HexastormCross scan error of new prism with 1 arcminutes parallelism is shown below.
A pixel in the image is 3 micron and the distance between lines is 111 microns.
Deviation angle of prism is
Orthogonal wobble is;
A deviation of approx 0.1 mm is observed at distance, d, of 30 mm from prism, which implies planes are not planar within 9.8 arcminutes.
New prisms reduce the cross scan error. The prisms are leveled within 1 micron prior to rotation and sides should be planar at 1 arc minutes (this is guaranteed by supplier)
To ascertain the motor is the cause, one could monitor the height deviation of the prism while rotating using a laser distance sensor.
Another option is to think about the results with the cylindical lenses.
If it is assumed the prism is perfectly planar, the refracted rays remain parallel with the incident rays. The second cylindrical lens maps all parallel rays with the same angle to the same point in the focal plane. Ergo it filters any issues with the motor.
Use of cylindrical lenses proved to be highly beneficial, which implies these problems exist.
If higher accuracy is desired it can be resolved by;
- using only a single facet
- reducing the distance between the prism and the exposure plane
- adding a cylindrical lens pair; this circularizes the laser bundle
and filters out problems caused by rotation.
I will look at the motor in the next post.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
Is the primary cause of the error repeatable polygon facet alignment errors or is it non-repeatable motor bearing flaws? Facet alignment errors can be compensated for with software while higher precision motor bearings will reduce the other.
Are you sure? yes | no
Thanks Gravis, conclusions are indeed a bit imcomplete so I updated the post. Motor bearings errors can also be reduced with cylindrical lenses see post
Are you sure? yes | no
It looks like you cut out the "price increases the BOM too much so it won't be included" part which is important.
Measuring the same line repeatedly should allow you to suss out what is repeatable and what is non-repeatable. If the measurements are done by electronic capture then a simple "percentage of occurrence" cut-off should be able to sort them for you.
Are you sure? yes | no
you r right... I should do more measurements, repeatability etc.. cylindrical lenses increase bom with 115 USD. This is the price of all the other components combined. You don't need the lenses to get a good feel for the technology. They actually only make sense if you would use the laser head in production. The technology is now getting near developer kit level, I am trying to iron out all the low hanging fruits.
Are you sure? yes | no