[jonstap]

Help! I am getting so so frustrated. have attached a pic (i hope!).

Built an OXCNC and all working just won't cut accurate circles. They come out a funny shape but squares and straight lines are fine. Strange off circle shapes are totally repeatable every pass and every time. Any ideas on what might be wrong?. Using 3NM NEMA 23 with 2 power supplies (24V 10A each)

Thanks

Jon

[ger21]

You have backslash in whichever axis moves up and down in the pic.
Could also be a loose pulley.

[jonstap]

Thanks for the fast response will have a check, see what I can find. I do find my grub screw on the Z-axis keeps coming loose, might have to grind a bit of the threaded rod flat (good or bad idea?). Anyone know a way I can more solidly attach the aluminium pulleys onto the shaft of the steppers if they are loose? The Y axis is the culprit with the slaved Y and A belts. Can I glue the sprockets or anything just to prevent movement or is this a bad idea?

Jon

[wizard]

Thanks for the fast response will have a check, see what I can find. I do find my grub screw on the Z-axis keeps coming loose,

The pictured problem isn't Z related but that Z axis problem may hint at issues on the other axis.

might have to grind a bit of the threaded rod flat (good or bad idea?).

The couplings, pulleys or whatever you are using need to be tighLy fitted to the shafts. Trying to run sets crews into all thread worn do it. At a minimal you need to turn the diameter of the threaded rod down to a size that gives you a clean diameter to mount a pulley or coupling on. Once you have a decent fit on the shaft you can then consider a flat for the set screw to hear against.

Anyone know a way I can more solidly attach the aluminium pulleys onto the shaft of the steppers if they are loose?

Tighten the set screws down tight. If there isn't enough aluminum to support the set screw thEn you have a problem. Generally if you want high reliability you install a pulley with some sort of taper lock. Pulleys like that are expensive though and I not even sure if they can be had in the smaller sizes. You can use Loctite bearing mounting fluid to lock pulleyes in place but then you might as well call the assembly disposable.

The Y axis is the culprit with the slaved Y and A belts.

That statement doesn't make sense to me, normally the A axis is a rotary.

Can I glue the sprockets or anything just to prevent movement or is this a bad idea?

By glue I'm assumining you mean loctite and similar solutions. Generally there are two camps when it comes to Loctite, one camp rejects its use at all cost and another thinks it is the best thing to happen to machine tools since cast iron. I'm generally in the middle here Loctite and similar bearing mounting compounds are great when used in the right situation. Knowing when and when not to use Loctite is very important.

Beyond that you mention sprockets. This makes me wonder about your machines design. If it is using chain then one thing you have to deal with is the need to tighten the chain extremely taught. If there is no chain involved than the term "sprockets" is out of place.

Jon

[jonstap]

Great reply, thanks for the info, I have checked and the problem was the X-Axis pulley (runs on timing belts and I gather now are pulleys, thought they were called sprockets hah). Cuts proper shapes now.

My problem now being, although all is working, i calibrated the steps per mm and the machine cuts small when running Aspire generated code in Mach3. Then I tried calibrating using Mach3s auto calibration over 900mm. Set these but again G code cuts small (only a mm or 2, but still smaller than I wanted).

I cut a 29mm circle and it came out 27.5mm cutting on inside profile with 6mm cutter. I will get some experimenting done over the next few days to see if I can fix it.

Thanks for all the advice

Jon

[wizard]

Great reply, thanks for the info, I have checked and the problem was the X-Axis pulley (runs on timing belts and I gather now are pulleys, thought they were called sprockets hah). Cuts proper shapes now.

My problem now being, although all is working, i calibrated the steps per mm and the machine cuts small when running Aspire generated code in Mach3. Then I tried calibrating using Mach3s auto calibration over 900mm. Set these but again G code cuts small (only a mm or 2, but still smaller than I wanted).

I cut a 29mm circle and it came out 27.5mm cutting on inside profile with 6mm cutter. I will get some experimenting done over the next few days to see if I can fix it.

Thanks for all the advice

Jon

Did you walk through the steps required to calculate the overall system ratio before trying to calibrate? That is millimeters per step (a stepper based system). Done right this should get you pretty close. You can check linear movement accuracy by using the spindle to bore or spot holes that you then can measure. With a good caliper. In other words bore some holes 5 inches apart and measure what you actually get. This type of check can highlight an axis specific problem.

Being off by 1.5 mm is actually a lot, so you may want to walk through your calculations again.

