CNC Making Discussion

[Don Noon]

4 hours ago, scordatura said:

Could you show what you are using to create the vacuum?  How did you create  the seal?

I have a 1/6 HP Gast rotary vane vacuum pump which works well enough most of the time, but some maple is very leaky, and then I use my 1/2 HP Welch Duo-seal industrial pump.  Not something you'd normally have lying around, but I got it for my wood torrefying chamber work.  I suppose you could glue-seal the maple if needed.

The seal is a 1/4" diameter silicone extrusion (eBay), with a slot milled in the fixture a fraction of a mm shallower than the silicone.  The ends have to be cut square, and sealed with silicone.

[NCLuthierWyatt]

On 1/7/2020 at 9:12 PM, Marty Kasprzyk said:

Accurately duplicating the geometry of great instruments is not very helpful for getting great results unless their wood properties are also closely duplicated.

I value multiple duplicity.

I agree, another thing we're ignoring by replicating arching is the warping that occurs through age as a result of string tension, humidity changes, and the sound post slowly altering the plates. If I recall correctly North Bennet Street School's violin making program is currently doing a study on this very issue. What's the point in replicating arching on 300 year old instruments if the the arching you're replicating doesn't even match the arching of the instrument of when it was new. If you're trying to replicate the arching of a 300 year old strad for aesthetics and to be able to say it's a 1:1 bench copy I can understand, but if you're aiming for matching sound characteristics, we're probably wasting time and lots and lots of money

[scordatura]

I can't speak for all of the makers that are pursuing CNC efforts, but reproducing arching distortions is in my opinion absurd and not worth discussing. I think that even though it has been stated before that CNC is used to do somewhat rough work or making of moulds, etc., many people responding to this thread are missing the point and expressing their opinions. That is all well and good but it is adding unnecessary opinion based content to this thread. Please use these threads to promote or criticize CNC in violin making:

 

 

[Steven P]

I am new to this forum/discussion. But I have made violin parts and many other things by CNC for nearly a decade. As stated earlier, CNC is a labor saving device and can't effectively be used to finalize a violin plate unless the entire instrument is machine produced, in which case you will get what you deserve. We would not use dull tools or those with inferior steel (for example) just because some ancients did. Modern tech like CNC makes the work easier and more approachable by many. My shop is presently in storage due to a move so I have been focused on design for CNC. This link is to stl files for the top and back plates I recently constructed and cleaned up: My STL files . The original design was in Rhino using classical drawings and contours available everywhere. They may have also been smoothed in Aspire. As for arching, there is much nonsense on the web about it, especially around curtate cycloids. You can read the article at the link for my take on arching, or go to the source here . In brief, arching comes from contours. My stl files are nowhere near final - they are meant to save labor and then be finished by hand, so perhaps they should have a generous Z cutting margin. I have made tool paths for these plates in Aspire, and there can be interesting challenges in constructing the paths and executing them. I have not cut these exact versions so I don't know the thicknesses that will result. My normal workflow would prevent this from being a problem, but yours may not. Care is needed!

[Dennis J]

Looks like it could do at least Mach 5.

[scordatura]

20 hours ago, Steven P said:

I am new to this forum/discussion. But I have made violin parts and many other things by CNC for nearly a decade. As stated earlier, CNC is a labor saving device and can't effectively be used to finalize a violin plate unless the entire instrument is machine produced, in which case you will get what you deserve. We would not use dull tools or those with inferior steel (for example) just because some ancients did. Modern tech like CNC makes the work easier and more approachable by many. My shop is presently in storage due to a move so I have been focused on design for CNC. This link is to stl files for the top and back plates I recently constructed and cleaned up: My STL files . The original design was in Rhino using classical drawings and contours available everywhere. They may have also been smoothed in Aspire. As for arching, there is much nonsense on the web about it, especially around curtate cycloids. You can read the article at the link for my take on arching, or go to the source here . In brief, arching comes from contours. My stl files are nowhere near final - they are meant to save labor and then be finished by hand, so perhaps they should have a generous Z cutting margin. I have made tool paths for these plates in Aspire, and there can be interesting challenges in constructing the paths and executing them. I have not cut these exact versions so I don't know the thicknesses that will result. My normal workflow would prevent this from being a problem, but yours may not. Care is needed!

