0602 Conversion

[tmbg]

yeah, it's the CDCO chuck. I had to order backplates from somewhere, I think it might have been MTW, and turn them down.

Turning the backplates was a real chore until I figured out that cast iron wants a really slow speed. After that, it machined like a dream.

[photomankc]

Nice! I wish mine were converted The ease of making radius cuts is not fair.

Hey, how much work did you have to do on that CDCO collet chuck? Mine proved to be a real dud and I exceeded the return period before I tried to mount it. It had huge conical runout and the register and taper were not concentric. It's a real mess. Did you have to fight much with it or was it pretty good as it came?

[doorknob]

Were you able to do that on the G0602 or did you need to use the Clausing to turn it down?

[photomankc]

You can turn the backplate on the G0602 no problem. You have to remove the splash guard to get up that far and you need to run at the low end of the RPMs.

On the chuck I got, after facing and cutting the register down (the standard 3 jaw backplate for the 0602 has the right hole pattern and diameter) when I mounted the collet chuck I was really disappointed. Almost 0.0015" of runout *per inch* you moved away from the chuck. When I checked the rear by mounting a trued bar in my 4 jaw and then mounting the collet chuck to the bar the rear face was WAY out of square and the register was close to 0.01" off concentric. I don't know if it was made on a friday afternoon or what but it's a disaster and the only way to fix it was to cut the chuck.

I finally got 2 new backplates to allow me to go at it again since in order to fix this I had to open the register on the CDCO chuck up to fix that. I did seem to cure the conical runout by refacing the rear of the chuck but I cant check the radial run out till I try a new register. It's now on the back of my project list though.

[tmbg]

mine wasn't nearly that bad. I had maybe .005" of runout when I first mounted it, but tweaking the mounting bolts while spinning a test bar I was able to get it down under .001". It's not perfect, but it'll serve my purpose for now.

[tmbg]

worked on the spindle encoder a whole bunch today.

I redesigned the index wheel to have more appropriate vane depths for the optointerruptors I'm using. I drew a solid model of the interruptors, and created a mounting plate model. I put them all in assembly and got everything lined up the way it needs to. The trick to doing the quadrature is two of the encoders have to be some even multiple of the vane angle plus a quarter. You want the signals 90 degrees out of phase. In my case, I ended up going with an 18 tooth wheel, which has a rising edge every 20 degrees. The two rightmost interruptors are spaced at 25 degrees. It could have been 45 or 65 and been the same. If it had been 35 or 55 instead, it would still work but the direction would read backwards.

One slot is deeper than the others, and one interruptor is mounted deeper to catch only that slot for an index signal.



Once I had everything lined up in cad and dimensioned the way it needed to be, I cut the disk and the mounting plate on the plasma table, and then threw the plate on the mill to drill and tap it. I didn't have 1/8" steel on hand, so it's actually two 16 ga pieces which were drilled and tapped together.



I fiddled with it a bunch to get the interruptors lined up exactly where they need to be. I ended up having to drill and tap new holes for the index interruptor to get it VERY deep on that wheel to catch the index pulse without getting false triggers from the regular teeth.




After scoping out the signals and confirming that all was as I expected, I changed the belt out to the high side and set it for max speed (2400 rpm I think). I was a bit concerned about rise time at this speed, and when I scoped it, my fears were confirmed:



I swapped resistors out to stiffen up the bias and reduce the RC constant against the scope's input capacitance, and it helped:



40us rise time is more than acceptable for my application. I'm not 100% how far apart the input capacitance of the Mesa's FPGA pins is going to be from my scope's input capacitance, but once I get it hooked up to the Mesa I'll scope it out and make sure it's not rounding too much. I also need to take into account cabling capacitance. If that becomes an issue, I may need to do a differential driver, and that's more work than I want to get into right now.

[tmbg]

and here's a video of the encoder running with the spindle at 2400. This is before I swapped out the bias resistors, and what you're seeing is the A and Z pulses only.

[nomedia="http://www.youtube.com/watch?v=IlVsiA06GqU&feature=channel_video_title"]YouTube - G0602 spindle encoder[/nomedia]

[michaelthomas]

I wish even a small portion of that made sense to me......lol. Cool!

[photomankc]

Awesome! Electronics and machine nerd content all in one post!

[tmbg]

I made a little board that lives at the encoder head which has the bias resistors. This way I can consolidate my wiring down to five lines: 3.3V, ground, A, B, Z.

[eartaker]

tmbg,

Why did you need 3 sensors?

[tmbg]

Quadrature coding (grey code) is done with two signals 90 degrees put of phase with each other. Grey code is useful because you get not only speed but also direction. The third sensor is for an index pulse, so the control knows where in the rotation the axis is.

[tmbg]

Got spindle synchronized motion working. Here's the obligatory threading vid!

YouTube - imcmahon's Channel

[kd5ahl]

<C-3PO> Well shut me down, machines making machines. How perverse!</C-3PO>

Whats the grumbling sound on the last 5-10 passes? Is that chatter due to a straight infeed? I did not hear it on the first few passes.

Very nice speeds though!

[tmbg]

I'm not sure about that chatter. I'm using a G76 threading cycle, and it has a compound angle that you can set, and I have it set to 29.5 degrees. The cycle I ran before the one I videod cut in much fewer passes, but the chatter toward the end was pretty significant. For the pass in the video, I set the "depth degression":

R-
The "depth degression". R1.0 selects constant depth on successive threading passes. R2.0 selects constant area. Values between 1.0 and 2.0 select decreasing depth but increasing area. Values above 2.0 select decreasing area. Beware that unnecessarily high degression values will cause a large number of passes to be used. (degression = a descent by stages or steps.)

I set it to 1.5, and it took way too many passes to complete, and still chattered.

Anyone know anything about that? I need to learn a bit more about feeds and speeds for threading and for parting.

Here's my G76 cycle, btw:


M3 S720

G0 X#<_driveline>
Z0.1

#<D> = #<_driveline>
#<I> = [#<_major_dia> - #<D>]

G76 P[1/14] Z[0-[#<_length> - .1]] I#<I> J.0025 K.0619 Q29.5

[Maglin]

I was wondering if it wasn't the insert rubbing instead of cutting with those very light passes. I've been told to dress up the inserts so they are sharp and cut instead of shear and it will help loads. That is what I'm told. Nice video btw. I'm definitly going to have to cnc a lathe someday.

[michaelthomas]

Very nice, tmbg. I can't wait to get mine going.

[tmbg]

So, i fitted optointerruptor home switches on it, but they suck. I need at least the X axis to be repeatable to a pretty high degree of precision, because all my tool offsets are based on where the machine thinks spindle centerline is. The opto is repeatable to about 0.010". Terrible.

I have a roller switch to use instead, I just need to figure out how I'm going to mount it

[michaelthomas]

That's a bummer about the switches. I was planning to use those as well. I really didn't want mechanical switches, and hall effect switches are probably not a great idea around ferrous metal.

I have seen a few people that use them is sealed housings.....I wonder if I could find those pictures again? Probably not......

[michaelthomas]

As I was reading about optical homing switches, I came across this .....


For real precision, I've used slotted optical switches. The secret is to use an op-amp comparator on the output to give you a constant "hard" cut off. I've got better than 5 micron repeatably, the limiting factor being vibration.

It means very little to me, but may be helpful for someone with electronics knowledge.