Advanced Knife Machining WIP, Pointy Fighter

[Eric B43]

Wow, subscribed.

 

[stereo.pete]

I just saw this and will be using the clay method going down the road.

 

[Nathan the Machinist]

Sorry, last weekend I had to choose between working on this WIP or going to the zoo with the kids and the zoo trip won...

Okay, to continue where I left off. I'm on the home stretch here on the blade (the scales are next) but I'm having problems with the surface finish on the swedge grind. Rather than just show you the solution I thought it might be educational to show you my attempts at fixing this before I got it right. This might not be the most interesting part of the WIP, but anyone who has struggled with surface finish will be able to relate.

I'm trying to achieve a fine finish on a small area on the back of the blade. Should be easy. I'm using the same insert and the same speeds and feed that I used before that worked so well. The problem is some wonkyness where the insert enters and exits the cut (there is always going to be a little bit of this if you look closely enough) and some righteous chatter when I flip it to do side two due to the reduced support. The SFM I'd been using was really high. High SFM is an easy way to get a fine surface finish when using carbide and it allows me to make the cut in less time. The obvious first thing to try was reducing speeds and feeds. I'd been spinning it at 1900 RPM, I dropped it down to 1000, then 500, 250 and lastly 100 RPM. At 100 RPM the chatter was completely gone, but the corresponding feed rate was so slow (1 IPM) I think I actually started to get into some stick-slip on the machine feed that showed up as regularly spaced crisscross marks in the surface finish. Regardless, at those slow spindle speeds the BUE (built up edge) was awful.

500 RPM:

250 RPM

100 RPM

Okay, that isn't going to work. I know, lets try something a little unorthodox. Lets use a super sharp Diamond-Like-Carbon coated insert designed for fine finishing of aluminum. These inserts are sharp enough to cut you and cost over $20 a pop:

Oh wow that turned out really bad

(that is some gnarly BUE, apparently steel likes to stick to carbon, who knew...)


I tried an insert designed for brass with similar results. Being uncoated carbide run at low speeds the chip likes to stick to the cutting edge, resulting in BUE. That BUE results in a nasty finish because the actual cutting edge is a glob of steel stuck to the insert.

Okay, so everything I'm doing here is going in the wrong direction. I'm being lazy and trying to fix a problem with cheap tricks rather than do the work and address the root cause of the problem which is that my fixture is not supporting the part well enough to prevent vibration. So, getting back to machining basics I go back to the original insert (which actually is the right insert for the application) and set it at an SFM that is high enough to prevent BUE and focus on eliminating vibration.

There are generally three ways of reducing chatter. 1: Reduce SFM (not an option, any lower and I get BUE) 2: increase feed rate (not an option, gives a poor surface finish and not effective in this application anyways) and 3: improve the setup, generally by improving rigidity.

I've been avoid that last one because it means I need to do more work on my fixture. However there is one last simple trick I haven't tried yet with is a glop of bubble gum or modeling clay stuck to the part to increase its mass, change its natural frequency and damp vibration. Okay, I'm not literally going to glob bubble gum onto my part but I'll do the next best thing. This is a piece as soft-as-chewed-up-gum rubber glued into the fixture:

That is 20 durometer shore A polyurethane thermoset elastomer. Nice and soft and squishy and good at absorbing vibrations. This actually worked. I shaped it to conform to the primary grind because I don't want it to hold the blade up off the fixture. I still want the blade to sit fully hard against the fixture, but I want this rubber to press against the part with about the same amount of force as finger poking against it. Machinists have long known that sometimes all you have to do to quiet down chatter is poke your finger on the part. Using my actual finger was not an option, but that glob of shaped rubber is doing the same thing.

The first cut looks promising:

And side 2, which had been squealing like a stuck pig:

That's gonna work. It's about time...

So, this was the last machining operation on these blades. They're ready for heat treat.

The next operation will be machining micarta scales utilizing vacuum hold down then fixturing for 3D machining.

 

[Big Chris]

Great thread. I have enjoyed each of your posts.

 

[Dan Pierson]

Very interesting -- not boring at all! Also interesting to see that the swedge is carefully designed
to *almost* reach the tip.