From a quick glance it looks like your design is Strad inspired. Is it a particular instrument (Messiah) P or PG mould? I have found it pays to be picky with your pattern source files rather going down the wrong path. Strad posters?

Also interested why you chose Aspire for smoothing and tool paths.

[Steven P]

It is a strad design, probably from sacconi drawings; not strad posters. I chose Aspire because it's something I use extensively and the cost is modest compared to many other options. Only part of my CNC work has been violins, and I am comfortable with the Aspire interface by now. Even so, setting up two-sided jobs in Aspire is not always as easy at they make it look. Some 3D work I did a few years back was easier using Cut3D, which no longer appears to be maintained by Vectric. I am mathematically inclined and use Mathematica for much of my work, but it is expensive and is a language unto itself. It does communicate now directly with Rhino/Grasshopper which is a plus. Mathematica objects can only get into Aspire via stl or dxf.

[Don Noon]

Here's what can happen if you have speed, feed, and depth of cut too high:

Actually, this is the beginning of a bit of rework.

1) X-Y axes were out of alignment by ~.3 degrees, which is enough to make instrument corners misalign to an unacceptable degree, and any tooling bolt patterns could not be flipped over and still match up.  Exceptionally annoying for the type of precision I want in my instruments and tooling.

2) Fixing some stupid design flaws.  Overall, I think it's a great machine, especially for the price, but there's something that HAS to be fixed:  the linear bearings are held in place by single, tiny setscrews with no locking features.  With all of the vibration that goes on, those setscrews inevitably come loose.  Since the bearings are a slip-fit in the aluminum housing, that allows a bit of slop, or even worse, the linear bearings can come completely out of the housings.  I will be re-machining some of the parts for bigger/additional setscrews, and adding thread locking compound.

I had to fix the Z axis earlier, but waited until I was ready to do the machine alignment to tear everything apart to get to the other linear bearings.

If anyone out there wants this reworked machine, I will probably be selling in in a few months to upgrade to the latest thing... and then I'll have to do this kind of work again, BEFORE I start cutting out parts.

On 1/13/2020 at 10:12 AM, Steven P said:

Modern tech like CNC makes the work easier and more approachable by many.

I would replace "many" by "very few".  Learning the solid modelling software, CAM software, Mach3 machine driving software, Gcode, and the care and feeding of the machine itself... these are fairly daunting barriers to most, and perhaps especially so to those makers who come from the arts and tradition side of things.

 

While I'm here... there's a separate thread for discussing the software for CAD/CAM software:

 

[Michael_Molnar]

I agree with Don’s remark about “very few”.

[Marty Kasprzyk]

There seems to have been at least four different ways of duplicating wood plate arch shapes that have been used:

1.  Hand carving using arching templets as guides.

2.  Duplicating router machines using patterns.

3.  Pressing thin plates into molds.

4. CNC machining.

5.  In the future we might see 3D printing if suitable materials/designs can be found

 

 

Each of these has some disadvantages so I make my plates flat.

[uncle duke]

1 hour ago, Marty Kasprzyk said:

Each of these has some disadvantages so I make my plates flat.

I do have visions of Don sitting beachside under an umbrella reading sports illustrated while the machine is at home doing all of the work.  We'll have to wait and see.

I use the drill press method for making my plates - just drill down to certain depths.  The downside is I'm sure I'm the only one doing this in the continental U.S., possibly worldwide - it's a lonely feeling thinking that.

[scordatura]

2 hours ago, Marty Kasprzyk said:

There seems to have been at least four different ways of duplicating wood plate arch shapes that have been used:

1.  Hand carving using arching templets as guides.

2.  Duplicating router machines using patterns.

3.  Pressing thin plates into molds.

4. CNC machining.

5.  In the future we might see 3D printing if suitable materials/designs can be found

 

 

Each of these has some disadvantages so I make my plates flat.

I would also add carbon fiber which would be laying material in a mold. I have played the Luis and Clark instruments. They are good but for the discerning player have an unusual tonal characteristic.

