Bit Deflection.

[nat_wheatley]

How much is my bit deflecting!? From what I can tell I've ruled out everything else, even raising my table 3" to curb any deflection in the Z axis, but I'm still getting a considerable amount of deflection showing up in my cuts. I'm able to work around it, with a combination of climb/conventional cuts but am curious more than anything how much can the bit deflect. I'm using a 1/4" compression cutter, running at 4ips, 16000 rpms, and my rough guess would be that I'm getting 1/32" flex somewhere.

[ljdm]

What material are you cutting, and how much per pass are you cutting? Too much per pass in a harder material at 4ips with a 1/4 cutter will cause deflection.

[nat_wheatley]

Hi Lou,

I'm cutting 3/4" veneer core plywood, going just about all the way through on the first pass. I realize it's asking a lot of the bit, but I would have thought the bit would snap before flexing like that.

[gabepari]

I'm sure some of it is bit deflection, but I would bet that you are still getting flex in the machine. There are lots of places it will flex, and it all adds up.

Get a dial indicator, place it on the table with the nib against the collet. Then go around pushing on various places on the gantry (even the table), you will see the indicator move. It doesn't take much.

Good Luck,

Gabe

[ljdm]

Gabe is also right - it's probably a combination. Figure even 1/64 bit deflection, 1/128 router/spindle play, 1/128 gantry deflection, etc, etc......... you might have to slow down, do two passes, or one clean-up pass, or all of the above. Play in spindle/router shaft, gantry flex, pinion wear, bit deflection, even though minute amounts, if present in all of the above at the same time, could add up.

[beacon14]

I get a similar amount of defection with the same bit and feed speeds if not a little faster. I think I could tighten up the machine a little to minimize it but not much. Some of it is bit deflection and that's not going away. You just have to allow for it in your toolpath strategy. The climb/conventional approach is a proven winner.

[Gary Campbell]

Nat...
I've gone thru the same things that you are trying to cut cabinet parts. Gabe & Lou are right on the marks. Every place on the machine that there is movement, there is clearance. If you add a few thousandths each from the wheels and pinions from the X car to the wheels and pinions of the YZ car and then the rollers in the Z you can see how it would add up. If each were just a few thousandths, you could be close .020 under load. Throw in .010-.015 of flex for the gantry mechanism and another .010-.015 for the bit and you can see how there is no benefit to pushing the tool.

To test, set up a full sheet of material on the bed. Using a V bit make .020 rectangles as large as possible and check dimensions and diamonds. Adjust gantry for square as needed.

When no load rectangles are within your specs, put in a 1/2" (little or no delection) bit and cut a 20" square 1/8" deep @ 2ips and check dimensional accuracy. Try another at double the speed. Then try another at 5/8 deep @ 2ips, then another at 4 etc, etc.

Then put in a 1/4" bit and try the same... this will show you whether the bit or the machine is flexing.

I like to Vcarve 8 22" squares on the sheet and cut inside, outside, climb and conventional at different speeds and depths to find the sweet spot for my machine. THe no load cutting of the V bit lines are a good benchmark to compare a loaded toolpath against.

Dont forget that all your rollers and pinions have adjustments. Making sure that they are all properly adjusted is critical to cut quality.
Gary

[nat_wheatley]

Thanks to everyone for the posts, it all makes sense. I'm actually looking forward to trying each of the tests and fine tuning things where I can. It's also somewhat of a relief to be able to use the climb/convential cuts to compensate for the play/flex/deflection... that I can't correct.

[gpari]

Nat, also some bits geometries flex more than others (compression vs straight). I try to use at least 3/8" tooling whenever possible. Today I had to cut some 3/4" MDF with a 1/4" compression cutter. Tool was running at 6 ips and 13k rpm, with the first pass climb(.650" deep), second pass conv(.010" into the spoiler). Although I got nearly 1/8" deflection on areas where the bot actually hit 6 ips, the finished parts were dead on.

Gabe

[jhicks]

Seems common to be off as much as 1/32 but I'm a bit suspicious of 1/8" as deflection. In fact I would guess its not deflection as much as table and gantry calibration. Might try this.
Take a nice new 1/4" bit and make a center drill pattern on your x and y with spacing between holes maybe an even 1.0". Run the pattern up the y 48" and down the x 96" and drill all holes 1/16" deep or just enough to get a true clean hole.
Now measure the distance from the outside of one hole to the inside of the next. Effectively center to center but easier to see clear edges to measure from. Should be 1" right? Unless of course your 1/4" bit drills .245 holes but all things considered, each space from one hole to the next should be equal. Now measure the composit like 12" from one to another and see how much you are off if any.
I have seen holes/spacing "rubber band" or be dead on for the 1st 12 holes/inches, then under 1" than back to 1" or more. In the end it cut parts on the lower Y side of the table on dimension and on the high of the Y under by 1/32".
After messing with it a while I recalibrated revolutions/ inch settings for movement on X&Y (Mach software option but should be similar on SB calibration for movement increments calibration) ,resquared gantry, tightened up everything including Z axis V rollers,pinion gears, and dialed it back in so running the same tests gave me spot on results between and overall with precision measuring scales. Knowing the drilling was accurate, all that was left was loose, out of square, or deflection.
Acid test is to cut two parts the same rectangular size, flip one over and lay it on top of the 1st one. If its equal in size and square with no ridges to the touch, you have arrived. I say this because your 1/8" and your 1/32" "deflection" variation could easily be partially out of square gantry plus minor calibration discrepancy over longer distances . Two or more identical parts at cut at different parts of the table flip one over ant 180 degrees and compared is the real proof.
good luck, time to dial in is worth the effort but out of square can creep in at any time. Especially after a bogged down cut.

I check square every day and often more than once daily and often find I'm off 1/32" to 1/8" just from the night before over the 96" length corner 0 to corner 60 of the Y gantry.

[gabepari]

Trust me, my table is square, and everything that moves is tight without slop. I pride myself in being one of the most anal guys I know


My Z is trammelled within 2 thou over a 8" radius, Y-axis is square to an indistinguishable amount over 60".

Try this, take a 1/4" (7/8 CEL) compression cutter (slightly dull) run it 6 ips 13k in 3/4" MDF. Cut a 12" square, first pass .650" in the climb direction, second pass .010" into spoiler in the conventional direction. You will most definitely see deflection. Stuff that is beyond our control (without, of course, reinventing the bot as it is) is, will and does flex.

Granted, I could slow the machine down and increase rpms to reduce cutting forces and see less deflection, at the cost of overheating the cutter. But the deflection in the first pass is directed into the waste material, so it doesn't hurt anything and my bits seem to be happy.

Also, save yourself the hassle of "checking" for square everyday. Adjust your stop blocks, then each time you start up the machine pull the gantry against the blocks before powering up. Presto, it's perfectly square every time. No more "checking", you "know" it's square the instant you turn it on.

I'm reminded of those high speed films in high school physics class demostrating how much the wheels and frame flex on a bicycle when it's ridden.

Gabe

[jhicks]

I hear you Gabe. But what happens when you run the hole drill test and measure between and overall?
Glad you're confident. Just offering another opinion for what its worth.
1/8" deflection still sounds big to me and unacceptable for parts that need to fit together but whatever works for you is always the best solution.