[bremerr]

I built a JGRO but I'm having a problem with the size of the parts that I'm cutting. I've thought about this for a long time and I have no idea what is wrong.

Some background info:
Aspire to create tool paths, outputting using Mach3 post processor
Mach3 to run the g-code
1/2-10 acme screws, all axes
I have probotix bi-polar drivers
PacSci bi-polar motors
The steps per inch are all the same, (1600 or 16000, I can't remember, running 1/8 microstep either way)

If you need any more info let me know!

I created an extremely simple program to cut an ID and an OD. Both were drawn to 2 inches in diameter. I'm using a 1/4" router bit. For this program I roughed the diameters leaving 0.050 (on the radius) and then went back and did a finish pass. In Aspire I used the profile command for both circles.

The ID is coming out undersize and the OD is oversize. If there was an incorrect setting or something I would expect them both to be small or both to be large, not one large and one small.

Both the ID and OD should be 2.000 inches. The ID comes in at 1.960 and the OD measures 2.045. Both of these features were cut during the same program and with the same tool!

Here's the kicker, the slots left from the router bit are exactly what they should be: 0.300 inch (remember I used a 0.250 bit and left 0.050 during the rough cut.

I want these features to come out the same size! I need some help badly!

[Mountaincraft]

Is this happening with just 'this file' or could it possibly be happening with all files but you just now noticed it?

Are you 'sure' that your 'steps per inch' settings are 'perfectly' correct in Mach 3?

Also, have you checked 'roundness' (ie, is it off the same amount in X and Y)?

[ger21]

Also, try cutting some squares and measuring those. You may have to adjust your steps/unit slightly.

Another thing to try is cut in the opposite direction. Tool or machine flex may result in different sizes depending on cut direction.

[louieatienza]

Aside from machine ridgidity, I'd check to see if your bit is truly 0.25" or maybe .02" undersize! You can test this by simply measuring the bit, or making the tool diameter .23" and regenerating the g-code. Also try running the same program on a different part of the machine, or run a matrix of 3 or so and compare the measurements.

If you climb cut, the tendancy of the bit is to deflect away from the "cut line," and the opposite is true with a conventional cut. Sometimes I'll rough cut with the climb direction, and do the finish pass in the conventional direction.

[bremerr]

Is this happening with just 'this file' or could it possibly be happening with all files but you just now noticed it?

I noticed it on other files, that is why I made this super simple file just for a test to see where I was at when cutting an ID and OD of the same size.

Are you 'sure' that your 'steps per inch' settings are 'perfectly' correct in Mach 3?

No I am not sure. If there is something I should check let me know.

Also, have you checked 'roundness' (ie, is it off the same amount in X and Y)?

It's not perfect but it's pretty close.

ID Cut - X Direction 1.960 - Y Direction 1.945
OD Cut - X Direction 2.045 - Y Direction 2.040

[PaulRowntree]

I use V-Carve Pro, not Aspire, but I don't understand the use of a Profile to give anything but the width of the cutter. Did you manually do this? The errors are very close to the 0.050 'finish pass' that you mention.

On the smaller disk there is a slight notch on the top of the circle; the large circle may have a flat in the same spot. Is this a ramp in or perhaps evidence of flex/backlash?

If you make the patterns exactly twice as large, do the errors scale as well? If not, I would suspect machine flexibility. If so, it looks like steps per inch entries need to be revised.

[Mountaincraft]

I noticed it on other files, that is why I made this super simple file just for a test to see where I was at when cutting an ID and OD of the same size.



No I am not sure. If there is something I should check let me know.



It's not perfect but it's pretty close.

ID Cut - X Direction 1.960 - Y Direction 1.945
OD Cut - X Direction 2.045 - Y Direction 2.040

Sounds to me like your steps per inch is off.. I'd start there... Adjust them by the same percentage your measurements are off... You set them in the motor tuning page of Mach 3.. The same place you set acceleration and velocity..

make a change, and measure the difference... make another change, measure the difference.. and so on...

