วีดีโอ

I plan to 🙂... hopefully I can sell them in Q3 or Q4

I use a FPGA.. You could use a microcontroller but you need to guarantee that you can loop say at 2 MHZ. This was done via the beagle bone pru in assembly (see ldgraphy). I looked at ESP32 but it has many abstraction layers and typically uses RTOS. You would need to remove these which requires a deep understanding of the ESP32 and makes your code not portable.

@@Hexastorm That was the issue I ran into, even at 240MHz (cpu clock) the code would not run fast enough. I initially tried to circumvent processing interruptions by using RTOS and making one core perform pwm output while the other monitors trigger conditions. Even doing this did not work past, say, 20 frames per second without introducing significant jitter (+/- 3 degrees). Abysmal. So I just today bought a Raspberry Pi 4 B off of craigslist. Assembly is not something I have proficiency in but I will learn it.

@@Hexastorm I realized I need to use an adc, so I will just have to use Node MCU 32s as the ADC for the raspberry pi.

@@lazarusmagellan2367 you don't need an adc.. a raspberry is too expensive.. I think you r are just starting too realize what the problem is . The algo is pretty basic.. a microcontroller can be cheaper but it will take you a long time if you r unfamiliar. For industrial solutions fpga prevail, as they provide a route to asic. Furthermore all the knowledge and time spent is hard to monetize, you are probably better off learning about time series and data science and fixing some boring business problem. You can buy a beaglebone and use ldgraphy or my beaglebone variant or the fpga. Fpga is cheapest and fastest ( current rate is 12 MHz). Interested in microcontroller solutions with a price below 5 Euro's.

@@Hexastorm I still don't really understand why my system doesn't work. I am essentially doing the same thing you are, just with a slightly simpler setup. I have an IR Led that reflects off of a mirror (polygon scanner) and onto a silicon photodiode (BPW 34). a void loop runs continuously polling an input pin, checking to see if the photodiode voltage is above the threshold value. if it is, then the laser is pulsed in a sequence of microseconds. Even with this incredibly basic code and setup, i cannot get a static image. The lines jiggle left and right and sometimes jitter >10 degrees off center. Yet when I check the photodiode and esp32 response times with an oscilloscope, nothing looks out of the ordinary. Hell, even a basic arduino should be able to do this without any problems. I just don't get it...

I am working on a product which is easier to use. Bringing a product to the market is not easy.

@@Hexastorm at large factory they use large high DPI UV LED screen to direct expose to PCB but your idea is great for small and DIY users .

Originally, a system was developed for large substrates (look for AMSYTEMS). My version would indeed be DIY, small scale. Advantages would be it can expose curved substrates. Also high power lasers can't be directed by LCD or LED screen.@agaberahmed6657

I use freecad in combination with kicad, there is another video where you can see this. It does indeed take minutes to burn design. I have used projector's. Most popular in ldi is polygon mirror with paladin laser 365 nm. I believe my design should be able to outperform most projector's..

PCBway produced the PCBs. Minimum order quantity is 5 at a price of 25 USD. I have a thickness of 0.8 mm and 4 layers. There are several plugins for Kicad which can be used to generate the tracks.

@@Hexastorm Thank you. Is that 25 bucks a piece? I read that they do up to 16 layers these days.

@@christopherd.winnan8701 no 5 bucks a piece

A breakdown of a dvd motor is shown here (th-cam.com/video/Qb-uDgbgd3Y/w-d-xo.html) . You see that is uses coils in combination with a ring that is probably a ferrite oxide which is magnetized alternating. Classical polygon motors are made similarly and balanced hereafter. My motor is not balanced very well, this is not an issue for my current speeds but obviously something that could be improved.

I am still working on it. But like 1-2 hours per day.. focus is now on some critical but boring parts

Indeed that is the whole point 😂, easy to use laser scanner for pcb at home

Partly agree, the prism can rotate up to 24000 RPM and has 4 facets, so 1600 lines per second and is much faster per pass than one line per 2 second. My system operates at 2000 RPM and is very slow in the movie as each line is exposed, I believe 16 times. I did not use full laser power and speed to simplify debugging and as it is easier.