Being a formula 1 fan, I am amazed what they and aerospace are doing with these materials. I also have a few carbon fiber bows that are pretty good. My favorite being the Rolland Spiccato. 

[Michael_Molnar]

I anticipate that eventually there will a simplified version of these high end CAD packages targeting luthiers, craftsmen, and artists who do not need n-th degree functionality and a myriad of tool choices. I must admit that I glaze over when I look at the Rhino tool-selection screens.  And don’t get me going about Mach3, the CAM CNC controller. 

Then there is the mechanic side of CNC. Look at Don’s repair job. I too am adding an adjustment to align the router vertically. We need better quality, affordable machines that do not need adjustments when they are unpacked and plugged in. Many of the desktop CNC’s are meant for making signs and small furniture.

[scordatura]

Spindle speeds during CAM seem to have two schools of thought. The fastest for the cleanest cut or slowest for heat and wear. I realize that the feed rate is the other variable and the two are interrelated. I find the calculators that are available (e.g.http://www.cutter-shop.com/information/speed-and-feeds-calculator.html) are fairly aggressive in terms of feed rate...My feeling is to have a higher spindle speed (mine goes to 21K) and slowish feed rate. That being said one is supposed to take into account the chip clearing rate for each bit. One can hear when the machine is cutting if everything sounds happy. Far less calculated but seems to verify if you are on the right direction. Thoughts?

[Don Noon]

The link to the calculator didn't work... but I have this chart:

Yeah, it's in inches... and I try to keep metric these days when I can.

But say you have a 6mm 2-flute cutter running at 21000 rpm.  The suggested chip load (per tooth) from the chart being ~.25mm calculates out to 10,500 mm/minute feed rate, which is more than 2X the maximum travel rate of my machine.  I commonly use a 1/2" 3-flute cutter at 15000 rpm, which calculates to an insane 22869 mm/min.

The cutter itself might be happy with these big bites, but that translates to big loads on the machine.  Slowing the RPM down to get those big bites means that the frequency of the cyclical loads on the machine get lower, and the mass/inertia of the spindle become less of a factor ... meaning greater deflection and vibration.  

I generally try to run at high spindle speeds, and then as high of a feed rate as I can before the forces on the machine (including the flexing during starts and stops) get ugly.  I suspect that more rigid machines would be able to handle the higher feeds and forces before complaining.  There is also likely resonant frequencies of the machine structure, and it would not be good to generate loads at those frequencies.  I've been meaning to check that out, but haven't yet.

[scordatura]

Interesting article on chip load:

http://talk.dallasmakerspace.org/t/understanding-chipload-and-the-cnc-router/7893

[Don Noon]

1 hour ago, scordatura said:

Interesting article on chip load:

http://talk.dallasmakerspace.org/t/understanding-chipload-and-the-cnc-router/7893

The chiploads in the linked article are essentially the same as the chart I posted, which again gives recommended feed rates well above the capability of anything other than full-on industrial CNC machines.  

Us common folks have to use lower feed rates.  The fastest I have cut is 2500 mm/min (~100 in/min) out of the flat-out maximum travel speed of 4000 mm/min (157 IPM), and even that is less than half of the recommended feed rate in the article's example.

[scordatura]

12 minutes ago, Don Noon said:

The chiploads in the linked article are essentially the same as the chart I posted, which again gives recommended feed rates well above the capability of anything other than full-on industrial CNC machines.  

Us common folks have to use lower feed rates.  The fastest I have cut is 2500 mm/min (~100 in/min) out of the flat-out maximum travel speed of 4000 mm/min (157 IPM), and even that is less than half of the recommended feed rate in the article's example.

Yes the mega quick feed rates seem to be for commercial production setups that are cranking through milling. No doubt pro level machines that are beyond the reach for most of us. This why when I went from the guestimation formula to a more "evolved" calculation I thought the feed rates are just nuts. As someone I read put put it, when someone is hand routing, they are going by the sound and feel of the process to guide their feed rate not a calculation.

[Michael_Molnar]

To use these theoretical feed rates you need a robust solid CNC that does not flex under these load conditions. My CNC “Shark” has a Delron (?) frame that will bend under these feed rates.   My feed rates are only 500-800 mm/min with a 1/4” end mill for most work. The router runs at 18k rpm. As with a hand router I listen to the sounds.