 

[Nathan the Machinist]

I promised a technical WIP, so I'm going to go into some detail about vacuum fixturing for those of you who are interested. Most of you will probably want to skip to the next post, but to anybody interested in trying it, this is an important post.



I am machining micarta scales. There are a number of different ways of approaching this depending on the design of the scale and the material, but in my shop it usually starts with a sheet held down to a vacuum table. A lot of us have vacuum pumps for wood stabilization that might consider using this on a mill or router for part fixturing. A vacuum is capable of generating over 4,000 pounds of hold down force across a 12X24 sheet.

This is my vacuum hold down table.

It is canvas micarta.

It is 12 X 24 X 2”. It has a big aluminum “chunk” attached to the bottom of it so it can be clamped into a pair of matched vises. I used to run it off a subplate but out of vises is a lot faster to setup.

This is not how it usually looks. It usually has a bunch of gasket traces carved into it from previous jobs that have been filled in with epoxy and milled flat. It was getting a little Swiss cheesy so I went ahead and mowed the whole top of it down so previous traces wouldn’t show for this WIP.

This is how you make and use a vacuum table for parts like this:

1. You need a vacuum pump. Most manufacturers use some form of a rotary vane for this and that’s not a bad choice. However, your hold down force is a function of your surface area and the depth of the vacuum. Some good rotary vane pumps get down pretty low, but many do not. Since I frequently machine relatively small parts I use a multi stage positive displacement vacuum pump that pulls a high vacuum to maximize my holding force on smaller parts.

Commercial CNC wood routers that use vacuum for part fixturing in furniture manufacturing are typically 200-600 CFM, but they're holding multiple sheets of wood against an MDF spoil board without gasketing so it takes a lot. For this small table I'm using an old 30 CFM “Absolute 1200” pump which is an unusual pump for this but it's working. You could probably achieve good results with just 1 CFM if you use a vacuum reserve tank and good gasketing and are careful. One advantage to a larger pump is you can run two lines from a manifold at the pump and have two vacuum zones that can hold a part in more than one location in case your vacuum becomes compromised you don't throw a part.

2. You need a vacuum table. Industry frequently uses an aluminum waffle table capped with an MDF spoil board that you can pull vacuum through or add foam rubber gasketing. For the small parts we’re making we want to maximize our surface area, so a precut grid of gasket channels is often no good because they don’t ever fit the profile of the part perfectly so I like something I can mill a channel in directly like phenolic. This channel can be filled with epoxy when you’re done and the table surface gets resurfaced every time you put it in the machine. You could use aluminum, but a 2” slab of micarta weighs the same as a 1” plate of aluminum, and is stiffer (stiffness increases by thickness to the 3rd power, so despite having a lower flex mod the 2” phenolic table is actually stiffer than 1” aluminum would be, and weighs the same).

You obviously can’t use an MDF spoil board with coolant, and I don’t like cutting this kind of stuff dry. Coolant keeps the dust and the fumes down.

3. You have to plumb the table to the pump with somewhat flexible tubing that won’t collapse under a vacuum. I’m using ½” LDPE tubing and push-to-connect fittings. The tube is relatively flexible and can pivot freely in the fitting. I have air water separator filters at the actual vacuum pump to catch any random drip of coolant or trash that may make it through. Take note that these units are installed backwards from how you’d normally install them on an air system, which makes sense when you consider the direction of air flow. You want ball valves at your table so you can turn the vacuum supply on and off at the table. It is good to have the line go up to the ceiling then back down to your pump. That way any little bit of coolant that leaks into your table has to go uphill to make it to your pump. The less stuff your pump ingests the better.

4. You need to distribute the vacuum in your table and to your part. I’m running two air lines (there is a good reason for this if anybody is curious) but you only need one. I drilled a hole into the edge of the table and drilled a hole down into it from the table surface. From this you need to cut a small air channel out into your part and (usually) out into the perimeter gasket. This helps distribute the vacuum all over the part. Otherwise you can have one section of your part pull down directly over your vacuum hole and can sort of seal off with the vacuum never making it all the way across the part. A lot of CFM makes obtaining the initial seal a lot easier. You either want a 2-3 HP pump, or a reserve tank to give you the CFM you need to obtain the initial seal. After that a small pump can maintain vacuum if your gaskets are good.

