On my taig, if I'm milling a circle and it is not coming out perfectly round, what could I be doing wrong? I'm milling bearing seats and doing both roughing using conventional and finishing using climb cutting (climb is only taking about 8 thou off).

Any tips on where to start looking for the prob?

I'm using the 280oz steppers and a xylotex drive. I've generated the G code using RAMS for the circles.

Thoughts or hints would be appreciated...

check your backlash

1-Check ur backlash
2-check that the setting for x-axis and y-axis leadscrew mapping is correct in controller software(Mach3 etc)

How is it not round? Is it ovaled? Or are there little notches at certian positions, etc?

For critical applications you should not make bearing seats this way. Bore it on your mill or turn it on a lathe, never profile mill it (if it's critical).

It will never be perfectly round if you profile mill it. How much out of round is it.

Regards
Phil

.On my taig, if I'm milling a circle and it is not coming out perfectly round, what could I be doing wrong? I'm milling bearing seats and doing both roughing using conventional and finishing using climb cutting (climb is only taking about 8 thou off).

Any tips on where to start looking for the prob?

I'm using the 280oz steppers and a xylotex drive. I've generated the G code using RAMS for the circles.

Thoughts or hints would be appreciated...

It depends where it is out of round and by how much.

Errors at the direction change points (12, 3, 6, 9 o'clock) are typically caused by backlash and take up errors in your drive'ball screws.

Out of roundness at 45 degrees off the vertical or horzonal axis are due to servo tuning errors. At this point, both servor are affecting a high rate or change in opposite directions. Unless the tuning is perfrect, you;ll start getting follower error and out of roundness.

Don't think for a minute that "backlash comp" will "fix" backlash problems - it won't. SOftware can NOT predict or compensate for uncontrolled/uncontrollable motion that occurs due to slop. You may not have a "perfect" machine which may be why you have some out of roundness. However, EVERY effort made to eliminate slop/backlash/etc or hother forms of lost/unpredictable motion will result in improved cuts and predictable geometries.

BTW, if you spend the time and money, you CAN tet mills to mill ROUND holes to amazing levels. However, it takes time and care and some5times more money than its worth, especially on a ''hobby" lathe/ or mill.

You could also use a rotary table to profile a bearing seat with an endmill.

Phil

For critical applications you should not make bearing seats this way. Bore it on your mill or turn it on a lathe, never profile mill it (if it's critical).

It will never be perfectly round if you profile mill it. How much out of round is it.

Regards
Phil

.

Thanks for the responses guys,

The holes are coming out out of round at about a 45 degree oval to the table

Assuming this is a tuning issue, any thoughts on how to go about fine tuning the stepper setup in Mach 3? I know how to change the settings but what should I look for in fine tuning?

I'm assuming this means i need to fine tune the # of steps/inch metric in the motor tuning area? Any references from other taig owners with a xylotex, 270oz stepper and Mach 3 setup?

re-reading the posts, I'll go back and tune up for backlash. Still confused on how to tune the steppers to help correct this though,

Check your gib plates, most likely you're dealing with missed steps like I was.

How deep are your passes ?


What's your feedrate ?

Sounds like a backlash issue, once you measure and input backlash into Mach it should come out round within a half thou or so. You will never get a perfectly round hole by this method though, it's always be off a bit.

In motor tuning all should be 32000 steps/inch with Xylotex.

Also make sure the head is trammed square in both x and y...

Thanks guys will do. I'm assuming something got moved/loosened in my recent move, but I'll go through it with a fine tooth comb.

I'm cutting plain old 6061 with 5/32 2 flute carbide for the roughing, (taking 6hundreths each pass in depth) and running at around 2500rpm. Doing my finishing passes at full depth taking about 8 thou of material off in climb cut with a 4 flute 5/32 carbide EM.

Mill does not sound like it is chattering or loading up really bad and I'm using flood coolant.

I am not altogether certain that it is backlash. I have just got my mill up and running. It has decent ground ball screws and fairly large THK rails. X are brand new rails and Y used, but looked and felt new. I have been getting perfect radius cuts, edge milling and drilling of holes in my parts. They line up perfectly with a printed version of the part. I am also doing some logo engraving with a drill mill. Extremely small and fast axis moves. There is no backlash in this machine that can be seen in the motion. It may mic some, but it would still have to be very small.
When I try to mill 1/4" holes with a 3 flute 1/8" endmill, the bottom side of the hole is flat. Kinda flat tire looking. The hole is correct on the other three sides though.
I measured this flat and put the opposite curve to get a round again in the drawing. This time it cut a perfect oval just like I had drawn. I have not enabled backlash in Mach 3. I am using SheetCam and Turbocad.
I am drilling the holes instead with a milldrill. I just thought this was very peculiar action for a machine with decent accuracy.

Backlash affects you at MOTION XDIRECTION CHANGES - the monkey motion problems at 45 deg are happening while the motors are attempting to do the most/fastest relative changes in velocity - one is speeding up, the other axis doing the opposite SIMULTANEOUSLY... A simple look and see at the motors will show/prove this

To make the motors "walk and chew gum" at the same time, they have to be timed EXACTLy the right speed relative to each other thought the motion cycle. We chased this all over hell and back until we FINALLY figured out how to properly tune for "follower error". Cutting ROUND circles is THE definitive test to assure you that you have servo follower error tuned properly.

Trust, me, it isn't back lash and it ish't backlash comp. It is a problem in the instantaneious RATE of feed change that the axis are making that is causing the 45 deg out of round problems. If the controller is NOT changing the RATE at which the step pulses are instantaneously being fed to the steppers, they won't/can't keep the poper relative speed to each other and they thus CAN'T cut round holes.

