Dimensional Accuracy in MDF vs Various Milling Combinations

What the heck does that title mean? I’m working on a project where I’m routing some MDF on a ShopBot. To help me get some good angles and built in guides for assembly, I decide to add an edge relief for a 90 degree joint. Basically, I milled out a 1/8″ deep recess for some 1/2″ MDF to sit in. Here’s a picture to explain it:


A vertical piece of MDF sits in that recess to create a 90 degree joint.

The problem is that when I put the pieces together, I had a very small overhang, pictured here. That grey piece should be flush with no line.


When I measured it, I discovered that the width of the edge recess was something like 0.485″, not .500″ which is the thickness of the MDF. I had everything set up right in Fusion 360 and the overall outside dimensions of the pieces were accurate. However, that recess was not. I went back and discovered that I had set the edge pocket clearing method to climb milling while the overall contour cut was set to conventional. Could this be a problem?

To determine this, I created a 2×2 DOE (design of experiments) matrix. One factor was climb/conventional milling. The other was the edge/pocketing milling and the outside milling. I created a 4″x4″ piece with a 1/2″ recess in Fusion 360. I made four of them, and created 4 separate tool path combinations that varied the milling used on the recess and the outer cutting. So, for example, for one of the pieces, the recess was milled with climb milling while the outer contour was cut with conventional. Since this is a 2 factor experiment with two levels, that’s 4 possible combinations. Here’s what the pieces look like in Fusion 360. (Click on the picture for a larger version that explain the edge/outside definition I’m using.)


I also threw in a 3rd factor: milling bit. I ran one set of parts with a downcut bit and one with an upcut bit. I did this to see if bit selection had any effect.

I then measured the outside dimension (4″ by design), the inside dimension (3″ by design) and the relief width (0.5″ by design). I did this without any sanding to make sure I was getting an “as is” measurement. Here’s an example of a measurement:


I then made a 3d chart of the data with the recess averages (of upcut/downcut measurements) plotted against “edge” climb/conventional cutting and “outside” climb/conventional cutting. Remember, “edge” is the recess that is pocketed out. “Outside” is the overall dimension.


And here’s the interesting thing. When you use the *same* type of milling for both operations (climb/climb or conventional/conventional), the recess comes out to about 0.500″. However, if you use climb on the recess and conventional on the outside, the recess is too short. On the reverse, it’s too long. Looking at the averages, we can see that climb milling results in dimensions that are slightly too long. However, when climb milling on the edge is combined with climb milling on the outside contour, this error cancels out, and you get about a 0.500″ recess.

The data also shows that there doesn’t appear to be a significant difference between upcut and downcut bits.

Averages Outer Dimension 1 (in) Outer Dimension 2 (in) Inner Dimension 1 (in) Inner Dimesion 2 (in)
Climb 4.012 4.010 3.019 3.016
Conventional 3.993 3.994 3.001 2.999
Downcut 4.004 4.002 3.010 3.011
Upcut 4.001 4.001 3.011 3.005

Some things to consider with this data:

  1. I didn’t record values by machine axis. In other words, I can’t tell if a dimension was cut in the Y axis or the X axis. This may have an effect as there could be differences in the axes on the machine.
  2. I only did one run of each cut and only one measurement. There’s not enough data to run a statistical analysis to determine how significant this data is.
  3. I ran all runs at 180 in/min with a max depth of .125″ per cut. Obviously, chip loading, feed rate, etc. would all have an effect on the cutting.

So why the difference? Well, if you think about it, in climb milling, the bit takes a small cut first which grows. In a compressible material like MDF or wood, it’s possible that the material gives a little and then flexes back. This would lead to material with a longer dimension. Whereas with conventional, the bit takes the largest bite first, which doesn’t give the material as much of a chance to compress.

So, overall, use conventional milling on MDF. But if you don’t want to or can’t, make sure that for any important dimension that is affected by two (or more) cutting operations, you use the same milling convention.