Any amateur builders using their own CNC?

[Jeff Siddall]

Looks like a good start Jeff. Odd how it seems to Z dive a bit at a regular spacing in some of the pics.

Do you mean the low spots at the ends of the cuts? If so that is CAM related, not CNC. Here is a pic after the rough passes:

You might guess those ends are because of some problem with the machine but they were definitely not. You can see the effect of chatter in some of the edges that are a bit wavy where they should be sharp. I am hoping the new parts will increase the rigidity enough to get rid of that entirely.

To get a small sample of the neck in CAD I just sliced off the rest of the neck with a cut so there is no change in "Z" around the ends, but it still seems like PyCAM is deciding to "rounding off" the edges. I have no idea why, but if I contour the ends I am pretty sure I could get rid of that. I actually did that on another test with a contour around the heel and it sorta worked, although the loft caused some other strangeness in the way PyCAM rendered the toolpath. Anyway, not a showstopper.

One possible improvement to the results would be to run a few finishing passes with the stepover reduced to half on the last pass.

The slop in the rails will be most prominent in the roughing passes and the finishing passes won't be pushing back so hard and allow you to get a bit tighter tolerances.

Yes, that would probably work but again, with increased spindle rigidity, I can hopefully do it in one pass.

Are you using a 1/4" ballend? A 1/2" will smooth out those cusps a lot especially if you use the same feed rate you do for a 1/4".

Yes, technically a 13/64" (5.1587 mm) actually, since those happened to be on sale. My collet is an ER11 and those only go up to 5/32 so no go on the 1/2". Still, at 80% overlap those grooves are only about 0.2 mm (0.008") deep so not much sanding there!

And yet another point, if your machine has that much slop make sure your fret slotting passes all run the same direction to try and keep the errors all pointing in the same direction and not compounding each other.

Since my frets on this one are going to be fanned that won't really be possible. Still, I don't think that will matter. At 0.6 mm (0.023") that bit will be going seriously slow -- somewhere around 40 MMPM (1.7 IPM). Again, rigidity should solve that.

It isn't obvious from most of my pics but when the bit is cutting smoothly that thing is actually surprisingly precise. Have a look at some of the surfacing experiments I ran in the bottom right corner of the block above. In areas where there was no chatter it is really tough to even detect the tool path.

These days I'm trying to get closer to finished on my projects. I've started by leaving a good deal of hand work at the end to make sure I don't go too far on the CNC but now I'd rather spend 20 more minutes on the CNC to save 1/2 an hour of sanding.

Yeah, I can always make the finishing passes even narrower (0.5 mm) and try to find a 5/32 ball end if I wanted to save even more sanding. Those changes could get the grooves down to a couple of thousandths.

 

[james condino]

In no way did I want to be critical of the original poster (postee?)...I fully understand the effort put into the project and celebrate the sub $200 cnc build more than I can express. I remember well my adventure into mechanical engineering graduate school back in the day so I could get the credential needed for an "entry level" 1/2 million dollar cnc 20+ years ago. I also remember trying to price out a basic F5 mandolin top and back plate as a production prototype while working for a boutique manufacturer from our cnc subcontractor. We spend over $50k year with this supplier and knew that they already had the CAD worked out and they quoted us $18,000 for one set of mandolin plates while we were projecting 500 sets a year as an addition to our current orders!

Now I regularly use guys like Andy Birko of Birkonium design, RC William of KC stings and RCCNC (a member here who hags out on the double bass side) and a couple of other folks. I also live about three hours from the Shopbot factory in one direction and Cobra in the other direction, so I recognize there are a LOT of locals who have cnc machines. The 100kgarages forum is loaded with people who have nice machines and are looking for other people's projects to help pay for their fancy machine. If you offer a $500 or $1000 scholarship to the local technical college for a student who can help you integrate a curriculum project with a real world practical application, you'll get 50 people in a few days interested in helping you design cool CAD guitars vs. the engineering depatment's boringsaf@#$ widget semester projects!

