Unterschiede

Hier werden die Unterschiede zwischen zwei Versionen angezeigt.

--- projekte:dvdlsm:start [2021/02/07 00:04] – thasti
+++ projekte:dvdlsm:start [2021/02/07 22:25] (aktuell) – [Laser Diodes, Lens and Actuators] thasti
@@ Zeile 1: / Zeile 1: @@
 ~~NOTOC~~
-====== DVD-Drive-Based Laser Scanning Microscope ======
+====== DVD-Drive based Laser Scanning Microscope ======
-{{ :projekte:dvdlsm:dvdlsm_full.jpg?400|}}
+{{ :projekte:dvdlsm:dvdlsm_full.jpg?450|}}
 Due to an abundance of one specific type of DVD drives, plans were made to give a useful afterlife to these devices. A laser scanning microscope seemed to be the most useful project.
@@ Zeile 28: / Zeile 28: @@
 {{:projekte:dvdlsm:opu_disassembled.jpg?500|}}
-==== Laser Diodes ====
+==== Laser Diodes, Lens and Actuators ====
-The OPU houses two laser diodes. As this is just a DVD/CD reader, the lasers are not very powerful but provide sufficient power for microscopic applications. The (red) DVD diode starts laser operation above approx. 60 mA and requires about 2-2.5 V of forward voltage. Its polarity could be easily determined by measurements.
+The OPU houses two laser diodes. As this is just a DVD/CD reader, the lasers are not very powerful but provide sufficient power for microscopy applications. The (red) DVD diode starts laser operation above approx. 60 mA and requires about 2-2.5 V of forward voltage. Its polarity can be easily determined by measurements.
+The primary lens is positioned using electromagnetic actuators. These come in the form of three coils, which tilt or move the coil proportionally to the current flowing though them. Their polarity and required full-scale DC actuation current can be easily found by using a lab power supply.
+The focusing control loop driving these coils is normally deriving its error signal from the astigmatic focus error detection method. It requires (in its basic form) a four-quadrant diode, and focus is optimized when all quadrants of the diode are equally illuminated. Deviation from this point will cause the circular illumiation to become elliptical and therefore illuminate two diagonal quadrants more strongly. A schematic explanation of that scheme can be found [[https://www.researchgate.net/figure/Astigmatic-detection-method-a-Schematic-diagram-of-the-optical-system-with-commercial_fig1_329138728|here]] and in many textbooks.
-==== Lens Actuators ====
-The primary lens is positioned using electromagnetic actuators. These come in the form of three coils, which tilt or move the coil proportionally to the current flowing though them. Their polarity and required full-scale DC current can be easily found by using a lab power supply.
 ==== PDIC ====
-The PDIC (Photo Diode Integrated Circuit), in a DVD reader application, serves multiple purposes. These include gathering focus, tilt and tracking information from the disk as well as recovery of the RF data stored on the disk tracks. For this reason, PDICs typically consist of at least 6 active diode areas, sometimes even more. The PDIC used in the available drive type is seen below. It was placed on a little break-out PCB to allow easy access to all pins of the device.
+The PDIC (Photo Diode Integrated Circuit), in a DVD reader application, serves multiple purposes. These include gathering focus, tilt and tracking information from the disk as well as recovery of the RF data stored on the disk tracks. For this reason, PDICs typically consist of at least 6 active diode areas, sometimes even more. The PDIC used in the available drive type is seen below. Given its a very niche device, oftentimes custom-designed for a specific drive vendor, datasheets are hard to come by. Initial pinout probing revealed only the power supply pin locations and a number of analog output pins, but I was unable to find any matching ICs datasheets, so some reverse engineering effort was required to figure out its secrets.
+First, the device was placed on a little break-out PCB to allow easy access to all of its pins, especially during microscope inspections.
 {{:projekte:dvdlsm:pdic_breakout.jpg?600|}}
-Initial pinout probing revealed only the power supply pins and a number of analog output voltages, however no datasheet for a PDIC with matching pinout could be found online. To be able to re-use this component, a little reverse-engineering effort was started.
 === PDIC Reverse Engineering: Optical ===
-Using an IC inspection microscope, a die shot of the PDIC was produced. This could be used to get a better idea of the internal structure of the device. This was made way easier by the fact that it is an optical integrated circuit, and is therefore housed inside a clear plastic package with good optical properties, suitable for direct microscope inspection without decapsulation. The picture below could be used to identify some functional blocks of the circuit, however full tracing of the bondpads through internal circuits to the diode quadrants was considered unfeasible.  However, it became apparent that the PDIC seemingly houses at least independent 8 photo diode channels. (Four quadrants in the center square, and at least two segments for each of the outer squares).
