UV-Laser PCB Exposure Machine for Hobby Projects (Version II)
Vložit
- čas přidán 21. 09. 2023
- This video shows a home-made exposure machine for printed circuit boards, ideal for small amateur electronics projects. One goal was to use only cheap and easily available parts. It operates extremely quiet because the used stepper motor is driven by an analog sine/cosine constant current source. I hope this video will be an inspiration for your own hobby projects!
- Věda a technologie
Woah, what an awesome build. Nice work on the custom sine drive circuit and calibration!
Thanks :)
Very cool! I've spent years trying to accomplish the same thing, and I kept running into issues that could broadly be broken into two categories: laser dot size/astigmatism and X/Y axis linearity. It's rather difficult sourcing a relatively powerful UV laser diode that has a small enough dot size and doesn't have extreme variations in dot size over varying distance due to focus. I don't know, I just had trouble locating lasers that had a published dot size, it always was about wavelength and power output. I mainly tried two beam movement methods: a galvo mirror assembly intended for a show laser, and a hexagonal spinning mirror assembly intended for a laser printer, with the idea that it would provide a very consistent beam sweep rate and therefore position. Both had their issues, but if I had been able to find a proper UV laser with a small enough dot size that could be maintained across a reasonable distance, it would have gone considerably better.
Anyway, kudos on creating a very workable and inexpensive system!
Looks cool! Would be glad for a link to some documentation/github page! I have made a poor man's CNC from my old 3D printer, but it does not work well. This could be a better way imo!
yeah agreed looks cool and i now want to build one , but i dont need one 😂and if i would, id just get a fiber laser wich can make pcb without the messy etching process
Nice maschine i like too use a old SLA 3d printer but nice idea
You can use a magnetic encoder, something like AS5600. Spin it constantly, turn on and off the laser.
Indeed I thought about the AS5600 for calibration but I doubt that it is precise enough to measure down to 7.2deg/1024, which is the actual resolution. Spining constantly? Around 360deg? Would be a waste of time since on the way round only a fraction is needed to "paint" on the board, wires would wrap around the axis too.
@@gumpomat Maybe not a stepper but brushed/brushless dc. For wires needs to be some sort of sliding connection. What kind of UV sensitive material did you use?
Typically I use material from "Bungard" (I think it is common only in Germany) but I tried also some unknown brand for the experiments that was laying around for more than 20 years. Worked well, but the protective foil left some ugly glue on the edges.
Very nice machine! Doesn't the laser dot get bigger towards the sides? But the results look very nice!
The center distance from the laser to the table is 240 mm, to reach a deflection of 30 mm the motor axis has to turn ~7.2 degrees. At that angle the new calculated distance to the table is just 1.9 mm more, the dot remains small enough for the required precision. In fact I did not do a lot of calculations after the first experiments with thermal paper went surprisingly well.
Nice project great efforts. What about large double PCBs 20cm x 15cm as example with lot of details details who long it will take to accomplish
At 2:12 you can see in the control program that the 50x15mm test board takes about 60s to "print", approx. 8s per 1cm^2. Since the machine does simple scanning line by line (like an old CRT) it is independant from the level of detail. A 20cm x 15cm board would take around 40min, of course the machine needs to be bigger than mine for that. For larger boards I would think about a more powerful laser and increasing the scanning speed.
I think high power and more speed you can use high resolution LCD with UV backlight like "3d resin printers" take all PCB at one shot like modern PCB manufacturing use direct image exposer machine@@gumpomat
wow, how do you keep it focused!
Surprisingly, the focus in such a system is not a problem at all, once the lens is adjusted (using thermal paper), it just works fine.
I imagine on the sides of the board laser spot will be a bit out of focus? And because of the angle light comes it also change shape from circle to elipse. How do you deal with that?
Of course you are right, focus will not be the same throughout a line from left to right and the shape is oval even because a laser diode has different divergence angles for each axis. But reality proves that the errors are small enough to create structures as needed for SMD boards with 0.5 mm pad distance. Would be interesting to calculate the theoretical dimensions of the laser spot and the error (anyone out there who can help?) we can expect but I think it is difficult because not all needed data about the lens and the laser diode is available. However, even with another brand of laser and collimator used in my first machine before, I had the same success so focus and beam size does not seem to be a major problem for this type of usage.