My other secret weapon: whenever I see someone else online who is doing cnc work that is appealing to me- I write them a simple email or make a phone call. 1) Suck up to them first! 2) Offer up something of help to them and their business- a review, a write up somewhere, et cetera (I've written probably 75 articles for other builders). If you can express yourself here for free, you have the skills to write elsewhere. Finally, ask for help with your project and offer what you would think is a reasonable price for their "expertise" for a reasonable number of pieces- a two or three to test the waters. You can ask for more as your relationship develops. Most important- BE NICE...flatter them... 95% of the time, they are happy that I'm willing to pay them for either help with my cnc work or for $100 or so they are more than happy to sell / share their design and tooling work for me or one of my cnc friends to use on our own projects as long as we are reasonable about it and keep it in house and do not mouth off about it all over the internet. Very few people ask for intellectual property these days, so probably 50% of the people I deal with think its cool to make a few hundred $$$ for little effort. Just as I have learned from each of the 2000+ students I've taught in the last four decades, they get excited to see someone else branch out with their designs in way that they may not have thought about or are afraid to try.

Building instruments of any type is an extremely satisfying and challenging way to spend the limited time here that we all have. Regardless if you use a $100,000 cnc machine or a stone age bronze tool, the key is to know what you want to say and then with directed intent let the tool help you bring out the beautiful instrument that is already waiting for you hidden in the piece of wood...

 

[TrustRod]

Do you mean the low spots at the ends of the cuts?

No. In the first 2 pics in this post Any amateur builders using their own CNC? it looks like every 4th or so pass went a bit deeper. Most noticeable in the triangular section on the second pic.

 

[turindev]

...A radius and fret slotting should take less than an hour including setup and clean up afterwards.

Whoa. This caught my eye.. if you don't mind sharing @TrustRod what feed/speed do you use for your fret slotting toolpathing? I cap out at 10 ipm on my 0.023" endmill and fretting takes a little over 2 hours.

(Sorry to hijack the thread)

 

[Jeff Siddall]

No. In the first 2 pics in this post Any amateur builders using their own CNC? it looks like every 4th or so pass went a bit deeper. Most noticeable in the triangular section on the second pic.

OIC. No, those are rough passes using a ball end with very little overlap (10% IIRC). For whatever reason those were quite deep. It could be toolpath or I may have changed Z at some point between the roughing and finishing files. I will have to run that test again and make sure I know why it happened and take steps to correct it if it was in fact a CAM issue.

Whoa. This caught my eye.. if you don't mind sharing @TrustRod what feed/speed do you use for your fret slotting toolpathing? I cap out at 10 ipm on my 0.023" endmill and fretting takes a little over 2 hours.

(Sorry to hijack the thread)

I mean, if you consider that a bass fretboard might have somewhere around 50 inches of fret slots, if you were going 10 IPM it should only take 5 or 6 minutes per pass.

Now, whether it is smart to take one deep cut (say 2 mm or 0.080") or several shallower passes is something I would like to hear some comments on.

I have read arguments that using deep cuts and going slower is actually easier on the tool since the lateral force is distributed across the whole flute instead of only at the tip. Also, the wear is spread evenly across the length of the flute instead of concentrated only at the end.

Finding a feed speed becomes tricky because the DOC in a single pass is about 3x the tool diameter, and I haven't found chip load factors for deep cuts like that. Info I have found for small mills puts chip load around 0.005 MMPT (0.0002 IPT) for a 0.3 mm (50% of diameter) DOC. But for a much deeper cut presumably you need to go slower.

 

[Jisch]

I don't much about this stuff, but I had a friend cut blind slots and edges on a Rocklite board. Once he got it running it took about an hour to run. It took considerable time to get everything set up though. I know he did two passes on each fret slot, if I recall correctly that was for width, not depth, I believe his passes were all full depth.

All of that is out of my memory, so it's practically useless information.

 

[turindev]

OIC. No, those are rough passes using a ball end with very little overlap (10% IIRC). For whatever reason those were quite deep. It could be toolpath or I may have changed Z at some point between the roughing and finishing files. I will have to run that test again and make sure I know why it happened and take steps to correct it if it was in fact a CAM issue.



I mean, if you consider that a bass fretboard might have somewhere around 50 inches of fret slots, if you were going 10 IPM it should only take 5 or 6 minutes per pass.

Now, whether it is smart to take one deep cut (say 2 mm or 0.080") or several shallower passes is something I would like to hear some comments on.

I have read arguments that using deep cuts and going slower is actually easier on the tool since the lateral force is distributed across the whole flute instead of only at the tip. Also, the wear is spread evenly across the length of the flute instead of concentrated only at the end.

Finding a feed speed becomes tricky because the DOC in a single pass is about 3x the tool diameter, and I haven't found chip load factors for deep cuts like that. Info I have found for small mills puts chip load around 0.005 MMPT (0.0002 IPT) for a 0.3 mm (50% of diameter) DOC. But for a much deeper cut presumably you need to go slower.

