[Bubba]

Have you put a "spider" in the back end of the headstock to center the ball screw at that point"? Secondly, when I turn long lengths of stock, I set up a piece of pipe that is larger than the stock (say a 1" or 1 1/2") to act as a stop in case the stock starts to "whip" and bend.

[photomankc]

It wasn't whipping from flex. It was just being held crooked in the jaws by that sleeve so it was exaggerated at the far end. Now it's running pretty true. there is only about 2.5" out the back of the spindle and the part is 3/4" by about 1 1/2 foot. I think I can get by without too much extra support on these short sections. The 3 foot X axis screw is going to require something back there for support.

[photomankc]

Been several days since I posted any updates:

1.) Got the tools and inserts to do the threading for the Z ballnut and got that job done. I didn't have a grooving tool to cut any relief at the back so I just used the threading tool and plunged in a bit deeper than the thread depth and that gave me a place to exit without leaving a massive burr. I chickened out and turned of the lathe each pass at the end using my carriage stop to judge the end. Worked out well. Got real nice threads.


2.) Found a process to machine the ballscrew and ran with it. The best I could do was to dial in using the outside flats to indicate off of. Was a bit of a pain but collars sleeves and all that other stuff wasn't getting it any better than this anyway. Took a good bit of work to get through the outside. The lathe was being knocked around a good bit in the interrupted cutting. If I have to make another Z screw for some reason I think I will give making brazed-on ends a go. Got the end made and tested everything for fit before I pulled it out. Everything on the bearing journal and threaded section was good but on my machine the coupler section was quite long.


I ended up with 0.610" of coupler shaft before the motor would fit on. Took the shaft out the grinder and made a quick flat for the set-screw and it was ready.

3.) Made my own Oldham for the ballscrew side. Used a 3/8" end mill to make a close fitting hole but I think my drill chuck has enough runout to make holes a bit oversize from the tailstock. It's not as close a fit as I would like but seems to be ok.

4.) Mounted it all up on the column to check it out. Seems to be working alright. I was not sure how snug to make the A/C bearings so I just went for no up/down end-play and locked the nuts at that. Spun pretty free like that. With the saddle on and the gib lose I could easily push the saddle down the column by hand. I attached a drill chuck to the shaft and I could easily spin the screw up and down. With the gibs tighter I can just barely turn the screw with finger power on the Oldham hubs but still pretty easily turn with the drill chuck. I figure that is just going to be an adjustment process to get smooth and easy motion without too much squishiness.

Mounted the motor and it fits up to the coupler great so tomorrow it will be time to do wiring

[arizonavideo]

Did you flip your Z collar upside down to let the head go lower?

[Starleper1]

Looking good. One of these days when I get time I will get mine finished.

[photomankc]

It's alive! Z movement. Just need to get the 3 bolt mod done and get the head mounted up and started testing. I'll try to get some video later of it. Electronics didn't even fight back!

[photomankc]

Looks like the pc is going to be the limiting factor on speed. I cant get past 85 IPM at 8 microsteps. I get following errors with EMC2. I'm not sure I care much though. 75IPM is looking pretty fast to me really.

[photomankc]

The good news:
- 0.0015 backlash. I'm good with that. Roton added oversize balls for me on the order and that seems to have paid off.
- 75IPM with the head back on - I'd like to get a PC that will allow EMC2 to hit 100+ but that's a low priority.
- Movement accuracy over 6" is inside 0.001". I'm happy with that.
- About 12.5" of travel I don't want to push to the very edges yet without limit switches.
- 3 Bolt mod is done.


The not so good:
-There is a wobble. I can't track down where it comes from. I checked the screw 6 ways to Sunday in the lathe but I think there may have been a slight bend in it or something. After I cut to length there was not much sticking out the back to do anything with but you could see a slight wobble at the very end. I'm picking up 0.009" at the bottom of the Oldham but part of that has to be the not-so-great fit on the 3/8 hole in the Oldham coupler I made. It's visible though which leaves me concerned. When I rechucked in the lathe to do the finish work I double checked against the screw and still was showing less than 0.001" on my cut diameters. I'm kinda stumped on where it's coming from exactly.

I guess I will see how it goes. If I eat bearings or ballnuts then I guess it will be time to revisit this screw.

- The holes on my three bolt mod didn't come out very nicely. The drill press I purchased with this as a project in mind was still not really big enough. I couldn't find a good way to clamp the thing down well and it moved around when the drill started to bite in so all my layout work came to little avail. However the 15/32" drill had enough clearance to make up for the out of position holes and it all fit up. Just not as neat looking as I had hoped. I guess I don't look at the rear of the mill head often enough to sweat it though. It fits and I doubt I could have done better hand-drilling.

I'll try to get it trammed in tonight and see if the three bolts help with the head nod I was showing before.

[photomankc]

Did you flip your Z collar upside down to let the head go lower?

Nope. I don't think it's possible on these. The dovetail is not symmetrical as I recall. With a setscrew holder or a collet chuck I can reach the table well enough to machine right into it if I'm not careful so I don't need it to go much lower.

[SCzEngrgGroup]

- 75IPM with the head back on - I'd like to get a PC that will allow EMC2 to hit 100+ but that's a low priority.

I'm baffled by that comment. Nothing in the PC or EMC could possibly be limiting you to such a low rapid rate. That represents an incredibly low step rate (about 1.6kHz) with direct drive steppers and 5-pitch screws. The worst PC in the world can do 15X that rate.