[jonstap]

ok, have a strange problem now. I am checking for accuracy using Mach3 and putting in coordinates G0 X0 in 50mm units from 0mm to 350mm on both x and Y axis.

These axis run on belts. belts are tight, pulleys are solid. If I move the axis in one direction eg to to the right, then all coords are absolutely accurate every time. Whenever I travel in the opposite direction, then there is a loss of about 4mm on each point (non cumulative loss) except on the G0 X0 which is always accurate on the X axis but I get the 4mm loss on the Y axis.

so, to sum up, whenever I move to the right I am accurate, whenever I move left I am off by about 4mm every time but G0 X0 is ok. The Y axis has the same problem but G0 Y0 is off by a few mm (always short)

Any ideas

Just confused now

Jon

I thought it might be backlash but the right direction is always spot on accurate by coordinate whatever value I put in. The Y axis does the same but G0 X0 still has the few mm loss.

[ger21]

I thought it might be backlash but the right direction is always spot on accurate by coordinate whatever value I put in.

Backlash will do that, as it'll only show up in the other direction. But 4mm is a lot, and my guess again would be a loose pulley, that slips slightly when you change direction.

[jonstap]

Hi Gerry,

Thanks for the reply. If I move say 50mm to the left then I lose 4mm. Which made me think backlash. If I then move to the right the actual coordinates I requested will be reached as if those 4mm weren't lost. Then if I go to the g0 x0 position it will be correct without the 4mm loss. That's what is confusing.

Still backlash? I think I'm going to go the route of loctite at this rate to make sure of solid pulleys.

Jon

[jonstap]

Ok, I glued the pulleys onto the shafts of the steppers. They cannot be moved at all when the power is on.

Exactly the same thing is happening. When I type G0 X100 I will get to X100, if I then go G0 X50 I will get to X54 If i go back to G0 X100, I will get to X100 and so on. So, the machine is completely accurate in one direction only and a few mm consistently off in the other.

The belt is tight, the pulley is glued aaaargh!

Same on both axis.

I am pulling my hair out here trying to figure things out.

Any ideas?

[wizard]

Ok, I glued the pulleys onto the shafts of the steppers. They cannot be moved at all when the power is on.

Exactly the same thing is happening. When I type G0 X100 I will get to X100, if I then go G0 X50 I will get to X54 If i go back to G0 X100, I will get to X100 and so on. So, the machine is completely accurate in one direction only and a few mm consistently off in the other.

That is almost the definition of backlash, however 4 mm is a lot. So much in fact that you might actually see the machine taking up the backlash if you slow it down a lot. To dig a little deeper ar the belts driving lead screws or are you driving the axis directly from the belts?

If you are driving a lead screw I'd look into slop in the lead screw support bearings. This is a significant amount of displacement and you should be able to feel it by physically pushing the axis back and forth. When doing this you will want gain on the steppers to allow them to hold lead screw position. If it isn't the bearing mounts or lead scre fit in the bearings the only other option I can see at the moment is that the lead screw nut is slipping in its housing. In the end gross backlash like this can only happen in certain places.

The belt is tight, the pulley is glued aaaargh!

You must then hunk about what isn't locked in place.

Same on both axis.

I am pulling my hair out here trying to figure things out.

Any ideas?

Well from what you posted I really doubt there is a set up problem with your CNC controls. Considering that you need to look at the mechanical aspects of your machine that might allow for extreme backlash. The thing to keep in mind is that backlash by definition comes from taking up mechanical clearances. Since 4 mm is huge you need to consider where on your machine you can possibly have that much clearance. It could be a combination of places too.

If you have access to old secretary supplies a bolttle of white out ink can come in handy here. It would allow you to make the portion of various mechanical component positions and ideally would lead you to the area of the machine with the problem. The reason to use white out (on clean shafts) is that it dries real fast and leaves nice cracks when broken. That is one possibility but honestly you have such a large error here that you should be able to find it by feel or observation.

[ger21]

If you do a G0 x100, followed by a G0 X96, then the machine shouldn't move on the second command. If you change it to a G1 X96 F10, it should be slow enough to see why it's not moving.

[Devastator]

I'd use the caveman method of pushing and pulling on parts of the machine vigorously, but that's just me.

[wizard]

I'd use the caveman method of pushing and pulling on parts of the machine vigorously, but that's just me.

4mm of clearance ought to become obvious with this method.