+Using an IC inspection microscope, a die shot of the PDIC was produced. This was useful to get a better idea of the internal structure of the device. This process is made way easier by the fact that it is an optical integrated circuit, and is therefore housed inside a clear plastic package with good optical properties, suitable for direct microscope inspection without decapsulation. The picture below could be used to identify some functional blocks of the circuit, however full tracing from the bondpads through internal circuits to the diode quadrants didn't really go anywhere for me. However, it became apparent that the PDIC apparently houses at least independent eight photo diode channels. (Four quadrants in the center square, and at least two segments for each of the outer squares).
 {{:projekte:dvdlsm:chip_res.jpg?600|}}
@@ Zeile 48: / Zeile 51: @@
 === PDIC Reverse Engineering: Electro-Optical ===
+On most of the pins, some change in output voltage could be seen when illuminating the device with a bright source of light. At first, the output behavior of each pin was simply put into different categories:
-To facilitate Pinout reverse-engineering of the IC, a simple test fixture was soldered (see above). This gave easy access to any of the pins. On most of the pins, some effect to applying direct lighting could be identified. Initially, the output behavior of each pin was simply put into different categories:
   * Low light sensitivity: 4 outputs
   * High light sensitivity: 4 outputs
-  * Same magnitude, but opposite polarity outputs: 2 pins
+  * Same magnitude, but opposite polarity outputs (differential pair): 2 pins
   * Negative output voltage: 1 pin
 It was conjectured that the smaller inner quadrants were responsible for the low-sensitivity outputs, while the outer segments would produce a large amplitude (due to their larger light collection area).
-In a second stage, the exact pinout was determined by means of selective illumination: For any given test, two IC output pins were compared using two DVMs, and an aperture was slowly moved along the surface, either vertically or horizontally. Normally this would result in reduction of the output voltage for one of the outputs earlier than for the other one. In this way, the positional relationships between all output pins could be resolved, forming a final assignment of diode quadrants to output pins.
+In a second reverse engineering session, the exact pinout was determined by means of selective illumination: For a given test, two IC output pins were compared using two DVMs, and an aperture was slowly moved along the surface, either vertically or horizontally. This would in result a change of the output voltage for one of the outputs slightly earlier than for the other one. In this way, the positional relationships between all output pins could be resolved, forming a final assignment of diode quadrants to output pins.
 {{:projekte:dvdlsm:pinout_bg.png?800|}}
-The output producing negative voltages is not considered to be useful in the final application, while the differential output pair is likely a summed output of the four inner quadrants (the RF output). With the assignment of pins to diode quadrants done, the OPU could be modified to fan-out the required connections to custom electronics
+The output producing negative voltages is not considered to be useful in the final application, while the differential output pair is likely a summed output of the four inner quadrants (the high bandwidth RF output). With the assignment of pins to diode quadrants done, the OPU could be modified to fan-out the required connections to custom electronics
 ==== OPU Modification ====
-A breadboard header was added to the OPU, fanning out all required signals to the custom electronics on a ribbon cable. The analog output signals are separated on one side, while the current inputs for the laser diode and the lens actuators are kept to the other side. This should help mitigate electrical coupling between these signals, even though the diode outputs are actively driven by the PDIC.
+The whole top assembly of the DVD drive is reused. This includes not only the OPU, but also the linear (worm gear) axis driven by a stepper motor and its base plate. A breadboard header was added to the OPU, fanning out all required signals to the custom electronics on a ribbon cable. The analog output signals are separated on one side, while the current inputs for the laser diode and the lens actuators are kept to the other side. This should help mitigate electrical coupling between these signals, even though the diode outputs are actively driven by the PDIC.
+The other axis (below the drive mechanism) is built from a commercial precision linear actuator that was donated to the project. Other people have simply used another DVD drive in its place, which would certainly be a possibility here.
 {{:projekte:dvdlsm:dvdlsm_head.jpg?600|}}
@@ Zeile 77: / Zeile 81: @@
 It was pretty obvious that at least all the mechanical actuation and laser driving could be handled by an IC foreseen for an optical drive. Texas Instruments offers a line-up of such chips, which are one of the few types that can easily be sourced from common distributors. The [[https://www.ti.com/product/TPIC2050|TI TPIC2050]] was chosen due to its all-digital interface and flexible output options. Given it's a pretty high-quantity component, its cost is very reasonable (€2.50 in single quantities) and can hardly be beaten by any other integrated options.
-A PCB was designed around this IC, with an additional 16 bit ADC plus analog MUX for the PDIC outputs. The PCB was built to house an [[https://www.crowdsupply.com/1bitsquared/icebreaker-fpga|iceBreaker FPGA board]] that would take care of all the interfacing.
+A PCB was designed around this IC, with an additional 16 bit ADC plus analog MUX for the PDIC outputs. The PCB was built to house an [[https://www.crowdsupply.com/1bitsquared/icebreaker-fpga|iceBreaker FPGA board]] that would take care of all the interfacing. All electronics were designed in KiCad and are available publicly in its [[https://git.loetlabor-jena.de/dvdlaser/dvdlaser_hw|GitLab repository]].
