Open hardware fast high resolution LASER

From RepRap
Revision as of 06:01, 1 December 2016 by Repraponion2 (talk | contribs) (some improvements)
Jump to: navigation, search

This article article describes a transparent polygon laser diode scanner. The article aims to make room for an open-hardware version.

Description

Commercial laser diode sources can provide light at wavelengths down to 375 nm. The maximum power for single mode commercial laser diodes of 405 nm and 375 nm is 350 mW (SLD3237VF) and 70 mW (NDU4116) respectively. The SLD3237VF is used in a 4x blue ray writer. Blue ray writers with 16x use stronger laser diodes, but no official data sheets are known (1 watt is typically claimed). At short wavelengths, laser diodes cannot give enough power to solidify large substrates fast. This is critical for a high waver throughput in printed circuit board manufacturing. Also a lot of energy is needed, for large area ceramic photo-polymerization printers. The critical energy dosage for 60 volume percent silicon oxide and 55 volume percent alumina is 60-400 mJ/cm2 and 500-3300mJ/cm2, see Halloran. As a result, it can be advantageous to combine several laser diodes to solidify a single layer. This article tries to present a cost-effective illumination system which is able to combine the power of several laser diodes. Light can have up to two polarizations, as a result up to two single mode laser diodes can be combined into one bundle. A beam splitter cube is typically used to combine the laser diode bundles. In the figures shown to outline the chosen solution, a single laser diode is focused directly via an aspherical lens. This results into an elliptical spot. A spherical spot can be achieved by first collimating the laser diode with an aspherical lens and then circularizing the bundle with an anamorphic prism pair. The bundle can hereafter be focused with for example an achromatic doublet. The polygon can be made by polishing the sides of a quartz sheet. Key is to realize that in reflective polygon scanners it is advantageous to hit the polygon with a large collimated spot. In transparent polygon scanners, the beam is focused and the spot can be smaller; which keeps the disk light. The bearing of the scanner can be created via a ball bearing, an air bearing or a self-sustaining air bearing. In the current solution, a self-sustaining air bearing is chosen as it is thought to be cheaper. After focusing, the laser diode bundle is displaced by rotating a transparent plate. The transparent plate is formed by the opposite sides of a regular convex polygon with an even number of sides. The illumination head is moved perpendicular to the scanning direction. As a result, the substrate can be solidified. Another solution would be to move the substrate relative to the now static illumination head. This could be desirable as self-sustaining air bearings with a high speed rotating disk are best kept in a static position. In figure 1 A, a possible embodiment is shown. In figure 1 A, a polygon cell consists out of three transparent polygons. Two polygon cells are shown in figure 1 A. More polygon cells can be added to the light engine if the exposure area needs to be enlarged.

Figure 1 a.

Another embodiment, where one of the polygons in a cell is rotated is shown in figure 1 B. Cells can be made out of; 2, 4 or more polygons. As can be seen in figure 1 B, the polygon may be rotated by 180 degrees.

Figure 1 b.

Key is that the transparent polygons project lines which overlap partially, if the illumination unit or substrate is moved orthogonal to the illumination direction on the substrate. Both embodiment's shown can be used to solidify six lanes. These lanes overlap partially. The overlap is required to allow for correction of alignment errors in the optical design. Four lanes and three overlaps are shown in figure 1C.

Figure 1 c.

The alignment errors can be detected by a camera which is moved under the illumination unit in the setup. Projection should be done directly on the CCD chip of the camera. A gray filter can be placed on the camera to protect the CCD chip. The errors can then be corrected for in a slicer based on, for example, the Visualization ToolKit and a spot detection module based on OpenCV. Naturally, these libraries come with a convenient Python interface.