Regards,
Ray L.

[photomankc]

I'm baffled by that comment. Nothing in the PC or EMC could possibly be limiting you to such a low rapid rate. That represents an incredibly low step rate (about 1.6kHz) with direct drive steppers and 5-pitch screws. The worst PC in the world can do 15X that rate.

Regards,
Ray L.

I'm saying that because the machine gripes about 'following errors' if I go any faster while the stepper shows no issue going at that rate, and there is no feedback system so I was assuming that to mean it can't issue the step commands fast enough. Have not yet dug into it.

My driver is on 8 microstep, That means 8000 steps per inch or 600,000 for 75 inches in an minute. Divide by 60 and you have 10KHz. The EMC2 Kernel requirements for latency are what's hammering me. I have bad latency and it can't be corrected on this model of PC so my kernel speed is terribly low. If I went to 1/2 step I could probably go much, much faster but I like the smoother motion on 1/8 step. I may try 1/2 step later and see what happens.

Still working around with it. EMC2 is nice but it's picky about the motherboard and it's well documented that new computers can give disappointing results compared to older systems with better real-time performance.

[photomankc]

Machinechick's 3 bolt mod is fantastic! I had 3-4 thous of forward nod that I had to shim out as best I could before. Now by torquing the upper bolts tighter I can pull the head back and get all but 1.5 thous of the nod out that way. Once they are tight I can torque down the rest and hold it. Glad I took the time to do that. Just need to make a cardboard spacer to hold the bolts in the circle next time to make getting the head back on easier.

[Starleper1]

Glad to hear that. I didn't re-assemble mine yet but I had .002 variance before, Hopefully that will be gone once its back together.

[photomankc]

Video of me yammering on about the conversion and of course a machine cutting air.

[nomedia="http://www.youtube.com/watch?v=MSjMbKRGo2M"]YouTube - 3 Axis Complete.mpg[/nomedia]

[machinechick]

Glad to hear you like how the three bolt mod works. Sure makes tramming easier.

[photomankc]

Just found this for EMC2 and it may solve my PC's latency issues:

Pico Systems Universal Stepper Controller

Basically it connects to the parallel port and takes care of generating the step/dir pulses based on data that EMC2 writes to the board from the much slower servo thread. This means that I can untie the CPU from generating the step pulses. As far as EMC2 is concerned this appears to be a servo system. Kinda spendy but I was looking at building another PC anyway so perhaps this might be easier. Makes some of the interfacing easier too with built-in watchdog and limit/estop.


On another note: I pulled the DRO off the Z axis. Running several tests showed me to be well inside 0.001" of commanded position over 6" of travel and with backlash comp it's within the same when reversing. In short, on the Z axis the DRO is redundant. I can hit a Z dimension no problem as long as my touch-off is accurate and the head is right where it says it is.

I still like my DRO on X because at the ends of the travel backlash and position error increases on the acme's and I can see if it's where I think it ought to be, or I can compensate if not.

The 570 motor shows an amazing amount of torque. It will still move the head with both gib locks tightened down. It will loose steps raising it but will still move it. Even my wobbling ball screw is really making clear what a difference screw friction can make.

Last night I lived up to the definition of CNC I saw once: "CNC: Making incorrect parts and breaking stuff, faster and with greater precision". I swear, I messed up a part damn near every way possible in no time flat. All my own fault too. I'm totally not used to dealing with Z moving around. Came within a gnat's hair of punching a hole in my new GMT vice too! :0 Gonna have to really pay attention to detail now.

[SCzEngrgGroup]

You could buy a lot of PC for $250! I paid $85 for a complete 2GHz Dell system....

Regards,
Ray L.

[photomankc]

I know.... just an option that's out there. I'm not sure I think it's completely worth it either honestly but it also makes a few hookups a bit simpler, adds a charge pump, and gives me spindle control and relays without having to interface in a bunch of other this and that boards. I'm going to spreadsheet out what it would cost to do all those plus another surplus PC and see how that compares.

[photomankc]

Ah, hah...... found the problem. For some reason the Base_Thread parameter in my ini was set to 100000 microseconds. I went looking after I double checked the math on the jitter in my system and found I should be able to manage 14 to 16KHz with a substantial buffer to boot so something was not right. When I looked in the INI file at the base_thread number then it was obvious why. Lowering that to 60000 lets me move at 120IPM and the head was still doing fine but the wobble starts to become more noticeable. I'll stick to 75 until I fix that.

Thanks to Himmy for planting the seed that made me keep thinking that was not right.

[photomankc]

One part, start to finish. This is for the replacement tool rests I have been making for the Horrible Freight bench grinder I got for my birthday. The sheet metal rests were useless for just about anything so I came up with this to replace them with something that was angle and height adjustable:
[nomedia="http://www.youtube.com/watch?v=6MVQFPNCdkg"]YouTube - First CNC Part[/nomedia]

(The finished product is shown at the end of the video)

I apologize for the audio. I did the narration track separately and I'm still learning to manipulate the levels.

Most of the code, except the drilling and counter bores was hand written. I couldn't get CAM to do it the way I wanted to since it likes to do a conventional pass and then a climb pass and I can't climb heavy-ish cuts with my ACME backlash. I also hate having to MDI/Jog my way through squaring stock. Drilling and counterbores were handled by Vetric Cut2D.