Hey Tesla atoms, I am open for collaboration. You can sent me an email on info@hexastorm.com. All code is shared via my Github profile github.com/hstarmans. I am willing to answer specific questions, clarify decisions or change code so it simplifies collaboration. I do not have a lot of time for support.

here is a summary presentation th-cam.com/video/b7ArZDhsyfI/w-d-xo.html

@@Hexastorm Thank you so much for sharing this, this content is GOLD!

Yes, i did this before building my own..

If you have board which you can expose from two sides, and align front and back within 0.3 mm you could.. biggest challenge is probably vias

@@Hexastorm can this be used for single layer aluminium mcpcb aswell?

@@bibinvarkey3041 it can.. you can even use it for other substrates. Coat it your self, expose it and then remove the exposed parts using chemicals and then deposit or remove chemicals on the revealed parts. Hereafter remove the coating. It is not my area of research however.

Voltage ligt nu rond de 5V, de motor driver kan tot de 11 Volt aan. Een probleem bij hogere voltages is de hitte ontwikkeling en opwarming van de magneten. Je wilt dus niet te hoog gaan. De wrijving in het lager is heel laag, daarom is er zo weinig nodig.

I will check it out, I am aware that the coil design is suboptimal and youtubers like atomic showed better designs

@@Hexastorm so I found his channel through your subscriptions. And yes the wedge shape coils is better than the circular traditional coils for the same reason. But faulhaber coils are a step above improvement over wedge coils for the very same reason. The radial flux version of a multi pole pairs faulhabor coil can be seen by googling images of "ThinGap motors" as a maker of motors with high torque. They have impressive power density specs. Alternately simply searching images of faulhaber coil may turn up many two pole radial flux. My motor seams to be one of only two motors that I can find that are axial flux faulhaber coils but serously you get more percentage of wire activity participating in providing force in the vector that torques your rotor around.

I looked into it build-its-inprogress.blogspot.com/2015/02/coreless-axial-flux-motors.html?m=1 and here the faulhaber is explained.. I would need to do a calculation, but it is highly likely it is even better than the leave motor

@@HexastormNice example of nearly 100% Copper fill minus only that which is needed to define one trace from another. You'll be able to do even better with traces that grow in width as they go from the inner diameter to the outer diameter. I wish i could draw this out here but, I will try to describe it.. Let us first simply to linear motor... Force=current×length×magnetic field strength. But if the wire is at an angle other than 90 degrees, then Force=current×length×magnetic field strength×sin(of that angle) Let us compare the serpentine coil against the faulhaber coil. With 1 amp and the magnet's field strength = 1 tesla and are 4 cm tall and spaced centered at 3cm apart. First is serpentine coil: The coil is a repeat of 4cm + 3cm = 7cm segments for each magnet. The 4 cm segment is at 90 degrees to the direction of force we want. So Force= F=1amp×4cm×1tesla×sin(90 degrees)=4 amp cm tesla. And F=1amp×3cm×1tesla×sin(0 degree)=0 amp cm tesla. Adding the forces from those two segments you get force = 4 amp cm tesla. I'll Work with this unit because there's no conversion. Faulhabor coil second: Using the same magnets makes the length of the diagonal 5cm using a²+b²=c².... 3cm²+4cm²=5cm². F=1amp×5cm×1tesla×(4cm/5cm) And reduces to F=1amp×4cm×1tesla and further reduces to F= 4amp cm tesla Take note that the force generated against each magnet is the same for both coils. But, the faulhaber coil only needed 5cm of while the serpentine coil needed 7cm per magnet. This linear faulhaber coil is (7-5)/7=.2857 or 28.57% shorter than the serpentine coil for the same thrust.

@@Hexastorm his motor was part of what inspired mine. But where he went went without an iron core,ninwentnwitj a slotless iorn core made from coiled strapping steel. While it does introduce some losses my motor really needed additional inductance. When I said my motor was one of two motors that i am aware of that are using axial flux faulhaber coils, the one you found is the other. And his was some of the inspiration in my motor.