Search Report

There are several other laser diode polygon scanners which can be used to combine the power of multiple laser diodes. KLEO Halbleitertechnik which was partly owned by Zeiss and later sold to Manz sells the Speedlight 2D. The Speedlight 2D is a system which uses 9 polygons and 298 laser diodes to solidify a substrate with a width of 600 mm. The reflective polygon has 32 facets and rotates at a speed of 50.000 rotations per minute, see patent US8314921B2. The Netherlands Organization for Applied Scientific Research (TNO) has done a lot of research in transparent polygon scanners. It developed a system which is similar to that of Kleo AG but uses a transparent instead of a reflective polygon, for details see patent application WO 2015/160252 A1. The transparent polygon scanner proposed in this article is different than that from TNO in that it uses a single and not multiple optical bundles per transparent polygon. In addition, the rotation axes of the the proposed system are orthogonal and not co-planar with the substrate. As said, the single optical bundle can be made with up to two laser diodes. It might be beneficial, to use laser diodes with larger wavelengths. These wavelengths could be used to sinter powders or metals. Another possibility would be to solidify egg whites with infrared radiation in a food printer. Explosives, piezoceramic materials and metals can also be fabricated via photo-polymerization. TODO: identify patent base of Ricoh's air bearing,

Business Case

Competitor Analysis

High end large-photo polymerization machines or PCB machines tend to be closed source and start at 250K euro’s. Companies operate in this business segment with a large patent base. For the business case, the following machines have been taken into account. In the photo-polymerization market competing machines are the PromakerL7000D (Prodways), ProX 950 (3D Systems) and XEDE 3SP(Envisiontec). Formlabs has been omitted as it uses a colimated bundle, i.e. the laser diode is not focused. The number of pixels offered by DMD printers like the Rapidshape or Carbon 3D is seen to be too small even with a 4K DMD. In the PCB market competing machines are the Nuvogo Fine 10 (Orbotech), Speedlight 2D (Manz) and the Ledia 3WL (Ucamco).

Competing Technologies

The following technologies can be distinguished;

  • Polygon scanner with F-theta lens and one laser bundle
    • Used by: Envisiontec, Orbotech
    • Limits: Three element F-theta lenses are expensive, telecentric systems are even more expensive and can, unlike transparent polygon scanner, not project outside the lens. For high-resolution systems, it can be beneficial to use a thick polygon, for a supplier see Lincoln Laser. An example calculation is given here. Transparent polygons can be thinner as the bundle is already focused.
  • Reflective polygon scanner with multiple laser bundles
    • Used by: Manz
    • Limits: the polygon tilt angle is smaller than 45 degrees, most likely costs 1 million euro's
  • Transparent polygon scanner with multiple laser bundles
    • Used by: LDI Systems (defunct)
    • Advantages: telecentric projection, system can project telecentric outside the lens, polygon tilt angle can be 45 degrees
    • Limits: at best proof of concept, light engine in the order of 1K euro's per mm of projection
  • DMD chip illuminated with LEDs
    • Used by: Ucamco and Prodways
    • Advantages: PCBs can be illuminated with multiple wavelengths which can be advantageous for PCB manufacturing, DMD is able to project several pixels at once, laser scanners are not able and therefore in the photo-polymerized lines are visible.
    • Limits: multiple beamers can be placed adjacent to each other but this expensive, as a result Prodways translates the beamer and illuminates a 45 degrees mirror, if the mirror is illuminated with laser diodes this can result into multiple-slit interference.
  • Galvanometer scanner with Nd:YAG LASER
    • Used by: 3D Systems, Materialise
    • Limits: low power and frequency of Nd:YAG laser, due to inertia galvanometer scanners are slower than polygon scanners

In the PCB market only Kleo uses laser diodes at 405 nm. The system should be designed such that it is a lot cheaper. The Grating Light Valve, sold by companies like Silicon Light Machines, is typically used for mask-less lithography with 2.5 micrometer features. Too few applications of the MEMS scanner developed by Fraunhofer were known to take it into consideration. The femtosecond laser, which can be used in two-photon polymerization to focus light in space and time and trigger a non-linear reaction, was seen as too expensive.