Other than the tool diameter Louie mentioned, that's the 'easiest' fix (if that's the problem).. but if you're noticing it with other files/tools too, then it's probably not a tool diameter issue..

[Mountaincraft]

Re-reading your original post, I would try just cutting the circles again without messing with a rough pass and finish pass.. because your dimensions are off about the same amount as this rough/finish difference..

Just cut a 2" circle outside the line with the 1/4" bit (set to .250 diameter in the tool settings), and another inside the line with the same tool setting.. and then measure it.... see if it's any better...

Because it's acting like it sees the tool as being about .30" in diameter... which is the .050 you added for the rough/finish thing..

My guess is that you have something set up in Aspire and it's doing exactly what you're telling it to...

[jsheerin]

You could always see what your g code says. If it has the correct commands in it, you know that it's your bit, something in your controller setup, or something on your machine.

[bremerr]

Also, try cutting some squares and measuring those. You may have to adjust your steps/unit slightly.

I cut 3 squares. They were drawn 2" x 2". One of the squares had the tool travel right on the 2" square. The next square was an outside profile, and then an inside profile. See the image for the measurements.

There was no rough cuts, just one pass with a 0.250 spiral upcut bit.





I'm thinking of adjusting the bit size in Aspire to 0.260 and running the program again to see where I'm at. If anyone else has any other feedback, that would be great.
.
.

[Mountaincraft]

Man, from looking at that, I'd say you have one of two issues going on...

The first issue, is that something in your machine is 'loose' allowing it to move from where it's supposed to be... the fact that one of the cuts you measured varies in width from one end to the other makes me think that either your router (or bit) has a lot of run out, or the machine is 'flexing/moving as it cuts...

The other possibility is that you're losing steps.. and this seems like the most likely to me from those pictures.. The problem is on 'one' axis only, and gets worse from when you started the top square to when you finished the last square on the bottom...

It looks to me, like your gantry is racking and gets worse as the project goes on... I had a similar problem, and it was because one of my X axis motors was losing steps..

So a couple questions..

Were those cuts made in order from top to bottom?
Was the machine homed/zeroed between cutting each of the squares?
Were the squares all cut in a single pass for each?
Describe your machine.. Do any axis' have dual drive?
Also, in the picture is 'X' from left to right, and 'Y' up and down?

Home and zero your machine, then measure the distance of your gantry to the far end of each of your rails.. Then run the program again (the longer the program the better)... then when finished, without any homing or zeroing, measure the gantry to the end of each rail again and see if they're the same as the original measurements or have they changed and how much...

[bremerr]

So a couple questions..

1. Were those cuts made in order from top to bottom?
2. Was the machine homed/zeroed between cutting each of the squares?
3. Were the squares all cut in a single pass for each?
4. Describe your machine.. Do any axis' have dual drive?
5. Also, in the picture is 'X' from left to right, and 'Y' up and down?

Answers:
1. The squares were cut from bottom to top.
2. There was only one file that cut all three squares, sequentially from bottom to top, using the same bit. There were no stops from start to finish.
3. Yes, one pass per square.
4. It's a basic JGRO, one motor/lead screw per axis. See photo below.
5. In relation to the photo with the squares, X is left to right, Y is up and down.

[PaulRowntree]

It seems like the problem is really in the bottom cut; the others are within 0.01" on mdf, and depending on how much tear is in the measured area, it is easy to loose several thou in the measurements.

On the bottom cut the width looks constant to my eye, yet the numbers show an enormous error : are those numbers right?

Again on 2 of the cuts there is evidence of positioning error in the corner unless you ramped in on those.

I think doing a series of cuts of different sizes would speak volumes, especially in seeing if the errors scale.

[Mountaincraft]

The bottom one is visibly out of square... Very odd that it 'started out' making the worst cut, then got better over time, finally being 'perfect' on the last cut...

If you're using Mach 3, just for grins try setting a delay of a few seconds to allow the router to come up to speed before the program starts the cutting... see if that changes anything...