My current aim is to make a single module, which has controller laser and motor. You can then buy this laserhead as is. As it is open source, it should be easy for hackers to integrate it in a a pcb machine, 3d printer etc. I also try to keep the price low.

@hexastorm that's great. I have also tried this. I would prefer it to use with PCB. But I won't recommend this for 3d printing due to laser life. If you want to make it cheaper. Than why to use fpga instead of microcontrollers. May be a stm32 and esp32 with some external ram. Can we some have discussion. I am also interested in this project. I might be able to make this work.

@@muhammadarsalan3412 , the first iteration was done in assembly on a microcontroller and based on ldgraphy. I picked FPGA, as it gives you more options and seems more likely for future industrial systems. This is where I think you can make most money. The current FPGA is around 5 euro's. Most costs still reside in assembly.

Do you have github or somewhere I can get the updates and maybe able to help you out ?

@@muhammadarsalan3412 the FPGA code is hosted here github.com/hstarmans/hexastorm The files are self contained so lasers.py describes how the laser head works and you can test it there as well. Some files don't have tests, but this is rare. The code for a microcontroller can be found here github.com/hstarmans/ldgraphy/blob/master/machine/binaries/stabilizer.p It runs on a PRU of a beaglebone.

Thanks for the idea, I actually looked into magnetic levitation, although not pyrolytic sheets. The article pubs.acs.org/doi/10.1021/ja310365k looks promising. The problems I see; - will the sheet stay planar (if it wobbles along the axis of rotation), there will be a cross scan error - ideally the prism can spin at high speed, I have seen videos (th-cam.com/video/AU0q4wVohF8/w-d-xo.html) at low speeds if the prism is not balanced at high speeds, I fear it will get out of orbit. Now the axis, counteracts the centripetal force.

@@Hexastorm Yeah, I think this is where you can add in hall effect sensors to compensate for wobble. I would also use a single neodymium magnet glued to the center under the pcb, and then use the remaining coil actuators to balance and control the orientation and direction of spin. I bet if you can get a good measurement of the reactive magnetic field on the pyrolytic graphite, you can use it to actively balance the prism in the center, as the constant rotation should create a perfect periodic signal, and lagging or leading magnetic wave forms from the hall sensor can tell you which direction to adjust the pwm strength per coil. The other neat thing is if it's self centering, then you just need a certain amount of tolerance in the design and it can handle not being perfectly centered. one issue with moving it parallel with respect to the force of gravity, your EMF from the coils will need to compensate for that, or the bearings will need to center the prism. It might be easier to put the plates on a linear gantry and move them across the linear array of light coming from the prism. It's how all modern flat bed scanners work and it's a lot easier to build a smoothly moving sled than a vertical high speed spinning low cost scanner element.

​​@@Hexastorm Are you moving the prism radially? If you are, the off balanced weight may be balanced by another equal weight moved the same distance in the other direction. On a reverse threaded screw. //[ ]////(O)\\\\[ ]\\

@@kreynolds1123 I mentioned radiallly but this is not correct. I mean that the prism is not balanced, the center of mass is not at the center of rotation. This gives a force which increases with speed and increases the friction on the bearing. I have done balancing experiments in the past. It is not an issue at low speeds like 3K rpm, I do think it could be an issue if you use magnetic levitation, but maybe you can search for the right rotation point !?

@@Hexastorm an unbalanced weight simply needs an appropriate counter weight to balance it. I'm not a balancing expert but I'm pretty sure their are at least two different kinds of balancing you need to deal with. First is that you can place the rotor axial on two points and see if the rotor has a preferred side up and balance this. To remove the preference. But then there is another way a rotor may be off balance with weight positioned significantly above or below the rotor disk that encourage a different kind of wobble. Where a motor is constrained and rpms are minimal the wobble is tolerable. But at high rpms it stresses the motor and bearings. And where the rotor is not well constrained on magnetic bearings, wobble quickly poses very severe problem. Static and dynamic balancing. I forget the guy's name but there was this guy who did dynamic balancing with a speaker to quantify level of unbalance and how adjustments improved or worsened. I'll see if I can still find that video.