Interesting points here. Yeah Im at 10 ipm making cuts of something like .012 (1/2 tool diameter), so it takes a number of passes. I was just reading on a guitar forum about someone who was running it at 40 ipm (.01" per pass) that might be worth trying, that would cut my time in 1/4. No idea what the CPT value should be though, so im not sure of spindle speed for proper chip evacuation. I was running 12k but the person running 40 ipm was running 24k.

 

[Arie X]

select a ball endmill for your back of neck finish profiling. or if for some reason you must use a flat tool, then bias you cuts at 45 degrees. this will invoke a ton of 3 axis code that it sounds like you machine or software or both might not handle, but you could experiment. but from the gouging going on i can see that your software is not capable of delivering clean toolpaths unless there are some settings that still have to be tweaked? if you look at the code your should see straight lines like (gross example)

G01 F10. X-2. Y0. Z0.
X2. Y-.025 Z-.025
X-2. Y-.05 Z-.05
etc..

notice in this example x is a straight line from +2. to -2. your post may be outputting splines or small steps breaking that linear move into tiny x movements and possibly allowing the machine to gouge. unless your rigidity of part, tooling and fixture is really bad.

what does you code look like?

 

[Jeff Siddall]

select a ball endmill for your back of neck finish profiling. or if for some reason you must use a flat tool, then bias you cuts at 45 degrees. this will invoke a ton of 3 axis code that it sounds like you machine or software or both might not handle, but you could experiment. but from the gouging going on i can see that your software is not capable of delivering clean toolpaths unless there are some settings that still have to be tweaked? if you look at the code your should see straight lines like (gross example)

G01 F10. X-2. Y0. Z0.
X2. Y-.025 Z-.025
X-2. Y-.05 Z-.05
etc..

notice in this example x is a straight line from +2. to -2. your post may be outputting splines or small steps breaking that linear move into tiny x movements and possibly allowing the machine to gouge. unless your rigidity of part, tooling and fixture is really bad.

what does you code look like?

Yeah, plan was to use ball end and then I switched to flat for testing. I can readily switch back.

For whatever reason the toolpath was very much not straight. The circular marks at the end of the cuts were made by separate plunge operations. Again, it seems to be because of something different it is doing where the cut area ends. Maybe if I chose to keep the toolpath inside the material it wouldn't do that. Anyway, experimentation will tell. I may switch CAM apps, especially for the 2.5D slicing stuff which is more readily available. The list of 3D free CAM software for Linux is very small. If I need to I can buy a new hard drive and resurrect my old Windoze laptop and maybe try Fusion360.

 

[TrustRod]

I use and love Fusion360. They are clamping down on what's free to 3 years for a student and 1 year for a biz making less than $100k a year. I just got an email to call them to verify my eligibility for free status.

 

[TrustRod]

Whoa. This caught my eye.. if you don't mind sharing @TrustRod what feed/speed do you use for your fret slotting toolpathing? I cap out at 10 ipm on my 0.023" endmill and fretting takes a little over 2 hours.

Looking quickly in the mess of files from past projects I see

29 frets following radius/32” scale 5 string bass
21,000 RPM (max speed of router)
40 - 50 IPM
.02”stepdown
Less than 3 minutes a pass, 3 or 4 passes.
I have lead ins around 8 IPM

These should not be taken as recommendations as I don't really know what I'm doing but it's worked so far in Ebony, Purlepleheart, Bubinga.

I'm also moving up to a .027 mill as I glue frets and like an easier press slot for less chance of back bow.

Milling the radius takes about 15-20 minutes.

 

[Jeff Siddall]

Update: before disassembling the CNC to start work on the Y-axis expansion I got LaserGRBL installed on a Windoze VM and then ran through some sample text for the back of a headstock. I exported the G-code and ran it through the bCNC. It took an evening of fiddling to get laser mode setup, and the speed calibrated for the laser power, but the results at the end were unbelievably good. I mean, this CNC has inexpensive parts all around, and the 1 W laser is one of the smallest you can get, but the quality of the engraving I managed on a scrap of walnut I had lying around was far beyond my expectations. I might not have believed it if I hadn't seen it with my own eyes (wearing the funky green safety glasses of course)!

