Monday, April 24, 2017

Completing the G0704 CNC Conversion

To the extent that a project like this is ever "Done", this one is done and what I write from now will mostly be either what I do with it, or improvements I make to it.  The mill itself is now done.  I think all CNC hobbyists end up tweaking their machines as part of their hobby, though.

I started putting together a project page which appears in the "Special Pages" list in the right hand column.  I've moved that up to just below the Florida Gun Blogs section.  It became apparent to me that the page needed an introduction and a conclusion; this is the conclusion.

The last thing I wrote about the project was that I had a small but persistent oil leak out of my oiling system onto the floor.  Small as in 3 or 4 drops in a day.  As my last fix, I replaced the push to connect fittings with plastic barb fittings.  Because I used 5/32 OD, 1/8 ID tubing everywhere, I needed a fitting that went from 1/8-27NPT male threads to a barb in that size.  I couldn't find one, but I could find a fitting that I could adapt to and make work by using a short piece of 1/4 tubing.  It ended up looking like this, and has not dripped one drop of oil in 5 full days.

People considering building something like this will wonder if it's worth it.  In response, I'd have to ask why you'd consider building it.  Do you need it now, urgently? Are you trying to learn what's in a system like this and make one, just because?  

If it's purely to save money, it might not be the smartest thing to do.  The G0704 is $1359 today (including shipping).  Automation Technologies will sell you a ready to run, CNC converted G0704 with an Ethernet interface for $4000 (I don't think that includes shipping while the Grizzly price does).  That's paying $2620 more to have it now as opposed to working on the conversion as long as I did.  I'm not 100% sure I've captured every expense in my journey, but I think I spent close to that amount, maybe more.  That prebuilt machine doesn't appear to have an oiling system, a chip tray, and an enclosure, all of which mine does, and all of which were considerable effort and expense to get working. 

How long did it take me to build it?  I actually started in June of 2015, six months before retirement.  Once I had all the parts made for the approach I was taking, in March of last year, I switched approaches rendering all of what I'd made useless except for the Z-axis motor mounts.  I started in early April '16 and had the system running on all three axes, ready to use this February 2nd; 10 months.  It wasn't full time work, but I worked several hours every day, weekends included. 

A really good machine that has more travel than mine, is the Tormach PNC1100 from Little Machine Shop at $8500.  Tormach is industrial quality without a doubt, and I have no doubt it outperforms mine.  It still isn't complete as it doesn't have an enclosure around it and I'm not sure if it includes an oiling system.  The next cheaper Tormach that LMS carries is priced at $7000; it has less travel than mine and a motor rated for less power.  Both of these machines include things I can only aspire to at this point - automatic tool changer systems, and tool holders (the Tormach Tool System - as an add-on, it's half the price of their milling machine) and very likely more.

So I spent about what the Automation Technologies G0704 costs but ended up with a more complete system.  I also made and remade a lot of parts and got experience I've never had before with the tools.  With my $2600 or $2800 I not only bought the conversion, I bought an education.  If my only goal was to have the completed machine, it's a perfect illustration of "penny wise but pound foolish".  If that's your goal, buy one of those machines.

On the other hand, think of a mill like these as a system; a chain of parts that's subject to the adage that the chain is no stronger than the weakest link.  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 spindle can cut off chips of metal without stalling.  How fast the spindle can rotate under load depends on how much power the motor is able to deliver to the cutter.  If you want to remove the most metal in the least amount of time, a CNC mill needs a faster spindle than a manual machine's.  Mine spins at 2300 max; those Tormachs can spin at 10,000 RPM.  The motor's horsepower determines how much force ends up getting put on the whole machine, which in turn affects how much the machine bends and deforms during a cut; which determines how rigid it needs to be and how big and heavy it needs to be.  Plus, I think that worrying about "removing the most metal in the least amount of time" is where industry lives, not hobbyists. 

Don't go thinking "I can put some ballscrews and nuts on this machine and it will be zero backlash and as good as Tormach".  First, it's possible to end up with backlash - I did.  That's in my list of stuff to tweak on and make better.  Second, like I said, this is a system and you're replacing parts in a system with upgrades expecting it to get better.  It's also possible you'll improve one thing and discover weaknesses in other areas.  The way I look at it, you generally get what you pay for.  The Tormach has the advantage that it was designed from the ground up to be a CNC mill on a small production line.  You don't buy the $4000, ATI-built G0704 and expect it to equal the $8500 Tormach - which itself is not going to equal a $50,000 HAAS. 

If you're the kind of person who approaches most problems with "close enough for government work", machining is probably not for you.  I don't have any measured numbers on what sort of accuracy to expect in the converted mill, but I think these are a good choice of a home machine for the vast majority of your mad scientist projects that will fit on it.  The mill as received from Grizzly was a good performer as a manual mill, and it should still be one.  You're not going to cut a titanium turbine for a 747 engine on it, but maybe you can cut one for a model engine. 
5-axis machining is possible with the G0704, but not the way I've configured it now.  I have all the parts of a fourth axis that I intend to get running soon. Maybe after the GB-22. 

EDIT 4/24/17 2025 EDT - Added the link to the project's page in the second paragraph


  1. Extraordinary work for the conversion. And your mention of buying an education as well as a machine is exactly right.

    Well done. (except that as you said, done is a sort of flexible concept)

  2. If you look around you can usually find a Robodrill in still-functional condition for about $8k. I see them all the time. What you got was an education, and that's probably worth the money you spent.