# Laser focusing optics calculations for mirrors and lenses (for a Laser Direct Imaging machine)

Mike Gaffer
TL;DR Summary
Need to select optics path and parts for building a high resolution laser scanner on a 2D plane.
Objectives:
- best path for optics needed to focus and "draw"/project a high resolution image onto a workspace around 500mm square (for a Laser Direct Imaging machine)
- where to cost effectively purchase or make the optics necessary to build a prototype

The problem:
I need to project a high resolution image onto photoresist (photographic paper/film).
The film is sensitive to UV in the 300-440nm spectrum. Therefore, I was considering a Violet (405nm) or Blue (450nm) laser.
The film is used for PCB (circuit board) etching - I'm not sure of the laser power, or exposure time/passes with the laser, needed to expose, but not overexpose the film. I ordered film and a 1watt laser diode at 405nm (multimode) off Alibaba.
The area needs to scan 500mm x 500mm (larger would be better, but may be too expensive).
The focusing and scanning accuracy has to be 0.08mm or better, as in be able to draw lines accurately and repeatedly on the film being exposed.
I'm considering using a galvometer, like for a laser light show, and ordered a $90 kit off eBay to test. Questions: - How would I go about selecting the best optical path so I get a nice, tight, controllable, focused beam to form my scan or image? I notice the SLA 3D printer, "Form Labs" uses a laser, focused on a galvometer, bouncing off a large flat first surface mirror, and a parabolic (focusing?) mirror, before it is projected over a build area of around 300x300mm (below). - Is this complex path needed? - How would I go about doing these calculations? - where could I find or build a parabolic surface mirror like that if necessary on a budget? - Is the multimode laser okay for this? I also considered a DLP (digital light processor) chip and UV laser (like a laser video projector), however I dont think the resolution would be high enough. The highest resolution projection of light I need will be 0.09mm. So, if we assume 500mm @ 0.09mm resolution, this = ~5500 px, or 30,250,000 "pixels." - therefore I've ruled out a DLP chip with a UV laser, which typically runs 1920px x 1080px. I've also considered scanning in 2D, the laser beam on the X axis, and moving the workpiece on the Y, but I think this would add complexity. Can anyone point me in the right direction for design ideas? where to source parts? starting calculations or CAD programs/calculators? Thank you in advance! Last edited: ## Answers and Replies Science Advisor Gold Member For a decent quality parabolic mirror, try one from a decent quality reflector telescope; unfortunately the larger sizes are in the $$to$$,$$price range. They whole telescope may be adequate for a one-off build. For initial experimenting, lower cost "toy" telescopes from a science toy store may be useful. If you don't want to tear apart a telescope, here is a link to some of the better quality mirrors: https://www.edmundoptics.com/search/?criteria=parabolic mirror&Tab=Products# Cheers, Tom Science Advisor Will you draw vectors or use a raster scan ? 500 mm / 0.08 mm = 6250 elements = 13 bit resolution. How accurate can your galvanometer and mirror be ? How long will the mirror take to settle to the required resolution ? How long must each element be exposed for ? 6250 x 6250 = 39 Mega pixels. The time taken to draw the image will be very much longer than you expect. 2 msec * 39 M = 78,125 sec = 1302 minutes = 21.7 hours = all day. The system you show is only a 1D system, so the flat mirror does not obstruct the parabolic mirror. 2D is impossible without a diagonal optical path and a 2D galvanometer. Note that a 2D parabolic mirror is needed to keep the ray perpendicular to the target, so it must be the same size as the target. That would be circa 710 mm diameter for the 500x500 mm target. Consider a raster scan, using a 1D system, as the bridge of an XY router. You will still need the 1D laser printer optical path. Mike Gaffer For a decent quality parabolic mirror, try one from a decent quality reflector telescope; unfortunately the larger sizes are in the$$$ to $$,$$price range. They whole telescope may be adequate for a one-off build. For initial experimenting, lower cost "toy" telescopes from a science toy store may be useful. If you don't want to tear apart a telescope, here is a link to some of the better quality mirrors: https://www.edmundoptics.com/search/?criteria=parabolic mirror&Tab=Products# Cheers, Tom Hi Tom, Thanks for your reply. Edmund Optics is a resource I forgot about since college. Thanks! Also, I love the idea of disassembling a telescope for optics, if I cant find the exact element cost effectively. Thanks for the ideas. Ted Mike Gaffer Will you draw vectors or use a raster scan ? 500 mm / 0.08 mm = 6250 elements = 13 bit resolution. How accurate can your galvanometer and mirror be ? How long will the mirror take to settle to the required resolution ? How long must each element be exposed for ? 6250 x 6250 = 39 Mega pixels. The time taken to draw the image will be very much longer than you expect. 2 msec * 39 M = 78,125 sec = 1302 minutes = 21.7 hours = all day. The system you show is only a 1D system, so the flat mirror does not obstruct the parabolic mirror. 2D is impossible without a diagonal optical path and a 2D galvanometer. Note that a 2D parabolic mirror is needed to keep the ray perpendicular to the target, so it must be the same size as the target. That would be circa 710 mm diameter for the 500x500 mm target. Consider a raster scan, using a 1D system, as the bridge of an XY router. You will still need the 1D laser printer optical path. Hi Baluncore, I didn't think about beam stability with the galvanometers. I guess they cant perform a raster scan efficiently without quite a bit of "vibration settling." For the beam path I provided, it actually is in 2-D --- the actual printer I own, and it scans the bottom of a translucent plastic build tray in the exact pattern of each layer of a 3D part... then moves it upward and does the next layer. I do believe you are correct with using a 1D scanning method. Something like: Laser diode -> focusing lens -> spinning scanning octagonal mirror with an optical encoder on the motor -> then a diverging meniscus lens -> concave lens... or or and move the workpiece on a set of acme ball screws for the 2nd dimension, slowly forward. Is there any good way of calculating what last 2 lenses I need? Is there like an Ansys of optics? or a SolidWorks focused on optics? Thanks for your thoughts! Ted Science Advisor Gold Member A source of Laser scanning motor. mirror, optics is a replacement scanner for a Laser Printer. However there tends to be no documention available for them, so you are kinda on your own for a drive circuit. If you can get a service manual for the printer you can likely find a little info though. Science Advisor @Mike Gaffer If you are going to do a raster print, in the style of a laser printer, avoid moving a bridge or gantry over a fixed platen. Instead, move the platen holding the PCB, on rails under a solid fixed bridge. It will require twice the floor-space, but will be very much more rigid, and so will remain aligned and repeatable. The position of the platen could be measured using a common 5 um resolution glass scale, as would be used on a machine tool with a DRO. (150 on eBay). Those scales produce a quadrature position signal. I would move the high-inertia platen at a fixed speed.

The fixed bridge would carry two lines of staggered parallel optical systems, taken from a common model of laser printer. Cables and power to the optics will be fixed to the rigid bridge, so will not need to flex.

You will need to map and control the boundary of each optical system using software. Semiconductors designed to control common laser printers can be used, since each optical module will be printing only part of the optical scan width. Since the platen is moving at the same time as the scan, the optical modules will need to be mounted one pixel diagonal on the bridge, but the bridge will need to be perpendicular to the platen rails. By using software mapping, it may be possible to get away without an autocollimator when setting the optical systems on the bridge.

To be laser wavelength independent, avoid lenses, use front-face mirrors for the polygon and optical system, except as fitted to the fixed laser module.

Notice that the curvature of the parabolic mirror will to some extent focus the laser along the optical-scan dimension, while the straight mirror is actually made slightly cylindrical, to focus the laser in the platen-movement dimension.

berkeman