Hello, I have been following this thread and it's very interesting about the rounds. I know a lot of people would be happy to find out how to figure out "follower error"
Any guidelines we can go by???
Thanks

can you tune out follower error on stepper systems just like on servo systems? i think the steppers would need to slow down the acceleration so they can keep up with each other. i'm new to steppers and slowly putting together a stepper driven mill just for fun, so i'm just curious how you would go about tuning a stepper drive?

I also use Turbocad (with the integrated Cam plugin) and have had a similar but not the same problem. If you "explode" an "entity" down to a certain level it becomes a multi-sided polygon which, when "camed", can produce a flat side on a radii.

You might spot it by looking carefully at the gcode or by zooming in on the problem area in Turbocad. Have you tried a tool path simulator. I think you need to do some work to find out if it is the software or the hardware before pointing the finger.

Regards
Phil

. I am using SheetCam and Turbocad.
I am drilling the holes instead with a milldrill. I just thought this was very peculiar action for a machine with decent accuracy.

I have noticed that with Turbocad 11.2. The second explode on a point and circle looks like a wine barrel shape. I didn't explode these at all. Perhaps this is where the circle (perfect) is loosing in translation. Perhaps a 64 sided circle would turn out better. ;)
I did play with my machines acc. and velocity this afternoon. Both X and Y screws and motors are identical in size and pitch. I did not try to go faster acc. on the flat side of the circle though. I see now this is what I should have done. I still have more of these to cut out tomorrow, so will try it some more.
I will try another test drawing as well. As I mentioned before, radiuses turn out perfect, so it does sound like either the drawing format or follower error.

Hello, I have been following this thread and it's very interesting about the rounds. I know a lot of people would be happy to find out how to figure out "follower error"
Any guidelines we can go by???
Thanks

There are two problems here and they are independant of each other.

Backlash has crept into the mix and is probably NOT the cause of the out or round at 45 deg points. Backlash tends to result in choppy shapes and /or flats at direction change points. ANY hysterisis or slop in your system results in lost motion which ends up generating choppy/irregular cuts. Simple back and forth moves of 0.002" (measured and verified) will determine if you have backlash - how you eliminate it is pretty much up to your budget as mechanical slop is usually solve by spening money.

When you start having out of round milled circles, it helps to know WHERE the out of round occurs. The OOR at 45's is occurring at the points of motion where the moros have to be perfectly timed (speed wise) as one is speeding up and the other is slowing down. Thus, your controler HAS to be in sync woth BOTH drive channels or else one channel will lag or lead the other. Result: speed is wrong and cut rate is wrong and OOR develops.

My understanding of 'follower error" was based upon a quick and dirrty explanatino offered by my mill tech. Basidally, you give the servo a command to go some place at a certain rate and it goes off and does so. However, the system MAY not be capable of doing it as fast or as readily as you ask/tell it to do so so the system lags mechcally. To make up for this, you can slow things down as this makes it easier for the machine to "follow" without getting behind the asked for/prescribed motion.

In our CNC servo system, we had backlash and hysterisis problems. After solving them, we uncovered the OOR cut problems. One by one, we solved the mechnaical ones and then attacked the electronic ones. Only when we were 100% confirdent that we had the mechanical slop isssues resolved (many measurements later) did we embark on servo tuning.

However, when the guy did do the tuning for us, he was rather impressed at the level of "roundness" achieveable via a tuned up CNC mill. IN our case, we can hold roundess within 0.0005 or less when we try hard - we've seen under 0.0003 uf we gi real slow. It is doable but takes time and patience.

First get rid of ALL your slop, then go after tuning.....

Tuning and following error is a bit different when talking servo's and steppers, following error in servo's is done using a PID filter process, ( a Google of PID should bring up an explanation).
Servo's use feedback devices, encoders etc in order to use the PID process, so the control knows where the servo's is.
With steppers, you have no feedback, so the controller expects the motor to be where the step output commanded it to, if you get following error with a stepper, it is usually because of lost steps, so tuning is basically a prevention of step loss.
The PID process can correct following error because of feedback, the stepper cannot.
BTW there is a tuning video for servo's on the Galil site that displays the trace of a typical following error when accelerating and at speed etc.
Al.

It could be a software problem

I have a Tig mill, and for ages the best circle I could make was a square that had rounded edges, almost a circle.

I tried loads of diff settings in mach3, I adjested my backlash.

A good test is to let mach 3 make a circle for you with one of the wizards, if it's ok, you know it is a gcode/cam issue.

I was using corel draw and lazy cam, I tried a few others and had simialr problems, I then used artcam with it's post processor and not looked back since

Something to consider at least

DJH

I just tried it again using four radii instead of a circle. It came out quite a bit better. Nearly a circle. For a 1/4" circle, I used four.124" radii. I think part of it is also the cheap end mills I am using. China express. I knew going in that I would have some trouble with small end mills. I'm not a machinist. I got 10 of these for the price of one USA made. They are decent for making sure that everything is setup right. I measured .004" difference top to bottom, both directions. The rest may actually be some backlash. I have yet to put a dial on it. Will post my findings then.

Not sure how your steppers are mounted, but on my system it included a trust bearing, shaft coupling, bore coupling. Sometimes the couplings get a little lose, might want to check those out as well. :)

whats your feed rate?could also be to fast,even in my hass if im too fast its not going to be round,try 5 imp and if it still out than its backlash or tuning,or even both