Sunday, June 7, 2015

This Weekend's Main Project Has Been...

Note:  This post has been edited since the project's completion to clear up some technical details I got wrong.  

Getting all the pieces on order to convert my Grizzly milling machine to CNC-based.  I spent a while considering whether I wanted to go that way or just add some sort of digital readouts (DROs).  My bottom line is that CNC costs more than DROs, but not that much more, and for that extra cost it does much more - as well as acting as a DRO.  After researching the topic on a couple of hobbyist machinist forums, I settled on the approaches used by "Hoss" at or  Hossmachine is aimed at the smaller SIEG X2 and X3 milling machines while G0704 is aimed at it and the other machines based on the same basic machine, the BF20L. 

Hoss has a couple of different approaches that he calls Phases 1 through 3.  That added extra confusion to me, and I bought his DVD ROM (based on recommendations on three different hobby machinist forums).  The three phases are three separate approaches to building a CNC system and comprise the two basic approaches to CNC'ing the mill.  Phase 1 is the simplest: it attaches motor mounts to the existing hardware (lead screws).  Add motors, controllers, the wiring need to get it hooked up, and off you go.  This is the way my CNC Sherline is designed and I'm going to do this on the Griz.  Lead screws are serviceable, but aren't considered ideal in the CNC world.

The problem with lead screws from the CNC viewpoint is backlash, that moment when you change directions and the wheel moves before the table does.  Backlash can be compensated for in CNC, just like you compensate for it when you're working by hand by turning the crank farther than you need to when you change direction.  I think I've said before that an old Master machinist once told me, "your machine has backlash; get over it.  They all do".  Unfortunately, software backlash compensation isn't a perfect process and while it removes gross errors, at the scale of "tenths" (.0001"), a circular cutout can take on almost a cloverleaf shape.  That puts a premium on making the machine as backlash-free as possible.

Lead screws have one major advantage, though: they allow you to use the mill with the CNC controller turned off (although you lose the DRO equivalence).  If the motors have dual shafts, you can attach the motor's front shaft to the lead screw and put a hand wheel on the other end.

The alternative to lead screws is to use ball screws (for example).  Ball screws convey motion not by screw threads pushing on each other, but by ball bearings pushing on their channels.  (There's a tremendous amount of information on screws of all kinds at on their "Application Engineering" page).  They say the big advantage of ball screws is lower backlash, more on this later as the project develops, but the price for that is they don't offer holding torque to keep things from moving when you don't want them to.  If I stop turning the hand wheel on my mill, the table stops moving due to that holding torque.  With ball screws, pressing on the table will make it move.  The holding torque is provided by the motor, so it's impossible to use a ball screw driven CNC without the motors running. 

The Phase 2 or 3 versions use different ballscrews from different manufacturers so the hardware for ballnut mounts and such is different.  This is a much bigger project.  The mill needs to be completely disassembled, since you need to replace the lead screws on all three axes, along with their mounts, and modification of the cross slide is required, but the result is a most likely a higher performance system.  I understand that Hoss created Phase 3 due to availability of a lower backlash ballscrew at a better price point.  It's important to note that the Phase 1 system can be changed to ballscrews in the future with just the additional work.  Furthermore, there are options for both Phase 2 and 3 that put the motor on the front or the back of the Y-axis, and an option that adds a spacer between the mill's base and the Z-axis column to increase the Y-axis travel.  Again, going with the Phase 1 lead screw approach doesn't mean you can never upgrade to Phase 2 or 3.  I believe, but am not sure, that Phase 2 is really obsolete, but even if you do Phase 3, there are things you'll need to know from the Phase 2 folders on the DVD.

The lead screws on the X and Y axes on the G0704 are 10 Turns Per Inch (TPI).  One turn of the hand wheel moves the table 0.100 inch (those numbers are 20 TPI and .050", respectively, on a standard Sherline).  A 200 step motor divides that into .0005" per step, although it's highly likely that the movement is not that accurate for every step.  Further, stepper motor controllers divide that .0005" into 8 or 16 steps, called "microstepping" (1600 to 3200 microsteps per rotation).  In principle, that would make your smallest step 62.5 millionths of an inch (8 microsteps) or half that, 31.25 millionths of an inch (with 16 microsteps).  That accuracy can't be counted on.  What it's really good for is smoothing out the motion of the motor.  The Z axis is a 6 TPI screw, so the headstock/cutting tool raises or lowers 1/6" or 0.1667" per turn.  The 200 step motor makes that .000833", and you can do the rest of that arithmetic if you want.

In reality, you can expect accuracies on the order of 1 or 2 thousandths out of a machine like this.