5. You need to gasket around the inside profile of your part and you need to cut little o-ring grooves everywhere you have a drilled hole in your part that would violate your vacuum zone. I cut my profile gasket channel with an 1/8” cutter along a path centered an 1/8” from the edge of the part. I program it this way, but you can also cheat with your offset compensation values and get the same result. I’m using a 1/16” o-ring cord stock (60 duro buna-n), which is actually .070”, so my channels are .057 deep giving me .013” o-ring compression. Too much compression and your part won’t sit flat against the fixture, and too little and it becomes difficult to obtain the initial seal.

I glue the ends of the cord stock together with super glue and I also glue the gasketing down into the groove with a few drops of super glue in strategic spots. I like the crappy consumer grade glues you can get at Lowes for this because it has a good reliable low bond strength and poor adhesion which makes it easy to get back out of your fixture later.

In use, you turn your vacuum on (with your ball valve) then you turn the vacuum back off (at the ball valve) and the part should remain unmovable. This confirms you have a good seal. Then you open your vacuum valve again before commencing machining. This seal retention makes removing the part tricky when you're done. I just pull my push-to-connect fitting to release the vacuum or if the part has holes in it I take a rubber tipped air nozzle and press it into a hole and use air pressure to blow the part loose from the table. You could use a three way ball valve and vent to atmosphere, but I don't think it's worth the trouble.

I think that sums up the use of vacuum for part fixturing.





I kinda doubt anybody actually read much of that, but I promised to go into technical detail and to anyone trying to figure this out on their own for the first time this ^ is gold.

The next post will be 3D CNC machining complex contoured knife scales. It'll be a little less dry. I'll try to get it up tonight or tomorrow.

 

[bear5177]

Wow! Great WIP! Thank you for sharing this.

 

[Nathan the Machinist]

The first thing I do is pull a sheet down on my vacuum table.

It looks like it's just sitting there but it is held down solid.

I like to run 12 X 24 sheets when I can because it is the most efficient, but this size here 8 ½ X 11 is not uncommon.

All I’m doing in this operation is cutting the hole geometry in the scales which I’ll use to mount it to a fixture. I suppose it would be possible to machine the entire scale here in its entirety, but that requires leaving a skin at the bottom of the cut so you don’t loose vacuum. Frequently I’ll put the holes in and then profile it within .015 of the bottom then pull it off, break the pieces loose and screw them down to a fixture individually, but this time I’m leaving enough room between scales for my ball mill and I’ll mount the whole sheet to my fixture and trim the scales out there. This reduces my sheet yield a little but better lends itself to unattended machining.

I drill my holes undersized, mill them with circular interpolation for position and then ream them for size, exactly the same as I did for the holes in the tang. This degree of accuracy isn’t particularly difficult or time consuming and makes fitting later on a non-event.

The holes are kinda counterbored/countersunk. I’m mounting my scales with flat head screws and I want the heads flush with the scale and for the countersink to transition into a shallow counterbore at the head diameter. My screw heads are .313” diameter which is sort of an odd ball size for a counter sink so I’m milling the 82 deg countersink with circular interpolation with an 82 deg carbide V mill. Compared to a countersink, this has the added benefit of leaving a cleaner cut, lasting longer and the cutter is double ended and is reasonably priced. You can’t beat that. I think this is the best approach with G10 and a good approach with everything else. If you have a CNC you should try this for your countersinks.


While those are running I made the fixture for the rest of the operation

Some things to note about this fixture.

1: It has a single accurately bored hole in the center of it. I think this is the best way to accurately indicate and locate zero on a fixture of this sort.

2: The scales are set up on raised pads. This allows you to mill the sides with a ball endmill (the ball end has to go somewhere). It also makes room for chips to flush out of the cut better. When I cut the profiles I make the cut full depth with a right hand cut, left hand spiral cutter. I like this type of cutter for that because it doesn’t try to lift the part, it pushes it down solid into the fixture. But it pushes chips down and they need somewhere to go, and that is what all that clearance is for.

3: There are little half moon cutouts roughly center with the part. That makes it easier to get hold of the parts to pull them back off the fixture.

I screw the sheet down and cut the profiles.