[Don Noon]

22 hours ago, Don Noon said:

There is also likely resonant frequencies of the machine structure, and it would not be good to generate loads at those frequencies.  I've been meaning to check that out, but haven't yet.

I tried with my usual hammer tap method (but much bigger hammer), but failed to measure anything that I would call obvious.  As a second estimate, I measured the deflection at the cutter when a load was applied with a fish scale.  It was fairly significant.  Guestimating the amount of mass involved puts the calculated first resonant frequency well below 100 Hz.  With a 2-flute cutter, somewhere under 3000 rpm is where you'd excite the 1st resonance.

So we're working the machines well into their flexy zones... meaning higher rpm and lower cutting forces (lower feed rates) will result in less deflection.

It also explains why  milling machines are essentially solid blocks of cast iron.

[David Burgess]

On 1/26/2020 at 12:51 PM, Marty Kasprzyk said:

There seems to have been at least four different ways of duplicating wood plate arch shapes that have been used:

1.  Hand carving using arching templets as guides.

2.  Duplicating router machines using patterns.

3.  Pressing thin plates into molds.

4. CNC machining.

5.  In the future we might see 3D printing if suitable materials/designs can be found

6. Hand carving using oblique light to obtain shapes stored in ones memory.

21 hours ago, scordatura said:

Yes the mega quick feed rates seem to be for commercial production setups that are cranking through milling. No doubt pro level machines that are beyond the reach for most of us. This why when I went from the guestimation formula to a more "evolved" calculation I thought the feed rates are just nuts. As someone I read put put it, when someone is hand routing, they are going by the sound and feel of the process to guide their feed rate not a calculation.

Back when I had some involvement in CNC machining (mostly on metal), the procedure was to take a "spring cut" (the "spring" allowance involved the deflection of both the machine and the cutter), and then take a final small cleanup cut with much reduced loads.

[Johnmasters]

On 1/5/2020 at 5:50 PM, NCLuthierWyatt said:

As an alternative to using CNC machinery have any of you seen the wood router offered by Gemini? Looks like a great option for those wishing to avoid the high initial investment of laser scanning along with having a lot less to go wrong. I work with lots of laser scanning and high end 3D printing in a dental laboratory along with some experience with CNC machinery in producing custom components for cars. Feel free to pick my brain if anyone needs help with figuring out how to export and edit STL files. 

Here's a demo of it. Not too bad for $3200. 

 

You can do it a lot cheaper than that.  Thompson makes precision ground rods and linear bearings.  Also supports for the ends of the rods.  Look up their stuff.   I had a violin/viola one with 1" rods,  three of them  You can make the router holder with wood ir you wish.  I had aluminum plate.    I also had two aluminum plates in 3/4"  jig-plate.  They were each held to the table with three screws each..   shim to a close parallel,  then a thin rubber washer could be compressed for fine tuning.   (The two must be parallel and close to the same height.)  I would not trust just a sheet of plywood.   Eventually the parts were canabalized to provide some parts for the new CNC.

And the linear bearings work better than a wheel rolling on a plated tube.  For cello,  I recommend 1.25" diameter components.  They are not cheap,  but still a lot less than $3200 plus it is fun to build.

 

[Michael_Molnar]

The “problem” with a Gemini is that you need a pattern to replicate. And you cannot easily change designs.  But if you have a winning design, this will knock them out for high production, if that is your goal.

[Don Noon]

You still have to stand there and move the router around, whereas with CNC you can do something else (like get out of the noisy shop!)  For $3k, you can get a decent CNC machine.

One of the nicest things about a CNC is the ability to make precision templates and tooling from a design; the duplicator can't do that.

[BenPearce]

Those look like pretty aggressive chip loads to me, esp if you are shooting for a nice surface.  I have tried numerous defaults within Fusion and find they almost always run better at half chip-load while still cutting nicely.  Some of the aggressive Feeds and Speeds can be great for hogging, but no good for "finishing" passes.  As mentioned by others, it can be much more informative tweaking both the feed rate and spindle speed while running test cuts (or even the real thing at a higher Z), as it accounts for machine resonances and other factors outside of the chart.