It's important to realize that a product like the G0704 is a system.  During the design, all of the components are chosen based on the expected use of the mill.  When you replace leadscrews with ballscrews, the table can go faster.  All well and good.  The spinning cutter, though, will only take so many cubic inches per turn, depending on how fast the feed moves and how fast the spindle rotates. How fast the spindle can rotate depends on how much power the spindle can rotate the cutter with.  If you want to remove the most metal in the least amount of time, you need a faster spindle than a manual machine's.  The motor's horsepower determines how much force that can be put on the whole machine, which in turn affects how much the machine bends and deforms during a cut; how rigid it needs to be.  Keeping the leadscrews and original spindle avoids some of those secondary effects.  Plus, I think worrying about "removing the most metal in the least amount of time" is where industry lives, not hobbyists. 

Hoss recommends 570 in-oz torque motors and a set of controllers from Automation Technologies.  I'm just going to go with his approach unless I find reason not to.  Add a power supply, (mine will be a 48V, 10A switcher) a breakout board from CNC4PC which turns parallel (printer) port signals into the signals to the motor controllers, throw in some some odds and ends (wires, switches, etc.), and you have a system.  To control the system, I'm going to keep using Mach3, which I bought years ago.  Mach4 was on display at Cabin Fever in April; I need to spend some time getting caught up on what's different about it.  I see it doesn't use the parallel port control, as Mach3 does, and that's a complication I haven't looked into in detail.  My plan right now is to use another old, 1-ish GHz, XP machine that's currently a door stop.  A stripped down old PC with XP works fine as a controller, if it's off the internet.  (In the Linux world, there's a program called LinuxCNC if you absolutely don't do Windows).  Modern computers have largely abandoned the parallel printer port, and getting it to work properly has always been a struggle as Windows takes lower level control of the ports with every version. 

The DVD also includes materials lists, and I went through my aluminum stockpile this morning to make sure I had everything I need to make the motor mounts and the few other pieces of machined metal it will take.  Didn't need to get any of that.

So with the rest of the day to play around in the shop, I played with the Griz mill making a guitar nut/string height adjustment tool based on that one.  Got about halfway through and realized I didn't have an end mill I needed.  I'll go order one from Enco or someone real quick.  This is it cut to size, and with a slot prepared for the digital depth indicator I picked up for woodworking. 
Hoss has a YouTube channel and shows a lot of videos of the things he's done to his Griz.  A ton of videos come on the DVD but several of the videos on the YouTube channel aren't on the DVD.  I'm not sure how much time to spend on this project here on my blog; for one thing, it's not my design and I can't say much about it.  We can talk CNC systems in general as much as folks want.  Once this project is done, and it should be over the summer, honest evaluations are what I'm here for.  I strongly suspect he wouldn't want me showing lots of details that he charges for!

As always, anything you'd like to see covered, let me know.

EDIT 4/24/17 at the project conclusion. to correct some technical errors, errors describing Hoss' DVD and CNC mills in general.


  1. Are you happy with Grizzly's quality? I'm thinking of ordering an 8" joiner...

  2. Watched an interesting movie last night- "Charge", about electric powered motorcycles at the Isle of Mann. Interesting machines and prototyping- you would like it.

  3. I did not convert my mill to CNC, but I did put inexpensive DRO on all three axis. After a fair amount of research I mounted Grizzly's Igaging DRO for a total cost of maybe $ 150.00 USD. One scale or the read out display is a little erratic.

    I jumped over to the Grizzly catalog and see that prices have gone up a little, and now there are two types of inexpensive DRO scales and readouts.

    I would like you to give some mention of the costs of the CNC conversion.

  4. John, the cost varies with the Phase you choose and how much you want to do, but that disclaimer aside, I think my system is going to cost well under $600. I spent about $450 over the weekend. The 3 motors and controllers were $330; the power supply was 75 and the glue pieces about 50. I probably will need to spend some more, but I really doubt it would get me up to $600. I plan to make this one better than my small, Sherline CNC, with more safety features.

    The real expense is in software. I should do a post on that, although I'm afraid my information may well be out of date.

  5. LCB, the quality from Grizzly is more variable than a top end suppliers' quality, but I haven't run into anything that's definitely their problem. Their drill chuck runout was pretty bad, - over .010" IIRC, I had upgraded that when I bought the mill, and bought one from LMS that does under .004".

  6. Thank you for the cost information, and I did check out the website. I was interested in Hoss's mod to use the mill as an engraver. But if I flash back to running a pantograph arm engraver in the military, the machine spun a small bit with a small motor.