@@gumpomat thanks for answer. yeah I see the result is good for small boards. I am just scared that let's say 10*10cm board will experience bigger distortion of a beam. you can try super wide board and make test pattern (usually different feature size pads \ lines) at the edges and at the center and measure them to see the impact. anyway the setup mechanically seems very simple (not sure maybe 0.9 degree stepper will improve something) and might be easy to replicate. that would be so much better that toner transfer pcb
Good afternoon. Really cool project! Please tell me, is there a freely available diagram and source code for replicating the device yourself?
Sorry, not at this moment. Putting it all together, with toolchains for the microcontroller part and for the control program, it would fill a small pocket book and I think most interested ones have their own, different preferences regarding compilers, microcontroller etc., connectivity (e.g. serial, sd-card, bluetooth... instead of USB). Also other stepper motors might need adaption (current, driver for the y-motor etc) and also for things like the ADC (I used an obsolete AD8303, just because I had one) most will have their individual preference so that a precise construction plan would most probably not fit for most hobbyists. :(
@@gumpomat I apologize for the possibly inaccurate translation (google translator). I was not interested in the source code of the firmware, but in the compiled ready-made file for the firmware. sd card, bluetooth, etc. - this is not the most important thing in such a project. The most important thing is that the project works, and this project is really interesting. According to the accuracy of DRV4988 stepper motor drivers,
DRV8825 is a very old article, and these drivers were primarily intended to reduce the cost of 3D printers and the size of the control board. But progress does not stand still, and drivers are constantly being improved. Trinamic has a huge selection, for example TMC2209, TMC5160. These drivers already have finer microstepping and higher precision.
It is thanks to this that they became almost silent. TMC5160 was used in the construction of a CNC machine on FluidNC firmware, together with an ESP32 microcontroller. The drivers really work smoothly and easily control NEMA23 motors with a current of 2.7 Amps. You can't hear the engines running at all. I think you need to take a closer look at them,
for this project, and if the drivers are suitable, then repeatability of the device will be easier.
What was the reasoning for one axis being linear, and one being rotation? Why not both rotation or both linear?
Well, the X and Y controls are not really needed for normal usage. So they came in for debugging at the beginning. The laser on the axis is more a "rotational" thing, so I stayed with rotational units, more easy to play with. While the Y axis was more convenient in mm since the Y-motor has to turn 2*360deg for 1mm. But however, decisions came on the fly. ;)
Wie rechnet man das jpeg(falle es überhaupt eins ist) in das Bewegungsmuster des Schrittmotors um? Das ist zunächst einmal unglaublich komplex. Sind das Vektoren, die Sie dann in Grad umrechnet werden, um den Schrittmotor auszulenken? Interessant, dass der Schrittmotor nicht die Genauigkeit hat oder liegt das am Modell?
How do you convert the jpeg into the movement pattern of the stepper motor? That's incredibly complex at first. Are these vectors that you then convert into degrees in order to deflect the stepper motor? Interesting that the stepper motor doesn't have the accuracy or is that due to the model?
youtube.com/@belgua-nh8rm (for English see below) Die Position und Bewegung des Schrittmotors wird über die Ströme durch die Spulen bestimmt. Durch eine Testbelichtung kann man dann eine grobe Tabelle erstellen, welcher Strom zu welcher Position passt. Eine feinere Tabelle lässt sich dann durch Interpolation z.B. mit Excel erstellen. Überträgt man die Tabelle in Array im Programm kann man den Motor mit einer Schleife und Lookup des nötigen Stroms in der Tabelle eine Zeile überstreichen lassen, z.b. in 2048 Schritten. Ist die Grafik ebenfalls 2048 Pixel breit, dann muss man nur noch testen, ob beim aktuellen Motorschritt (z.B. 500) auch das Pixel der Zeile in der Grafik (bei X=500) gesetzt ist. Entsprechend schaltet man den Laser bis zum nächsten Pixel ein oder aus. Nimmt man ein einfaches Grafikformat wie PGM (Export z.B. aus GIMP) lässt sich der Pixelwert durch Laden der Datei in ein Array und Berechnung des Bytes über die Koordinaten ermitteln. Ich bevorzuge PNG, da die Dateien kleiner sind. Der Zugriff ist komplexer aber es gibt fertigen Code dafür (z.B. auf GitHub). D.h. ich exportiere aus dem Platinenlayoutprogramm erstmal nach GIMP, skaliere dort auf 50 Pixel/mm und speichere dann in als PNG. Natürlich steckt der Teufel im Detail aber das Prinzip ist eigentlich gar nicht so kompliziert. In der Tat sind Schrittmotoren generell nicht so präzise, d.h. ein einzelner Schritt (z.B. 1.8 Grad) wird nie perfekt entsprechend des Stroms durchlaufen. Das ist aber eben auch ein untypischer Anwendungsfall, da Schrittmotoren meist eben mit Einzelschritten verwendet werden und nicht mit Mikroschritten, die einen Schritt z.B. nochmals in 1024 Mikroschritte unterteilt.
