Official Luthiers Forum!

Following my exit from the adhesives industry I took some time to carefully analyze my knowledge, skills, and dreams. I also got some fantastic input from several members of this forum. For the past 15 years or so, in one way or another guitars have been a common theme. Together with a partner who I've known since I was five and taken on countless ridiculous projects with already, I set about founding a business and making plans to acquire and fabricate equipment. Among the most critical was a 3 axis CNC router we designed together to be constructed primarily from aluminum extrusions and off the shelf parts. For all of the structural and mechanical parts, including all specially machine mounting plates, lead screws, anti backlash nuts, and fasteners... We pulled it off on a budget of just barely over $1000, which my partner was able to finance with his pile of scrap metal. Since three phase is not an option in our garage shop space, I opted to acquire a Porter Cable 75182 3-1/4HP variable speed spindle. With a relatively cheap upgrade of the collet, nut, and spanner total indicated runout is now down to 0.0002" and when everything is dialed in, we expect overall accuracy of 0.001". While commercial machines can exceed this by an order of magnitude, this is more than sufficient for building guitars.

Here's a picture of the beast from about a week ago after our first successful test of the motion systems:



And a video of us doing a quick dirty check of function using scrap MDF, a piece of paper, cardboard, and tape.

http://www.youtube.com/watch?v=J8W1S6HbF4k&feature=plcp

I'm working on a Strat style neck for a project of mine to test this out on as its first official machined part. At the moment I have two commissioned projects in progress, which will be the next in line and our first production instruments.

There will be lots of pictures to come, and I'm sure I will make some type of announcement when out website goes live and we officially open for business. For the time being, unlike most shops, I plan to take on special projects without any charge for design time. Anyone willing to supply their own materials will be able to have parts made for the cost of shipping and machine time. I suppose I've wandered from the original topic, but I'm more than a little bit excited to take the exciting and risky step of following my dreams, refusing to compromise, and founding my own business.

Now... Back to designing a nice torsion box to mount this puppy on.

Lookin' good Hugh, especially for what you have into it. I've had my machine running for about a year now so if you have any questions, feel free to ask as I was probably there not too long ago myself.

Thanks to the "generosity" of my former employer, I've got the summer off and am in a similar boat of choosing a direction. Interesting times.

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

I need to stop by the shop soon so we can build our control computer, it's going onto a nice 7ft server rack and will make a great setup. The machine itself is technically done (pictures and videos will follow shortly) and just needs to be dialed in to maximize accuracy and precision. During the past week I decided it would be a good idea to build a torsion box in order to provide a nice rigid surface on which to mount our gantry. Fortunately, this coincided with one of the hottest days of the year and later a power outage. The design was pretty straight forward: the CNC gantry is a bit over 2'x3' so I figured 3'x4' would make a nice size for the torsion box. I drew the whole mess up in SolidWorks to verify all of the geometry and finally ventured out to grab some MDF. I've seen all manner of construction methods for torsion boxes, and figured life would be much easier if I broke out my stacked dado blades and created a web that would simply slide together into an interlocking grid. After a bit of cutting, here's what I had:



A couple of things turned out to be great about this: using a scrap of MDF from a previous cut, it was possible to index a stack from exactly the same point. After all, parallelism is more important than perfect spacing and they all ended up within 1/32" of where the plans called for. The next torsion box I make will most likely use spacing derived from the table saw miter slots. There is also a very good chance that I will have the strips cut where I buy them... As long as the thickness is close, it will be very consistent and the sawdust won't be my problem.

Assembly has been more exciting than I intended. One of the pieces was installed in the wrong orientation, which threw a wrench in the idea of easy assembly from everything being parallel... it was close, but it created a bit of tension and made assembly more challenging. I used Titebond Extend, and prepared some sizing from it (2:1 water:adhesive) to brush on all of the gluing surfaces.. Definitely overkill, but if it's worth doing it's worth overdoing. Ultimately, we switched over to using some clamped aluminum extrusions to keep stress off the freshly glued joints and used a system of clamps to slowly press the pieces into place. I'm proud to report that it is both flat and square. Here's a picture of the process by candlelight. (Most of central Ohio was out of power, and my shop was included... fortunately I live in a part of town where my place still had electricity.)


Flash was on for this one to see what we were doing


This is much closer to reality.