[video=youtube_share;zOb3v7iqHuA]http://youtu.be/zOb3v7iqHuA[/video]

After all the profiles are cut there is a programmed stop so I can reach in and remove the drops. You don’t want a big chunk of debris to turn up in the wrong place and break something. Then, for the next 88 minutes that 3/8” ball endmill slowly chugs along cutting out four sets of scales.

And I get this:

These are surface milled to a smooth finish around their perimeter but have a scalloped surface finish on the top surface made by following the underlying geometry with large stepovers and a bit of an axis shift down into the part. I like this scalloped surface and variations of it because it looks cool, it feels good, it’s grippy and it isn’t too time consuming to make. A really fine stepover surface milled scale that was a relatively smooth perfect representation of the underlying model geometry would take considerably longer to cut, and it will always have some small scallops that would need to be sanded out. Why pretend we’re grinding this with a perfectly even surface finish. Let’s calibrate the fact these were milled and use those milled grooves as a design feature, they’re cool!

I evaluated the first set on a blade and felt it was really close to what I wanted. It was a relatively faithful copy of the last hand made prototype, but after handling it I wanted to smooth out the transition to the pinky and add a little meat there. It was okay and I liked the way it looked, but I feel this tweak is more comfortable. Before on the left, after on the right (it’s a subtle change)

And onward into production!

I’m making some of these with the scallop and others with a diamond quilt pattern in black and in brown micarta. I like the way the checkering looks, but the scallop probably feels a little more comfortable to my hand. You want your texture to improve grip by increasing surface area with your hand, not by trying to take a bite out of your flesh, so all of these will get smoothed out a little on a dull fine grit slack J flex belt to smooth round and burnish the peaks a little so they won't cause blisters on your hand during prolonged use.

This part of the machining operation is slow. You can’t rush it and get good results and the machining center I'm using here is pretty slow, so a batch of 80 sets will run all week. The 4 up fixture is nice because you can load it up and go focus on something else for a while as it runs.

If you want to bring out the color, a minute or two on a buffing wheel loaded with black compound cleans them up nicely without making them too shiny.

These are going to run a little while. I'm expecting blades back from Peter's heat treat any time now, so the next step will be finishing the blades. I’ll be grinding some of these lengthwise on a shaped platen. If you’ve never tried this before, it is a cool technique to obtain a nice satin finish running the length of the blade and clean perfect plunges, check it out.

 

[WilliamWoods]

Very cool!

 

[timos-]

holy...! I run one of those vacuum tables you were talking about with the mdf deal at work .. and you are right this stuff is solid Gold! My mind is a little blown right now.

 

[Big Chris]

This is incredible to read. Enjoying each post. Ready for more.

 

[Fellhoelter]

VERY cool Nathan.

Makes me want to learn how to draw in 3D so I can try this kind of stuff...

 

[Shawn Hatcher]

Makes me want to learn how to draw in 3D so I can try this kind of stuff...

Well, you have the genius part down, Brian. But what about the evil part?

 

[stereo.pete]

Very nice work Sir!

 

[jdm61]

Nate, a while back, you mentioned burnishing with a wire brush. How does that work compared to the slack belt?

 

[Fellhoelter]

Well, you have the genius part down, Brian. But what about the evil part?

Man, I am a hack in comparison to guys with real skill like this!

 

[Nathan the Machinist]

Nate, a while back, you mentioned burnishing with a wire brush. How does that work compared to the slack belt?

A wire wheel does a better job of getting down into the little nooks and crannies and eroding everything more evenly, where a slack belt just hits the peaks. Both do a good job of taking some of the bite out of the surface prior to buffing. I'm going to use a slack belt on these because I want to accentuate the appearance of the texture and grinding the tops of the quilt pattern with a belt creates contrasting little diamonds on the peaks. If I was wanting to demphasize the pattern and wanted a more mottled texture I'd use the wire wheel.

 

[Nathan the Machinist]

Man, I am a hack in comparison to guys with real skill like this!

I've seen your work man, you're no hack. :thumbup:

 

[sloth357]

Just wow... thanks for taking the time to share this with us... I couldn't resist.. I am looking forward to one of these!

 

[Huntsman Knife Co. LLC.]

Man this makes grinding feel totally obsolete. Now I want to go back to school and take some machining courses.