Competing Light Sources

The following light sources have been considered;

  • Light-Emitting Diode (LED)
    • Used by: Ucamco and Prodways
    • Wavelengths: 405, 395, 385, 374, 365
    • Frequency: set by other element in the optical path, e.g. the refresh rate of the DMD chip
    • Power:<4 watt
    • Price: 5 euro's per LED
    • Note: LEDs offer less contrast and depth of field than laser diodes but can easily be combined as they due their non incoherence interfere. Texas instrument seems to have a monopoly on DMD chips. Projection systems are sold by other vendors; for example, the LUXBEAM Lithography System sold by Visitech for >10k euro's. DMD chips can handle less optical power at shorter wavelengths. For wavelengths below 405 nm, the power limit is currently 4 watts per chip DLP9000UV.
  • Laser Diode (LD)
    • Used by: Manz, Envisiontec
    • Wavelengths: 405, 395, 375 nm
    • Frequency: 50 MHZ
    • Price: 40 euro's at 405 nm, 3870 euro's at 375 nm
    • Power: 0.4 watt at 405 nm, 70 mW at 375 nm
    • Cooling: most laser diodes do not operate well at temperatures above 30 degrees, but some can handle 80 degrees.
  • Diode-Pumped Solid State Laser (DPSSL)
    • Used by: Orbotech
    • Wavelength: 355 nm
    • Frequency: 80 MHZ
    • Power: 24 watt
    • Price: 190k euro's (only laser, you also need a chiller and a power supply)
    • Vendor: Coherent
    • Sizes: LASER 305 x 200 x 1100 mm, power supply 482 x 177 x 505 mm,
  • Nd:YAG LASER
    • Used by: 3D systems and Materialise
    • Wavelength: 355 nm
    • Power: 1 watt
    • Frequency: < 1MHZ

Proof of Concept

To understand the business case, please use the above to realize that competing systems are expensive and that not a lot of systems use laser diodes. For a viable business case, the price per transparent polygon and laser diode is key. As a result, only a single transparent polygon scanner has to be built. Firstly, it has to be shown that a single quartz sheet can be rotated at high speeds at low costs. For the proof of concept, the Ricoh Aficio AF-1027/270 can be used. The reflective polygon can be removed from the polygon and after flattening a transparent quartz sheet can be placed on top. This quartz sheet should weigh less than 8 grams. Key is to check whether the transparent sheet can rotate stable at high speeds, e.g. 20.000 rotations per minute. If this has been shown, the price per polygon is known and the price per diode and throughput can be calculated. At the moment, the price per transparent polygon unit is thought to be 300. (edge polished quartz sheets are not readily available and in the order of 150 dollars). The price for a light engine with 80 diodes would be something like 24K euro's. If the rest of the machine costs 30K euro's, the total price is estimated at 54K euro's . With a 70 percent margin it could be sold at 180K euro's, which would allow it to compete with commercial machines in the proposed business segment in the 3D printing with a 50 micrometers spot size. For the PCB market, the resolution target is 10 micrometers. This market is best targeted after 3D printing. From an investment point of view, the technology can be interesting as it can be funded in multiple rounds. The transparent polygon is build in the first round. This will be cheap (i.e. 10K) but have a high risk. In the second round, a complete engine can be sponsored (e.g. 50K euro's). Finally, in the third round a proof of concept tool could be built. This would cost something like 100K euro's. Only R&D costs have been accounted for in these investment rounds. The fact that the laser diode spot can be kept into position with a laser diode driver, photo-diode and a Field Programmable Gate Array (FGPA) has already been shown for several reflective polygon scanners. It would, however, be nice to make an open source laser diode driver based on (MyHDL) and for example, the Lattice iCE40, used in project (Icestorm).

Physics

In the following, an analytical description of the system is given. The equations have been divided into the following categories: polygon properties, spot properties, transparent parallel plate and system design parameters. The section starts with a parameter definition.

Parameter Definition

The polygon rotates about its center, i.e. the point inside the polygon that is equidistant from each vertex. The substrate moves under the polygon in a certain direction. The smallest angle within the plane of the substrate between the center line of the polygon and the line parallel to the movement of the substrate is defined as the polygonal tilt angle.