We're planning to put the skins on this week, probably cover the whole mess with laminate, and finally mount it to some cabinets before bolting on the CNC.

One more thing, I'm getting the cutting paths ready to go and believe I have decided on the first thing I will test the router on:



So far all of my business seems to revolve around 8 strings. This design, which I've just been referring to as the Mark I while it's in development is the first that isn't an artist signature model stemming from an endorsement. It's fun having endorsement deals right off the bat, but I also like designing from my own vision.

I finally have new progress to report. The torsion box is built, mounted onto an old kitchen island for stability and cabinet space, and the CNC router is sitting atop the construction for the time being. It's not pretty enough yet and will be covered in laminate shortly... most likely inlayed with our company logo for good measure. Starting this weekend we will be building a computer and integrating all of the control systems into a server rack mount and isolating it from dust exposure. I would be even happier to run a test cut on a body but that may need to wait still. Much more exciting news is in the works and I hope to be able to announce it soon.

Getting so close it's scary. Being the perfectionists we are, the rackmount system and torsion box will be fully decorated soon. Everything is coming to life, and I'm personally excited to finally start designing fixtures and cranking out parts. For those of you who are interested in getting into the CNC game and don't mind building things yourself: We're discussing the possibility of making this system available as a kit. Our total budget for this build has come in at right around $1500 including the router and control electronics. We have tons of spare computer parts, but even figuring in a complete computer and monitor... This is very doable for $2000. It also gives me an excuse to use my technical writing skills. For those of you who need templates or short runs of parts, we plan to make such services available for less than we should.

Anyways, lots of exciting stuff coming up soon (such as the launch of our corporate website.)

It's alive! Our first all up test of the router system has been successfully conducted. The video below was a cut of what is actually an old version of our company logo. Without pushing the machine to its absolute limits with microstepping, it already has a verified resolution of 0.0005" and cut the test pattern dead on with no variation down to the mil. The bottom line is that I'm a happy camper, especially given the budget of this project and that for glue joints precision of ±0.003" would still have been acceptable. It's the typical satisfaction of doing all your homework, taking care to properly execute the design, working around obstacles along the way without compromising, and seeing it all pay off with an end result that exceeds expectations.


My business partner had a old 7' tall server rack complete with several rackmount cases and other goodies. The control circuits were built into one of the cases. This is all hand soldered, including all of the ports. Judging by the lead dress it looks like commercial hardware, despite the fact that it is sealed inside the case and hidden from the world. Aside from having known him since age 5, it's this type of pride in workmanship and attention to detail that contribute to the reasons I trust founding a business with him.




Here's the whole mess as it is right now. Since everything is up and running, fabricating everything from inlays in the laminate that will cover the torsion box to making a set of panels to fill in the rackmount setup will be a much faster job.




Finally, the test video. Nothing extremely impressive to look at, but I promise the next videos will be much cooler since it will be cutting guitar parts... And a couple of custom pistol grips. The machine was running at 300 IPM, router at 10k RPM using a 90° conical bit.




Questions, comments, and criticism are welcome as always.

Your machine is looking good but, that cut speed is no where near 300ipm. To me that looks somewhere in the vicinity of 100 to 150ipm. For comparison, here's a little vid I took of my machine cutting at 150ipm and to my eye, it looks like it's cutting faster than yours. For reference, the rapid movements at the beginning of the cut are at about 400 to 450 depending on the axis.



(BTW, that is 3/4" ply cut at full depth with a 3/8" compression bit. I forget the spindle speed)

If you've programmed a cut speed of 300ipm and you're only getting 100, it could be a few things. Since your parts are measuring out correctly, it's obviously not your "steps per" setting. Things to check are the max velocity setting, feedrate override, and accell (though I don't think it's that). Run that program in air and look at the feedrate display in Mach 3. If it's showing a rate of 300 then something very weird is going on.

Another tip - the stock Mach 3 screenset is pretty dang cluttered. I'd highly recommend going with this one: http://home.comcast.net/~cncwoodworker/2010.html Ger is a very good dude and the screenset works very well. Especially read up on the automatic Z-set for the tool changes he's got in there. That alone is worth the $20

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

I just re-watched your video with the speakers turned on and it in fact, sounds like your accell is set very low and your machine is not able to get to 300ipm in that very small test pattern. I noticed this by the sound when you were jogging the long axis out of the way. Do the air test and report back with what's displayed in the feedrate box during the cut.

