Introduction: Optimizing CNC Feed Rates

About: Interested mostly on cost-effective alternative power for vehicles and furniture. Ah, electronics of course!

CNC making is a long but rewarding journey. I've been doing it for a bit and started to realize that some projects take from one to few hours to machine the part.

I then started to wonder how to honor the name of "High Speed Machining". Indeed on more advanced machines like the Haas Mill have an extremely useful feature for that -- operators can apply a speed up (or down) factor to the original G-Code feeds.

Unfortunately more popular and inexpensive machines don't share the same feature and their firmware is a black box, it should be hard to implement such a feature, if even possible.

In addition to that, cutters wear out and will no longer do a good job with the same chipload as specified by the manufacturer.

This is a framework I came up with so that we can optimize feeds and speeds for the exact combination of factors we have : material, spindle speed, cutter conditions and so on.

By the end of this process you can be confident that for a given set of factors, you did the most efficient process at the cost of a little calibration time at the beginning. And if you repeat the same factors in the future, you can still benefit from the same optimization.

Step 1: Defining the Benchmark's Feed Range

Depending on the type of material you're working with there are already good references in the cutter's manufacturer documentation. For some other tool manufacturers, little documentation will be found and we will have to start more conservative.

I'll take my journey as an example. I was in a Shopbot Buddy with an Onsrud 65-380 Endmill. I was able to find very helpful documentation here :

With recommended feeds and speeds.

Getting the references and recommended values will save you a lot of time. So you can pick up where someone left off.

In some cases we will have no reference, even from the manufacturer. In this case we either have to be more conservative, or have to analyze a wider range of values, which translates to more time spent. It really pays off to do your homework here.

Step 2: How

Basically it comes down to this - get a piece of stock that preferrably is the one you're going to cut, if not something very close.

I was lucky to have enough project stock so that I could cut a piece of it for my feeds and speeds tests.

We will do this considering that we're using Autodesk's HSM solutions* or anything that has the same features as adaptive clearing. It's worth to quote why :

It is unique in that it guarantees a maximum tool load at all stages of the machining cycle, and makes it possible to cut deep and with the flank of the tool without risk of breakage.

It's also worth mentioning that conventional strategies such as pocket clearing do not guarantee that. You have been warned.

To the actual strategy, we will just make sure we use our stock the most efficient way possible, in visually identifiable toolpaths so that we can go from a lower limit to an upper limit in multiple adaptive clearing operations and observe forces and noise as the chipload goes up. Eventually we will find which feed is too much so likely one or 2 points before that one will be our sweet spot.

This needs to be done responsibly and you should have a minimal sense of what is too much for the machine/tool you're using.

* Note that Fusion 360 has a free license to enthusiasts/startups and it comes with the standard Autodesk HSM features such as Adaptive Clearing.

Step 3: Defining Stock Size and Toolpaths

It's important that the model is a dummy, smaller version of your stock size (I set it to 0.2 times my stock size and pushed it to the bottom). The reason is that our goal here is to pretty much remove stock material as much as we can, to see how it goes. Not save material.

I then made sketches of offset rectangles in order to be able to constrain the different speed toolpaths to only a section of my stock, not all of it (see pictures).

One important setting for all toolpaths, is that you should chose a reasonable flute length to use. In my case, with a 1/2" Endmill I'm using also 1/2" of flute in the cuts. Remember then to go to the heights tab and set the bottom plane as an offset of X (0.5 in for me) from the stock top.

I don't know why, but if you do actual rectangles instead of offsetting the center one, Adaptive Clearing will create toolpaths inside of it, if you offset the inner ones, the toolpaths will be from outside to the contour of the rectangle. That's what I did.

The toolpaths will have useless overlaps but it's the best I managed to do. Tried more efficient ways but didn't find it. Called it the Efficient Level of Care and moved on. Let me know if you find better ways to do it.

At the end of the day you should simulate something like the second picture.

Step 4: Cut!

If you did a good job on your CAM (hard in the first time), go dry-run then run your cut. Mine turned out like the one above.

Step 5: Observe, Write Notes, Leave It for the Next Person/cut

If you made it so far, congratulations! Time to get our reward. At this point we will have enough information to judge which feed rates are the optimal for our material and set of conditions. A few things to look for during the cut and consider :

  • "Chatter"/vibration during the cut - this cause bad finishing - which you normally care little for roughing but you want to avoid excessive as you don't want the whole machine vibrating too much.
  • Tool deflection - this is a bit more tricky as is minimally visible but a good illustration of it is this :
    • It should probably come with some chatter.
  • Noise
    • What is good or bad here is up to interpretation, but basically you need to trust your instincts to ensure that you're not having a lot of cutter-stock impact, not a lot of vibration. If I had to describe it as a square wave, I'd say high frequency and low amplitude. If high frequency is not achievable fine, but if it gets too loud (high amplitude) you probably have a problem.

Based on these constraints, you can find your practical limits without endangering the machine, yourself and your projects.

As promised, you'll probably know the best feed rate to keep you much less time in the machine than if you were doing a conservative cut. Enjoy!

If you have a workshop or even a single machine, it's probably worth keep notes written close to it as a reference for next projects.

Keep in mind that by doing this you're pushing the limits of the conservative and you're walking into a space where you should be reasonable and responsible for yourself. These machines are not smart, meaning that you have to be.

I hope you enjoyed this Instructable!