  • <math>\alpha</math> denotes the polygonal tilt angle. In the proposed system, this is 90 degrees.
  • T defines the thickness of the polygon, T is equal to 2r.
  • r defines the inradius of the polygon
  • a defines the polygon side length
  • R defines the circumradius of the polygon
  • v is the number of vertices of the polygon.
  • n is the refractive index of the polygon, typically BK7 glass is used
  • d is the diameter of the aspherical lens
  • <math>\lambda</math> defines the center wavelenght of the laser diode beam

Polygon Properties

Figure 2.

In figure 2, a regular convex polygon is shown; here r is the inradius, R is the polygon circumradius and a is the polygon side length, v denotes the number of vertices. In figure 2, v is equal to 8. Earlier, we defined r to be equal to T. The number of facets of the polygon has to be even for opposing planes to be parallel. If the number of facets is uneven, there will be an edge crossing during illumination which is sub optimal.

  • <math>a=T\cdot tan(\pi/v)</math>
  • <math>R=\dfrac{a}{2 \cdot sin(\pi/v)}</math>
  • The interior angle of a simple polygon with v vertices is <math>180-360/v</math> degrees.

The maximum angle of incidence for an octagon is 67.5 degrees.

Spot Properties

For a collimated and hereafter focused bundle the spot size is <math>w_0=\dfrac{4\lambda}{\pi}\dfrac{f_{efl}}{D}</math>. The Rayleigh range is given by <math>z_r=\dfrac{\pi w_0^2}{\lambda}</math> The Rayleigh length or Rayleigh range is the distance along the propagation direction of a beam from the waist to the place where the area of the cross section is doubled. Typically, the beam is assumed to be within focus in the Rayleigh range.

Transparent Parallel Plate

A transparent parallel plate is typically used to transversely shift a collimated bundle. For a converging beam, a parallel plate will also give a longitudinal focus point displacement away from the lens and optical aberrations. The idea is to maximize the tilt angle of the parallel plate, while still keeping the aberrations below some acceptable limit, or equivalently the Strehl radius above some acceptable limit.

Displacement

  • longitudinal displacement is <math>\dfrac{n-1}{n}T</math>
  • transversal displacement defined as D is <math>D=T cos I(tan I-tan I^')</math> which can also be written as

<math>D=T sin I(1-\sqrt{\dfrac{1-sin^2 I}{n^2-sin^2 I}})</math>.

  • The spot speed can be derived by differentiation of D with respect to I. The speed at the center is smaller than the speed at the edges of a projected line. As a result, the amplitude at the center should be smaller and is ideally corrected for by the laser diode driver by varying the pulse frequency or current.

Seidel Aberrations

The aberrations are calculated up to third order, it is known from literature and experiment that these can be important. The Seidel aberrations of a parallel plate are given by Wyant. In the following, the f-number or <math>f_\#</math> equals <math>f_\#=\dfrac{f_{efl}}{D}</math>. A transparent plate does not have a Petzval field curvature aberration. As a result, the projection of the transparent polygon is telecentric. The other aberrations are:

  • spherical wavefront abberation is <math>=-\dfrac{T}{f_\#^4}\dfrac{n^2-1}{128n^3}</math>
  • coma is <math>=-\dfrac{TU}{f_\#^3}\dfrac{n^2-1}{16n^3}cos\theta</math>
  • astigmatism <math>=-\dfrac{TU}{f_\#^2}\dfrac{n^2-1}{8n^2}cos^2\theta</math>

The wavefront aberrations can be used to calculate the Strehl ratio. If the Strehl ratio gets below a limit during illumination, the aberrations will become dominant and the system will not image properly. As a result, it must be ensured via calculation that the Strehl ratio is larger than some acceptable limit, e.g. the Rayleigh limit of 0.71. Systems with a spot size close to 10 micrometers are possible. The main challenge is the placement of the optical components.

Simulation

The calculation can be verified with an open optical ray tracing and lens design framework, such as the Python library made by Dr. Jordens named Rayopt.