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

The accel makes a huge difference. When I originally retrofitted my machine, I started running some numbers. I thought you guys might find this interesting.

I have my machine set up to hit 600 IPM (full rapid) in the first inch which puts the accel at 60 inches per second per second.

It hits 600 IPM at around 0.83" and 0.16 seconds
It hits 300 IPM at around 0.21" and 0.08 seconds
It hits 150 IPM at around 0.05" and 0.04 seconds

The stock Fadal acceleration was closer to 20 inches per second per second.

It hit 600 IPM at 2.5" and took 0.5 seconds to get there
It hit 300 IPM at 0.63" and took 0.25 seconds to get there
It hit 150 IPM at 0.16" and took 0.12 seconds to get there

To hit twice the speed, you need four times the space. If you include deceleration at the other side of a cut, then at 4X the 'ramp distance' on a cut you're averaging 2/3 of your programmed feed rate. That means I'd have needed a feature at least 0.63 * 4 = 2.5" long or so to be getting around 300 * 2/3 = 200 IPM average cutting rate on the original control if I capped speed at 300 IPM. The new control would average closer to 270 IPM on the same cut, with the same speed limit of 300 IPM, because it only needs 0.21" at each end to accel and decel.

On the original control, the highest speed the machine could hit on a 2.5" long span and turn around would be 420 IPM, with an average speed of half of that, so in reality I'd never average more than 210 IPM over that distance no matter what the programmed feed rate. The new control caps out on speed at 0.83" and 600 IPM, so it would average about 400 IPM on that 2.5" cut but would cut the middle section at 600 IPM.

You can test what your 'real world' acceleration is by making a program that runs your machine back and forth on one axis over say a 6" span 20-30 times and timing how long it takes to run the program. You can solve for your real acceleration from that, and you'll be able to estimate program run times much more accurately after that. If you're running steppers and trying to max acceleration then you'll want to have some way of telling when it starts missing steps. High end mold making machines are running accelerations of 380 inches per second per second (1G) or even higher, just crazy.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

OMG! How'd you do that? I've never had the guts to cut that full depth. I usually go down incrementally with a bunch of step. Now you have me wanting to try it!

_________________
"I want to know what kind of pickups Vince Gill uses in his Tele, because if I had those, as good of a player as I am, I'm sure I could make it sound like that.
Only badly."

You guys don't miss anything. 300 ipm was a typo, it was indeed programmed and running at 100 ipm. Thanks for all the feedback! We haven't started on fine tuning things, but you've provided great additional considerations. More videos and pictures are sure to follow.

Well, couple things...first and foremost, I heard it was possible. Watch this video: http://youtu.be/rxVa05i1kbQ at about the 3:15 mark you can see them cutting 3/4" ply at 1300ipm (not a typo) at full DOC. That's a chipload of .072!!!! I just scaled my feed rate down based on an estimated chipload that my spindle could handle plus a little extra "pucker reducer" (around .004/tooth but, I think I could double that).

Second, when I first got my machine I forgot to zero Z one time and did about a 1" deep cut at about 250ipm in a 2x4 I had clamped to the table to play with and nothing broke

Important reminder though - hardwoods aren't plywood! When I'm profiling a fretboard I do lots of small cuts at about 300ipm.

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

Indeed it does, and even more so for Mach 3 users. Mach 3 has a pretty lousy constant velocity mode. It will round corners like mad if the accels are too low for the feed rate. There are some tricks to help it but over all, you get what you pay for. Exact stop isn't a good solution either.

Supposedly, Mach 4 is going to bring a bunch of improvements as far as that goes - S-curve acceleration, deviation from path spec etc. etc. but I'll believe that when I see it.

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

Per the comments, you'll definitely want to improve that acceleration. You'll get much better performance that way for sure.

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

You can run a pretty angry feed per tooth on wood. I use 0.012" on aluminum, and you can manage at least 1.5X that on ebony safely. The problem I've noticed most when running really high feeds in wood is the material shattering, so that needs to be taken into consideration. Ebony, especially, loves to split right down the grain if you're cutting a slot in a quartered piece.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Hugh, is this a closed loop machine or open loop? Also I'm curious to it's repeatability, if you dial in a set point and run the machine around the work area of table for a while and return to the set point, how close do you come back to the set point?