 {{:projekte:dvdlsm:dvdlsm_pcb_unpopulated.jpg?700|}}
@@ Zeile 83: / Zeile 87: @@
 {{:projekte:dvdlsm:dvdlsm_pcb_res.jpg?700|}}
-The PCB was designed in KiCad and is available publicly in its [[https://git.loetlabor-jena.de/dvdlaser/dvdlaser_hw|GitLab repository]].
 ===== FPGA Design =====
 The FPGA had to fulfill a couple of interfacing tasks, and the design uses the FOSS FPGA toolchain for the Lattice ice40 UP5k FPGA on the iceBreaker board. It mainly does the following things:
@@ Zeile 109: / Zeile 112: @@
 The main laser microscope class provides a few methods for focus control and scanning a predefined area to an image file. The focusing is the most interesting aspect here. Generally, the focusing first performs a wide scan of the Z-height to find the linear region of the focus error detector. Following that, it implements a simple PI control scheme for successively minimizing the focusing error.
-Given that relatively large areas should be possible, focus tracking is required. Initially, the software uses a relatively simple algorithm to achieve this:
+As relatively large scan areas should be possible, focus tracking during the scan is required. Initially, the software uses a relatively simple algorithm to achieve this:
-  * Given a rectangular scan area, place a virtual grid inside this area with at least N points per millimeter
+  * Inside the rectangular scan area, place a virtual grid with at least N points per millimetre
-  * At each grid point, perform a focusing and store the focussed Z-height
+  * At each grid point, perform one focus acquisition and store the focused Z-height
   * Least-Squares-Fit a 3D plane through all the points on the grid
-  * During the actual scan, calculate the optimal Z-height for a given position based on the Z-height of this plane
+  * During the actual scan, calculate the optimal Z-height for a given position by evaluating the Z-Coordinate of this plane
 Another noteworthy aspect is the Y-Axis (implemented using the DVD drive motor and mechanism). Given this whole mechanism is a made from a few cheap plastic pieces,
@@ Zeile 145: / Zeile 148: @@
-Higher resoultion scan of the same device EPROM, scan area 500x500 um. The two routing layers can be seen above one another. (Picture Below)
+Higher resolution scan of the same device EPROM, scan area 500x500 um (top right of previous image). The two routing layers can be seen above one another. (Picture Below)
 {{:projekte:dvdlsm:eprom_0p5x0p5mm.png?600|}}
-Scan of an I2C test chip from my colleage Szymon (fabricated in 65nm CMOS, picture below).
+Scan of an I2C test chip from my colleague Szymon (fabricated in 65nm CMOS, picture below). Scan area 1x1mm.
 {{:projekte:dvdlsm:i2chip_first_scan_dt.jpg?600|}}
-This picture shows some of the aforementioned nonlinearity in one axis, which introduces periodical distortions in the image. These could be corrected in principle, but I didn't bother. The nice CERN logo was placed on the top metal layer, and is composed of many small pixels that are on the order of a few micrometers large.
+This picture shows some of the aforementioned nonlinearity in one axis, which introduces periodical distortions in the image. These could be corrected in principle, but I didn't bother. The nice CERN logo was placed on the top metal layer, and some of the lower level metals are actually already visible from the top (the vertical and horizontal power distribution stripes are placed on the two metal layers below).
 Also from this chip, here's a few very small scans performed at 200 nm step size. Small metal filling patterns on the top metal layer are imaged, each of which is only about a two micron wide square. Lithographic imperferfections can clearly resolved using this imaging technique. The horizontal (black) lines were caused by the erratic X-axis stage, sometimes going to places it was not told to go...
@@ Zeile 161: / Zeile 164: @@
+Finally, here are two videos of the device in Action:
+  * Focusing: https://youtu.be/iFzVEHeG0xE
+  * Scanning: https://youtu.be/cj5zdhJTQUs
+===== Summary =====
+For not a lot of money, a rather capable little laser microscope could be built. At some point, more scanned objects will be added to the collection above. If you are interested in replicating this project or a part of it, all developments are publicly available in the following repositories:
+  * Hardware: https://git.loetlabor-jena.de/dvdlaser/dvdlaser_hw
+  * FPGA Gateware: https://git.loetlabor-jena.de/dvdlaser/dvdlaser_fpga
+  * PC software: https://git.loetlabor-jena.de/dvdlaser/dvdlaser_sw
+===== References =====
+I'm of course not the first one to build something like this. Here's some sources of my own inspiration:
+  * http://www.gaudi.ch/GaudiLabs/?page_id=652
+  * https://www.instructables.com/Laser-Scanning-Microscope/
+  * https://pubs.acs.org/doi/10.1021/acssensors.8b00340
+  * https://www.researchgate.net/publication/321233124_Generating_SEL_and_SEU_with_a_class_1_laser_setup