Educa's BigFoot

[jv4779]

Thanks for the hint about finding focal. I'll test it out since I have motorized Z table. I believe that in your config that is then the W axis ? what axis number would that be for the parameters in the ini file ?

W axis is AXIS_8. In Ben's 2x_laser config he had the table on U axis, but in conventional CNC machines U is parallel to X, V = Y, W = Z. So I changed it to W. A side benefit of W is that when doing a top down render of the tool path, W doesn't effect the X/Y positions.

Jeremy

[educa]

I am having some unforeseen problems with my machine.

I tried to cut a 1" circle using the different microstepping possibilities of my stepper motors.

The motors are 1.8° nema23 steppers. The biggest problem is that 1 full step on my stepper is 0.5mm of travel.

So I tried 25x microstepping and thought that I might get 0.02mm step resolution.

The result is something completely different. Instead of 0.02mm step resolution I get the results you see on the picture (taken with different microstep settings on the motors)



As you can see, in 1/4 step mode you clearly see all the steps of this motor and the cut is UGLY. 1/8 isn't better and from 1/16 and 1/32 I start to get better results.

its getting BETTER, but not yet good

The reason is probably because in microstepping mode steppers always tend to want to go into fullstep position. So if you are inbetween steps, you get actually not the wanted resolution but something inacurate.



I was looking at planetary gearboxes, but it seems like these would not be a good solution for me (I wonder why).

I was also considering using 0.9° motors and a 10:1 reduction between motor and axis by using 2 steel spur gears with 12 and 120 teeth.

And my last possibility is of course the most drastic one .... ballscrews.


Any ideas / hints / tips ?

Kind regards,

Bart

[BenJackson]

Sorry I'm not getting this out of the history, but what's the controller? A microstepping controller should easily be able to hold a stepper motor between "detents" as long as it's enabled. BUT the microstepping positions (other than about half) aren't really linear and you can't rely on them for accurate positioning (just smoother positioning).

I'd also say that given the nice clean cut you're getting in plywood that you're using pulsed cutting and the edges look more like the artifacts of a really low PPI than stepper positioning errors.

[educa]

the cuts you see are 400PPI, I also did 800 and 1200 PPI and didn't change a lot.

I must admit that the cut edge is indeed very nice (absolutely no burn) and for a lot of purposes the cut quality is more then enough. Its just that for things like acrylic this side finish is ugly. For plywood I guess its more then ok.


1 relatively cheap thing I can do is that I can replace my 1.8° steppers with 0.9° steppers. That would certainly not make the quality worse and if its really a problem of microsteps pulling towards full steps, then this would even be a good step.


There exists also stepper motors with 0.45° / step resolution. These are manufactured by Lin Engineering (1 company in the whole world I know of) and are quite expensive. Around $300 for 1 motor with around 1.8Nm torque

[educa]

This http://en.nanotec.com/steppermotor_st5909.html looks like a nice stepper motor to try out.

[kbob]

The motors are 1.8° nema23 steppers. The biggest problem is that 1 full step on my stepper is 0.5mm of travel.

Bart

Can you reduce the diameter of the motors' output pulley/gear? (I tried to find a photo of the drive in your buildlog, but if it's there, I missed it. Are you using a belt and pulleys?)

For comparison, the pulleys in the 2.x design are 20 teeth × 0.08 inch/tooth = 1.6 inch/rev (40 mm/rev). A 1.8° motor gets 0.08 inches of travel (0.20 mm) per full step.

Edit: I downloaded the Sketchup CAD file. It looks like you are in fact using a belt and pulleys.

[educa]

I use steppers nema23 with 20teeth T5 belt (so 5mm per tooth) that makes 0.5mm travel per step on the stepper.

If I would replace the 1.8° stepper with 0.9° steppers then I will effectively double my resolution.

I was looking for 0.45° steppers too but they seem to be extremely expensive (although I might still consider them because as you can see my machine is quite an achievement for me and I want to make it perfect now).


I was first looking for planetary gear reduction, but thats not a good solution (they told me) because of backlash and the fact that I wouldn't reach acceptable raster engraving speeds.

On the other hand, the quality is also not THAT bad. Its just a search to get it better:)


yesterday I put my 100mm lens instead of the 50.8mm (2") lens and tried again and suddenly a lot of mu problems were gone. I suspect that this is because the 100mm lens makes wider cut and the cuts overlap better making the error vanish more.

I don't think higher PPI will solve this, because I tested it upto 1200PPI. I must confess I did not do ANY test without ppi. Maybe thats a fault of me.

[cvoinescu]

FWIW, I agree with educa and disagree with BenJackson: they really look like stepping artifacts rather than PPI artifacts.

Some controllers are better than others at microstepping. Surprisingly, the $12 Pololu A4988 boards are quite good. One important thing to consider, with any controller, is that its operating point must be well into the current-limited region for microstepping to work. So if your power supply is 12 V and your motors are rated 8 V, 1.5 A, that won't work very well. Ideally, the voltage of the power supply will be about an order of magnitude above the rated voltage of your steppers. If your supply is 48 V and your motors are rated 4 V, 4 A, that's great.

Low-impedance motors (those with low voltage ratings but high current ratings) tend to be better at microstepping than higher-impedance motors, even after adjusting for power supply voltage.

Drivers are optimized to run close to their rated current. If you're driving 1.5 A motors from a 8 A driver turned way down, the driver may not have very good control over the current, and the microstepping may suffer.

Another important thing is that the torque of the motor is very low when its position deviates only a fraction of a full step from the commanded position. If your carriage doesn't move extremely easily, the motor will stay still as you microstep away, then it'll jump a number of microsteps at a time.

Finally, the detent toruqe interferes with microstepping, to the point that at very low current the motor may move full steps at a time anyway. As with the drivers, the motors microstep better when driven with currents close to their rated current, and less well when the current is much lower than that.

So, short of replacing drivers and motors, your best shot is to increase the driver current, if possible.

If replacing motors, definitely get 0.9 degree steppers, and do not oversize them too much. Both motors and drivers run better (as far as microstepping is concerned) when driven close to their rated current (let's say at least half the rated current for the motor, as a rule of thumb; for the driver, it depends, some regulate low currents well, some don't), with a power supply many times higher voltage than the motor rating.

[educa]

The motors I currently use are http://www.cnc4you.co.uk/index.php?rout ... duct_id=65 and I drive them bipolar parallel at 4.2Amp with a 36V PSU

The stepper drivers are http://cnc4you.co.uk/resources/CW5045.pdf


If you look at my cut quality then the errors I see are really quite small. I wonder if I should replace my 1.8° motor with 0.9° or 0.45° motors.

Both 0.9° and 0.45° motors are nema23, but the 0.9° is around 70 euro and the 0.45° is around 250 . So a little big difference

[cvoinescu]

That's a big motor there! Specs look good to me... Maybe the driver isn't very good -- I have no experience with it at all, but I remember I read recently about someone with a Chinese driver with a Toshiba chip who got bad microstepping performance and then switched to the much smaller A4988 and it fixed things for him. But he was using NEMA17 motors and a 24 V PSU, so much smaller. Could you borrow a different driver from someone else?