RE: Re: Scratch-built CNC mill

(conversation between Andrew Werby and Carlos Guillermo...Formatting this
reply to show who said what and when confuses me. Any tips? Carlos)

> -----Original Message-----
> From: Andrew Werby [mailto:drewid@...]
> Sent: Wednesday, August 25, 1999 5:53 AM
> To: CAD_CAM_EDM_DRO@onelist.com
> Subject: [CAD_CAM_EDM_DRO] Re: Scratch-built CNC mill
>
>
> From: Andrew Werby <drewid@...>
>
> Carlos wrote:
>
> >...for what I want to do, I really need a reworked quill feed and fast
> feeds. I saw the CNC Junior a few months ago, and I'm curious as to why
> they only get 50 ipm with such beefy motors. I would think the drivers

may

> be under-rated for the motors, the pitch/gearing to the ballscrews is not
> optimized, and/or the mill/drill just has alot of friction in the
> dovetails.
> I've seen alot of systems that just don't match the motors/drivers/gearing
> very well, and end up losing alot of potential performance.
>
> [I wonder if this is the case- how would one figure this out? My guess
> would be that there is a lot of inherent friction in the dovetail gibs of
> the imported mill-drill they use, since it wasn't designed to be a CNC
> machine. But optimization of the motor/gearing would be a step forward, I
> agree.]

You need to know the loads and the motor performance. You can measure
masses (inertia), slide friction, drivescrew nut friction, and other loads
on the motor, but it's most helpful to have a torque vs. speed graph of the
motor/driver/power supply system. I've been looking at the Microkinetics
catalog for this. If you have the right match of voltage, phase current,
phase winding (parallel, series, half coil), you can really get alot out of
a motor. Techno-isel seems to do an excellent job of this, from what I've
seen. The torque-speed graphs show how quickly torque drops off with
increasing speed. That's the part I'm most concerned about.

> I would probably want to do the conversion myself if a mill/drill did the

job for me.

> [You probably would want to do so much work to it - replacing or
> rescraping
> the dovetails, redoing the quill, etc, (it seems like you're rather
> particular about this stuff) that you might decide it's not worth
> it and do
> what I did- get an older but better-built CNC milling machine and
> concentrate on replacing the obsolete control system with a
> modern PC-based
> one. This in itself turns out to be a fair amount of work...]

> [I'd never heard of this material, (castable low-friction polymers -

Moglice,

> Belzona, Super Alloy) but it certainly sounds like just the
> thing for the job. Thanks, everybody who brought it to my attention. Has
> anybody tried using it for linear bearings, or is it uneconomical for

this?

> Where would I get some?]

I've been designing around Super Alloy Black 1500LFH from ITW Philadelphia
Resins (http://www.Phillyresins.com) They've got both liquid and putty
versions, as well as many other neat epoxy systems. Yes, I realize I'm
giving away all the secrets of my machine, but if someone beats me to it
with better results (and better pricing) I'll gladly buy one!)

> > [Why go through all this work and then use aluminum for the working

parts?

> > You didn't believe in it above, in the case of Sherline and MaxNC- why

put

> > it in your scratch-built machine? Or am I misunderstanding you?]

> The main problem I have with Sherline and MaxNC is (I believe) that the
> sliding parts are anodized aluminum against anodized aluminum.

> [This is true.]

> If this is true, wear will be much greater [sic - lower] than with

low-friction polymer cast onto one

> of the surfaces, especially for the speeds and loads I hope to see on my
> machine.
>
> [Oh, I see what you mean. But for flat-to flat, wouldn't using
> sheet teflon
> make more sense than casting in place? I can understand using it for
> matching odd contours, making nuts for odd threads, etc, but if you were
> just trying to make a sliding gibbs, would it really be economical?]

Teflon alone is very soft and flows under load. There are many filled
versions, but the problem with glue-on sheets is that you still need to
accurately machine BOTH surfaces, and they would have to match each other
precisely and be parallel to the other side via scraping, honing, grinding,
etc. I'm hoping to get away with making sure one piece is very straight and
smooth, and the other piece is very rough and very loose-fitting, and then
injecting the polymer to fill the void. Many times you can get away with
jacking screws built into the slide to align the parts, and a couple of
other alignment aids.