Novel thing is the prism, most scanner use polygon motor and a bldc with wires. Envisiontec has patent on rotating mirror

It is a four layer pcb, 0.8 mm thickness..

The overall concept is quite simple but it is quite a lot of work to get it right.

I improved my controller quite a lot and hope to release a video. This might still be some weeks.

A bldc motor makes it spin.. there is more information if you look for open hardware laser scanner

Progress has always been slow but steady. I worked on it for over 5 years. My current aim is simplifying assembly of the module.

Added new video where position feedback of hall encoder is used. Motor driver died so couldn't make a better one.

@@Hexastorm there might be an interesting use case I need to verify and I will in a bit of time

@@x321y Sure look forward to it

Roel, you r right that for this sort of layout, I could also have used a LCD screen. Video is mainly used to show that the technology works. Maybe I will add more examples later.

​@@Hexastorm at 2:46 you talk about the imperfections, this is mainly caused by the resolution of the design (file) and the resolution of the machine. both need to match the same or a multiply of it. in this example (photo from Mr. Heller ) i blame the mechanics for it. never the less, it's a promising concept. I prefer to make the artwork on negative film and therefore I am thinking of using an drum instead of the mirror.

I've had some great success using a resin 3D printer also. My choke hold is finding a simple way to get from eagle Cad to a 3D image that I can slice. I've used photo sensitive boards and when I have enough fails, I reverse the image and use resin as the resist. messy but very cheap. "Learn how to fail fast and cheap" Elon Musk.

​@@alexportiiii6414 You can also buy the display and an adapter to convert HDMI to MIPI DSI off aliexpress (or just the adapter). I considered using the resin printer and its firmware... but I wanted something more compact and would drive it with raspberry pi. I haven't finished it because the laser method is more interesting to me.

The areas between the tracks look over exposed. Note that, the exposed areas are removed. If I move back I don't end up at the exact same position, also my sideway move is inaccurate in 0.1 mm. This is small but very visible. My current focus is on replacing motor with pcb motor (carl bugeja) and raspberry with esp32. Hope to post a video soon! My current focus is not on improving the exposure, I am confident it can be fixed with more effort.

I have made a list of alternative exposure technologies here reprap.org/wiki/Open_hardware_fast_high_resolution_LASER#Alternative_Exposure_Technologies Acoustic optic deflector, reflective polygon scanner, optically addressable light valve, DMD, grating light valve etc. Most are covered on the list.

Thank you v much your the best. Only a few of these Tech can be used in the LDI for Lithography right? and i thought that Orbotech uses Reflective Polygone Scanner is it the same as Rot Mirror with reflective F-theta lens and one laser bundle?

@@cosanostra8341 orbotech is refractive f-theta lens, reflective f-theta lens is used by next scan technologies. Correct, not all are commercially used for LDI, some are not suitable

Thanks maxwell!!

exposure result is shown in this post... hackaday.io/project/21933-prism-laser-scanner/log/200330-new-laserhead-exposure ... Plan to do measurements in the following days.. first Formnext

Thanks Arthur, moved to a new chip and will use dsp with casteljau algorithm. I couldn't do multiplication fast on this chip so had to stick to addition. My current solution requires a lot of LUT. Beagleg is also working on this problem. They don't use nmigen however.

Thanks for watching. I never heard of pyrate so will definitely check it out. Documentation on this topic is still a bit scarce in the freecad community so I just picked something.

Jan, you are right code moved to github.com/hstarmans/hexastorm/blob/master/src/hexastorm/movement.py . I updated the link in the video. I am moving to a different chip with a dsp this will change the code

The ideo of using a prism originates from Dr. Jamar. Multiple person have looked at this problem. You can find a description of the original module here hackaday.io/project/21933-open-hardware-fast-high-resolution-laser/log/195579-image-of-deep-state-hexastorm-released .

Thanks for your reply.. my aim is not creating tutorials for Freecad but getting a working prism scanner.

@@Hexastorm great sir

thanks zeller and for putting LDgraphy out there it lives on :-)

Made a new video; you should check it out! Results are on hackaday; th-cam.com/video/SofTXG5s_MA/w-d-xo.html