The CNC has 1.8 degree steppers and a 4 mm lead screw so it has a step size of 0.02 mm or 0.00079". But that is all theoretical and the claimed precision something like 0.1 mm. So I figured my small text, with some adjacent lines nearly that close together, would be a jagged unreadable mess. Instead, it looks nearly perfect with minimal signs of overshoot, slop or steps.

For scale, the metal ruler in the picture below has 1 mm markings (about 0.040"). The parallel lines on the forward slash, for example, are something like 0.2 mm apart, and the linearity error is well below 0.1 mm, perhaps into sub-thousandths of an inch!

 

[Beej]

Nice, that is some fine detail

 

[GuitarPaul]

Wow...sort of interested in that little machine. I've been sort of obsessed with CNC routers lately, but will probably be moving and would love to get something small that I can figure out what I don't know on.

What extension kit did you use? I see a kit that extends the bed to 14.6" which isn't enough for a guitar board. Is there something longer?

 

[Jeff Siddall]

I ended up designing my own as I couldn't find anything off-the-shelf to get up to my goal of 700 mm (27.5") of travel. I basically got the longest lead screw 800 mm) and linear rails (1200 mm) I could find at a reasonable price. Getting any bigger meant belt drive and a whole different design. Unfortunately I am still waiting for the linear rails so haven't finished this yet. Once I get those and get it assembled I will report back.

Having said that, if I knew then what I know now I probably would have built my machine from scratch. The only things I ended up using from the original CNC were the steppers, controller and a few other misc parts. The real value for me in starting with an off-the-shelf machine was the inexpensive CNC education. For woodworking the original CNC 3018s are pretty useless.

 

[GuitarPaul]

huh. Thank you! Useless because they are not rigid enough, not powerful enough, etc etc?

Just looking at the Highline plans now...seems like a relatively cheap way to get involved.

 

[Jeff Siddall]

Yeah, the 3018 is just not rigid enough. You need to take a fraction of a mm per pass to prevent severe chatter.

The Highline machine looks nice, but I would need a bigger budget and bigger garage than I have!

 

[GuitarPaul]

Gotcha. I was watching his video about it...he says it cost about $1300. All the parts that look like metal are actually plywood. It's interesting at least.

The problem with improving cheap Chinese stuff (or perhaps with things made of plywood that probably should be metal) is I don't want to blow $800 on fiddling with something instead of paying $1500 for something that works. But paying like $150 to figure something out is worth it.

The Shapeoko looks really good too. I'd like to get something that's at least big enough to do necks and bodies.

My hope would be that the CNC router would be replacing a lot of other tools and be the centerpiece of making. (warnings made by a poster above are absolutely taken to heart btw). I just did 6 telecaster necks and 1 strat neck by hand, they all came out decently but it seems like 90% of the work could be done on the CNC and it could do it better than I could, and then I could really put my energy into the design, fit and finish instead of constantly being stressed about getting all this manual work done.

 

[Jeff Siddall]

Yeah, if you are looking at producing anything in quantities then a $1500 investment could make sense. My general advice would be to buy an off-the-shelf model unless you are more interested in learning how to make a CNC than actually having one. Making anything from scratch for the first time is a ton of work. Rewarding, yes, but it will take more time than you would expect.

I make things in quantities of 1 so can't justify the space or cost for a commercial CNC. Rather than the centerpiece for my work, a CNC is a precision tool to do a few things that would be hard for me to do (or do precisely) by hand: engraving, blind fret slots, compound radius, etc. FWIW, all the moving parts on my upgrade are metal, including the entire Z-axis. Only the base/table and gantry (which is fixed) are plywood. My costs were $300 US including everything (tax, shipping etc. on all parts, plywood, little fasteners, wiring etc. that some people ignore.)

Good luck!

 

[Bruce Johnson]

I'll mention this here on this thread:

I know a guy who is looking to sell a complete turn-key CNC guitar/bass building operation. A large serious CNC machine, all the tooling, a wide belt sander, resaw bandsaw, and many other smaller machines. A beautiful complete shop. Some software/machine training can be included, if wanted. Everything you need for a one or two person business turning out 100-1000 instruments per year.

The shop is currently in operation. Not a big hurry, but the owner wants to retire out sometime soon. He'd like to stay anonymous for now, but he asked me to put out some feelers to see if anyone is interested. I have no business or financial connection to him. The price isn't set, but he's thinking around $50K.

This is an opportunity for someone to step up into a small guitar building business. Ready to run.

If you are seriously interested, PM me and I'll put you in touch with him. I'm not going to discuss it any more here.