Ball screws vs. belts

[Paul Gallagher]

I'm starting to design my machine. I'm sure this is a big debate. To get maximum accuracy, ball screws are required. I've found with my small machine, that as it gets warm, the positions drift too much due to belt stretch.

Has anyone successfully used small, fast lead ball screws on a PNP? Stepper motors max out at about 1000rpm, so it I think servo motors are necessary if using ball screws.

Or maybe it possible to use belts, and to have the machine constantly self correcting with some fidicials and the top camera.

Belts are obviously way less expensive and easier to design around.

[Mark]

It has been discussed a few times already. Thermal expansion is usually not an issue for a pick and place machines, for various reasons. To sum it up:

1. The needed precision is merely ±0.1mm or so.
2. Continuous absolute precision is only needed over the small area of the PCB. Unless you are placing 0201 component on meter panels, thermal expansion is simply too small to matter.
   
3. For anything else, the whole machine thermally expands and contracts as one, including feeders, camera positions etc. and the belt is stretched along.
   

Furthermore, the belt would not typically be the culprit as it is stretches with the frame. I'm not sure but using ball screws could even be worse in terms of thermal expansion, if the machine construction mixes aluminium and steel.

So if you observe significant drift, it is usually something different. Some have reported camera drift, for instance, some thermal tension on the sensor mount perhaps.

_Mark

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/2d916625-3fac-41ac-a8aa-481a6a8cd6c4n%40googlegroups.com.

[Paul Gallagher]

If I had the ability to pick a fiducial next to each PCB, it would help with the accuracy issues. And yes you might be right the belt is not the issue. The machine I'm using is pretty flimsy. So any temperature change could certainly cause some warping and twisting and throw everything out of calibration. Ball screw expansion is a known issue on CNC machines. They often have cooling systems and also the best machines rely on glass scales or lasers, and not the screw for accuracy. I would really prefer to use belts, since the design is 5x easier.

[bert shivaan]

you are correct, much debate over this.

in the end, Mark is correct IMHO.

commercial machines have been using belts for a VERY long time. The difference is they do not use putry 6mm wide GT2.

But you do not need huge belts either. 9mm wide GT2 is MORE than enough in most cases. One huge drawback to belts is they "bounce" This is worsened when the lengths are longer. This does not affect accuracy, only speed in the form of settling time.

Design away using belts, enjoy your machine, never look back :)

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/e2bac44c-6c0c-4df0-88c8-5e199a571ffcn%40googlegroups.com.

[Paul Gallagher]

I was wondering about that. From the videos of some commercial machines, even Siplace, it appears they use belts. If its good enough for them, its good enough for me. As for the bounce, probably some decel would help. And yes, I was thinking should use larger than normal 3D printer belts. The belts in my machine now are 15mm. Lots of advantages to belts other than cost, including light weight for better speed.

From what I have experienced, the 2 biggest factors for success are 1) PERFECT solder printing. 2) reliable part feeders.

Up to 0.1mm placement error is no problem at all if the solder printing is precise.

I build my own semi-auto solder printing machine. Its been incredibly precise. I had used the table top hinge type, but they are all crap.

[Jim Young]

@pdgall - Regarding your solder printing machine, did you publish your design anywhere?

[Paul Gallagher]

I have not. Its very simple. All 4080 and 2040 aluminum, some linear slides, and 3D printed parts. But I did use the jig plate and adjusters from one of my Chinese table top machines.

Here's a picture of it. If anyone is interested, I can send the Solidworks files. Its not a 100% complete assembly model. For example, I didn't bother to model the joining brackets.

The plan is some day I will add an automatic squeegee to the top, but not a high priority.

[Jim Young]

Interesting, for sure. I would be interested in the design files. How does it work and why is it so tall? Does the PCB fit in the recess on the top plate?

[Paul Gallagher]

Its 1meter tall. Sits on the floor. The PCB is placed on the carriage, and air cylinders move it up to the stencil.

[Paul Gallagher]

If I did it over, I'd make it a little taller. I used a relatively low weight 4080 aluminum, and its still stiff enough. Happy to package the assembly here if anyone is interested.

[Jim Young]

I would be interested in the design files. The SolidWorks files should work. I think I can import them into Fusion 360. Thanks!