Hi All
I had my first go at digitising the other day, and whilst the results were fine for the job in hand (machining off weld from a cracked crankcase) the machined are had lots of random little dimples on it, as if the cutter was digging in see attached picture. I don’t think it was due cutter flex, the cuts were very light, cutter was 0.125 carbide ball nose. Anyone any suggestions why this was happening?


Paul

The simplest things confuse the cleverest minds.

Okay, I'm confused are you digitizing or machining.....those don't appear to be dimples (in other words they don't appear to be below the surface, but above the surface).....so, I would suspect that you're not using cooland and bits are being welded back.....I'm assumming this is aluminum.....

Sorry for the confusion. The program to machine the shown area was digitised from another part of the same crankcase. The work zeros were then shifted and the welded area machined. Yes the material is aluminium. The picture is not very good but they are definitely dimples. The cutter was 0.125” ball nosed carbide slot drill, run with flood coolant. It appears that on random points the cutter has moved below what should be the surface leaving a spherical impression from the ball nose. They are not very deep, or large, just unsightly. Surface finish in the dimples was the same as on the rest of the machined area with no indication that it was caused by cutter deflection.

I hope this makes things a bit clearer

Just looked at the picture, your right, it is a crap picture and they do look like they are raised

I'd guess that it's from irregularities in the digitising.

What diameter are the dimples?

How did you go about making the surface from the digitized points? Could it be such a thing as your program created the surface, but then machined everything including the points as discrete entities? Supposing that the points were used as a basis for a NURBS surface, they would not necessarily lie right on the surface. Perhaps it would be as simple as checking the code to watch where and when the dimples were created. Download a copy of NCPlot and check the backplot, one line at a time for a ways. This should uncover the mystery.

Gummy aluminum can machine in horrible ways. It doesn't matter the amount of coolant at times if the cutter was dull and even if it was sharp, some aluminums just don't machine well. Is this weld, or is this that psuedo aluminum brazing rod? Lots to consider when machinining aluminum and even then some of it tears instead of cuts.

MrWild may have touched on something...the dimples could be caused by poor flow during the welding process, or gas or contaminants caught in the weld...

Hi All
The dimples are about 0.030” in dia and are not inclusions in the weld they were created by the cutter, yep it was TIG welded, not alu brazed. I think the problem lies in the digitizing, I have down loaded NCPlot and will run the program through it tomorrow and see what it comes up with.

The job was done on an Ajax CNC DM 45. Built into the Ajax software is the digitizing package, which I guess is fairly basic, but is very simple to use, which is good for me. I guess I could have a backlash issue with the machine which might cause the problem, I hope not the machine is new this year.

Paul take a look at the gcode does it show the minor dips in the z axis code ? Terry

Hi All,
Haven’t had much time to look at this, but had a quick look tonight. The Z axis numbers are identical for each pass, so that’s ok. Back plotting in NC plot seems to show the dimples. A quick look at the raw data from digitizing and the CNC code produced from it, it would appear that the numbers are the same. Does anyone know of a program to compare the two files automatically? I will have a deeper look at the files as soon as I get time.



Paul

Like the UNIX diff command....

in DOS or a Windows command prompt use fc.exe the usage is:
FC FILE1.EXT FILE2.EXT
that will put the differences on the screen
if there are no differences then it will say something like "no differences encountered"
if there are a bunch of differences you can output the data to a file
FC FILE1.EXT FILE2.EXT > DIFF.TXT

where diff.txt would be a text file containing the differences and file1.ext and file2.ext would be the files with the appropriate extension (usually .cnc)

paul
try giong to another dec place (.001 /.0001) it could be the software rounding up/down the values .with a surface it takes very little to achive this effect
set the resolution to 1 dec place and see the differance for an extreme scinario
mike.

I didn't see any pattern on the physical dimples....maybe looking at the gcode and using Mike's approach with the data...you may see something.