The position and movement of the steoper motor is controlled by the current through its coils. Making a test exposure it is possible to create a table that lists which current is needed for which position. A table with finer granularity can be calculated by using interpolation mathematics, e.g. in Excel. Exporting the fine table to the control program of the machine it is possible to move the motor over a line by iterating through a loop, looking up the coil current in the table and sending that current to the motor, e.g. in 2048 steps. When the graphics is 2048 pixels wide too, you just need to test if the motor step (e.g. 500) corresponds to a pixel set in the graphics (at X=500). Depending on the pixel brightness you can switch on or off the laser until the next point is reached. Using a simple graphics format like PGM (e.g by exporting from GIMP), the pixel values can be checked in the program after loading the PGM file into an array and calculating the array index by the given coordinates. I prefer PNG because the files are smaller. Decoding them is more complex but code that does that is readily available (see GitHub). That means I typically export from the PCB-CAD to GIMP, there I scale to 50 pixel/mm, saving it as PNG. Of course there are many tiny details in addition, but the general principle is not that complex. In fact stepper motors are generally not so precise when it comes to microstepping. Means a single step (e.g. 1.8 degrees) will never be moved through perfectly according to the current. This is a not so typical use case because stepper motors are usually operated in single steps and not by microsteps that divide a normal step by 1024 sub-steps.
@@gumpomat Danke für die Einsichten und Gratulation zum Projekt.
have software and detail ?
Sorry, no, not planned. I hope the presentation motivates the interested ones (kids, students,...) to be creative, not by presenting detailed plans to follow blindly. There are many ways to realize something like that, different motors, controllers, DACs, some already have their preferred microcontroller (ST, Atmel/Microchip, Raspberry Pico, or Arduino...), development environments, for the PC (or Linux) too... Finally the code is not that complex, toggling an I/O pin while sending a table of data to a DAC, might be easier to build with a toolchain one knows than reworking my (ugly) code specialized for my type of hardware. However, the most delicate details of the electronics (constant current amplifier, laser and driver) can be seen in the video, maybe that helps. ;)
nobody ask about height. what height must to be?
Maybe a physicist in the audience could comment on that? ;) In fact the 24 cm distance to the table was just spontaneously taken when looking at some small laser marking machines available on the market (they use expensive galvanometers for deflection, not really an option for a hobby project). The actual height turned out to be useful, so I did not try to make it smaller yet. The theory about collimation optics, beam profiles, divergence etc. is available but I even don't know the exact focal length of the lens and I do not trust the datasheet of the laser diode. So I worked more or less with experiments only.
@@gumpomat you actually have a brilliant mind. I would need more information, especially about the lens and the laser diode.
@@gumpomat Great job. Those kinds of unknowns keep slowing the development of my projects. Your practical approach might be the solution.
Hi this is Harvey from PCBWay reaching out, do you have any interest in working something together with us? Look forward to your reply
Data available over internet shows copper has best absorbance for infrared laser lights Rather than VU
So is you clever data as valid for UV-exposure as for copper etching?
@@MuellerNick I'm telling about direct ething on copper ,no ink not anything
@@Abubaida-wx1iu Not the subject of that video. -> useless comment.
@@MuellerNick don't drink wine
@@Abubaida-wx1iu I don't drink any alcohol, if that helps you with your derailed argumentation.