_________________
Jim Watts
http://jameswattsguitars.com

What's the HP on your spindle? I've only got 2kW.

I agree with you on the ebony. I'm typically profiling ebony at a relatively shallow DOC but at about 300ipm.

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

Jim- This is an open loop system. We're still early in the testing and tuning process, so not enough information is available to describe repeatability yet. Given the initial precision along with home and limit systems, repeatability is anticipated to be excellent. I'm sure updates will be common as more tests proceed and data becomes available. Realistically, resolution and repeatability within 4 to 5 mils would be more than adequate to ensure optimal fit for wood glue performance. I'm lining up a series of tests for Monday that should answer a lot of questions.

This machine is only finished as a first iteration. As resources become available the design will evolve. While we are first and foremost a guitar manufacturing company, along with my co-founder excuses to acquire awesome shop toys is an unavoidable priority. If there is sufficient interest the designs can be made available. In terms of bang for the buck, extruded aluminum is hard to beat.

It's a 10 HP spindle, so 7kW, but the spindle is under almost no load cutting wood. The only place I've found where I can get the spindle load up to full is doing a HSM path cutting aluminum at a 0.750" depth. My new machine will likely be in the 2-3kW range, since 95% of the time I spend cutting is finish work and a lighter spindle with maybe more RPM would work out better for that.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Quick update on how the machine is running. Still tweaking acceleration curves a bit, but it is cutting at 300 ipm. Repeatability is inside of 1 mil and testing precision at returning to the starting point after running test patterns for an hour or so showed that it was not a measurable difference. I'm working on more of a torture test to find any drift.

For the time being, my time is devoted to finding some thick scraps of corian to make a t-slot surface from. Also, ultra low runout collets are on their way and with any luck will show up by Friday. Very soon we will cut our first test neck joint and rough out a body blank.

excellent Hugh, way to go!

_________________
Jim Watts
http://jameswattsguitars.com

Mach3 really is a bit of a chore to work with. We were fabricating a dust collection system the other day, shortly after installing new steppers using G-code my business partner generated. The design includes some relatively small internal square cuts that came out looking very rounded on two of the corner each time. After reviewing acceleration the first problem was that after the stepper change for some reason a different acceleration value was running on the Y axis than on the X, which explained part of the problem, not to mention we found the y-axis anti-backlash was slipping. I recommended splitting the operation into a couple of cutting paths for testing purposes... Dropped the cutting rate way down for the square cutouts and ran the rest at normal speed. It turned out beautifully. In fact, it turned out that there were nice friction fits on the interlocking parts. I'll post a picture soon of how it looked after polishing.

My business partner is a truly brilliant guy, but has no formal training in CAD, materials science, or CNC so I'm bringing him up to speed. I can't wait to teach him inlaying. Back in high school he made origami under a dissection microscope using forceps and has a picture of a crane he folded that fit on top of a dime. We both enjoy automation and hand tool skills.

Anyways, I get that picture uploaded soon. We have three spare steppers sitting around and have decided to build a arduino based CNC pickup winder that can be self contained or wired and run off a standard controller. It should be a lot of fun, and I can finally start running exhaustive tests on pickup response and performance.

Hi Hugh, A quick question. what settings are you using in the general config in the CV control section? If Plasma mode is checked it will round the corners. Also not having stop on CV angles checked will cause this as well. I usually set the CV angles on/checked and a value of 89*. I put a CV distance tolerance of a min 50 units when using micro mills. Most of the time I set the Tolerance to 100 to 150 units for 0.125+ endmills.


Just something to look at from someone that has had the same problems years ago.

Mike

_________________
I'm outside looking in, just farther from the window than most.

А +1 to this advice. The worse your acceleration, the worse the performance is going to be. I've settled on an exact stop value of 85 degrees as when I'm moving fast (I'll cut up to 400ipm) I get rounding even on oblique corners. Raising the CV look ahead to 150 gives a much smother running machine but also too much rounding for my satisfaction. For parts where I want sharp 3D corners I'm keeping it at 20. The machine runs jerkily (is that a word?) but the parts look really nice.

I haven't played with the CV distance tolerance yet though as I still don't understand what it means.

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com