> Also, the damping characteristics of the polymer is one of it's
> biggest selling features for use in new machine construction and machine
> tool rebuilding. I realize the stiffness of aluminum is much
> less than iron
> and steel, but I'm hoping that if I reduce or eliminate the "weak
> links" in
> the slide assembly design, such as gibb adj. design and fit-up of
> surfaces,
> and I mount the slide assemblies to a welded or cast steel frame,
> I can have
> a reasonable well damped, stiff machine.

> [You're using aluminum because you want to save weight on the
> moving parts,
> or is it because of the design conveniences afforded by the use
> of aluminum
> extrusions? ]

That and more. True, saving weight on moving parts reduces the demand on
the motors during accel/deccel, but the potential for extruded slide
components is the biggest motivator. Anodized aluminum works very well with
the low-friction polymer I'm using, as long as the surface is uniform
initially, which is a plus over steel or iron. The hardness and durability
of hard-anodizing is usually superior to heat-treated steel or iron (Rc
70+??), as long as you don't break through the coating to bare aluminum.
Plus, aluminum cuts like butter.

> > [This sounds like a good design for a gantry-type router, along
> the lines
> > of the Techno-Isel, but with more Z travel and less Y. I think
> there is a
> > market for something like this, especially if it had more Y. The routers
> > out there seem mostly oriented towards sign-carvers, but a more
> > general-purpose machine would be better for the rest of us. I'm in the
> > process of putting together something like this myself,
> although I wasn't
> > thinking of it as a steel-carving machine. I wonder if you will
> > really have
> > as much rigidity as you need without having a big "C" casting in there,
> > like the ones real milling machines all seem to use.]

> Well, with the modular nature of the extrusions, just such a machine might
> be possible. I considered a similar layout with the table moving
> instead of
> the gantry. I think you could get good rigidity that way

> [I've been going back and forth about this one myself. I think there might
> be a lot more slop in a gantry system when one starts adding the
> off-center
> force of an extended Z axis- perhaps that's why I haven't seen
> any machines
> like this.]

I've actually been finding alot of high-performance, high speed machining
centers using this layout. I think they call it a "portal" layout, where
the gantry is fixed and the table moves.

> > [This sounds cheap. Where are you getting these inexpensive
> > ball-screws and
> > linear bearings? Last I checked, they cost more than that for a machine
> > this big. The source I asked wanted about $1000 per axis for the
> > screws and
> > nuts to retrofit a Bridgeport. If you really can build all that
> for $1500,
> > sign me up for one.]

> The linear bearings are replaced with the low-friction polymer
> and extruded
> dovetail slide system. An excellent source for cheap high-precision
> ballscrews is Thomson. They have rolled, .001/ft accuracy ballscrews for
> $1.50 per inch.
>
> [Does this mean you lose a thousandth in accuracy for each foot of
> length? So if I had an 8-foot screw I could count on being out
> +/- .008? Is
> this because of torsion on the screws while in use, or just manufacturing
> imprecision adding up?]

The specs they give with the screws doesn't account for torsion. Just for
comparison, Techno-Isel ballscrews claim 0.1mm/300mm pitch accuracy. That's
.004in/ft. With a CNC system, this is no problem. The deviation is VERY
linear (constant slope, I mean), so in a CNC control you could change the
multiplier to account for it. Even high precision ground ballscrews can
have pitch accuracies of around half a thou per foot, and you pay for it.
I've talked to some European ballscrew reps which tell me they are amazed
that us Americans are so concerned about screw accuracy in a CNC. Why don't
we just use glass scales??

> [Thanks for running this down for me, Carlos. I think the .631
> screws would
> work fine for my router. But do you think they're sufficient for your
> steel-eater? The screws on the mill I've got are about 3/4". By the way,
> these things have to be lubed constantly- we've been fixing the oil system
> in our old mill, which is something of a chore. You need to add the oiler
> and (expensive) oil-lines to the cost estimate you're working up.]

Well, I don't know if I'm hoping for a "steel-EATER". Maybe just a
steel-snacker. I want to be able to run the occasional steel project, at
slower speeds of course, without (the machine) screaming, hollering, and
spitting. I'm going with the .631 screws for cost reasons, and I'll just
have to see how they do.

I've been fearing the oiling chores. I've got some ideas on lube ports, and
maybe I can integrate a simple plumbing system into my design. II've
sketched out a simple spring-loaded oil reservoir that might do the trick.
Does your oiling system feed the ways and the screw nuts with the same
pressure? And are you using ballnuts?

> [Andrew Werby

Carlos Guillermo