Also, the fact that it made a single line with differing 'width of cut' is strange too... Is there any play when you grab the bit and wiggle it?

[louieatienza]

Man, from looking at that, I'd say you have one of two issues going on...

The first issue, is that something in your machine is 'loose' allowing it to move from where it's supposed to be... the fact that one of the cuts you measured varies in width from one end to the other makes me think that either your router (or bit) has a lot of run out, or the machine is 'flexing/moving as it cuts...

The other possibility is that you're losing steps.. and this seems like the most likely to me from those pictures.. The problem is on 'one' axis only, and gets worse from when you started the top square to when you finished the last square on the bottom...

It looks to me, like your gantry is racking and gets worse as the project goes on... I had a similar problem, and it was because one of my X axis motors was losing steps..

So a couple questions..

Were those cuts made in order from top to bottom?
Was the machine homed/zeroed between cutting each of the squares?
Were the squares all cut in a single pass for each?
Describe your machine.. Do any axis' have dual drive?
Also, in the picture is 'X' from left to right, and 'Y' up and down?

Home and zero your machine, then measure the distance of your gantry to the far end of each of your rails.. Then run the program again (the longer the program the better)... then when finished, without any homing or zeroing, measure the gantry to the end of each rail again and see if they're the same as the original measurements or have they changed and how much...

Paul I think is onto something here. If your gantry sides are racking, or your carriage assembly has play, then when you climb cut, the bit will "push" away from the cut line. You can try to do a second pass with a conventional path to see if that helps. You could also make sure all axes run smooth; this may mean disengaging the leadnuts and making sure each axis moves evenly with no play.

I found with my Solsylva build that there is an inherent "spring" to the bearing setups, and the bearings can be run snug without significant drag, since the "springiness" will take up a little inaccuracy in the rail.

[bremerr]

Based on some of the posts here, I think there is some mis-understanding what I was trying to do.

I drew 3 squares, all 2 inches by 2 inches.

Bottom square: I had the center of the tool follow the 2x2 square. Based on a 0.250 bit diameter, I would expect the island to be 1.750 x 1.750 and the outside of the groove to outside of the groove to be 2.250".

Middle square: I had the tool go around the outside of the square. The the software takes into account the radius of the tool and moves the tool path outward that radius. I would expect the middle square to measure 2x2.

Top square: similar to the middle square except I had the software traced the inside of the square. I would expect the outside of the groove to outside of the groove to be 2 inches for this square.

My tool path screen shot is below. The black squares are the 2" squares, the pink line is the tool path.

[bremerr]

On the bottom cut the width looks constant to my eye, yet the numbers show an enormous error : are those numbers right?

Again on 2 of the cuts there is evidence of positioning error in the corner unless you ramped in on those.

Can you show me the math you are using to conclude there is an enormous error? I see that 1.760 + 0.260 + 0.260 does equal 2.280, which is what I measured.

I did not ramp anything. The corners that look funky are from dwell plunging in and coming out.
.

[PaulRowntree]

Can you show me the math you are using to conclude there is an enormous error? I see that 1.760 + 0.260 + 0.260 does equal 2.280, which is what I measured.

I did not ramp anything. The corners that look funky are from dwell plunging in and coming out.
.

I am sorry, my brain is on vacation or is just out to lunch. I just added the cutter once.

So the measurements are within ~0.010" in all cases. Could these errors just be because you are measuring mdf cuts? To check this, I would use a dial to measure exactly how far the gantry or Y carriage is moving, using air-cuts or the real thing.
Again, my apologies for not enaging brain before commenting first.

[acondit]

I am guessing that it is backlash in one axis on your machine. Measure the inside and out dimensions of each square in both the x and y directions. That should tell you where the error is coming from. You need to figure out the source of the error before you can fix it.

Alan

[Goose80]

Hey.
Old post, but I have similar problem.
Did you find out what was wrong with your CNC?