Shaded vectors show - help

[Murilolana]

Hi everyone,

This is my first post, so first I'll introduce myself. My name is Murilo, I'm a designer and intusiast of DIY stuff, but I'm a newb on electronics, programing and of course, lasers...

I've been browsing for everything related to laser scanning but somethings are still not clear for me.

I wish to build a simple monochrome system with XY galvos, but this system need to project shaded vectors. So, for example, this system would show shaded letters instead of just the outlines, or a shaded circle instead of just the outlines. No faded shades or "dégradée", just simple filled vectors.

So, I thought about two options of doing that. The first would be generating a "toolpath" of the vector on a CAM software and outputing this path to the galvos, showing a full filled vector. In this case, I would use Mach3 or similar software to interpret the Gcode and control the galvos. Has anyone here ever done this before? Is it possible?

The second option would be converting stencil images to full shaded vectors. Is there a software capable of that?

I wish to build a true SLA DIY 3D printer, that's why using a CAM software to generate the path and mach3 to control the galvos would be the best choice for me, because it would be possible to control other axis and functions too.

Please, does anybody know how to do this (or could give me some help)?

Regards,

Murilo

 

[Cyparagon]

I think what you're calling "shaded vectors" is what we call raster graphics or raster scanning. X/Y galvo systems are not suited for raster scanning in a projector setting. It's certainly possible and will work fine for your application though, since you can live with <2Hz image refresh. Not sure on the software.

 

[Murilolana]

I think what you're calling "shaded vectors" is what we call raster graphics or raster scanning. X/Y galvo systems are not suited for raster scanning in a projector setting. It's certainly possible and will work fine for your application though, since you can live with <2Hz image refresh. Not sure on the software.

Thanks for the response.

I think that's not a problem, since what matters is that the laser cures the resin. So, no problem at all if the image blinks a little bit. It can even be displayed in a "scan line" fashion, like up/down projection.

Can you tell me if it's possible to control the galvos through mach3 or similar software? I know mach3 outputs the axis signals (step and direction) as TTL, which are interpreted by the motors drivers and converted to analog.

In case of using galvos, I assume that the way would be: PC -> breakout board -> DAC -> galvos

If that was possible, I would use a smoothstepper board to comunicate with the computer and output the TTL pulses to the DAC. This board is capable of generating pulses in a rate up to 4mhz, wich I beleive is enough for driving the galvos at full speed.

So, is this possible?

Thanks

 

[Bionic-Badger]

So many people interested in this resin curing stuff...

Galvos are the wrong technology for this type of work. Use them for actual vector work, like drawing lines, not raster-type operations. Even if you control your galvos to scan across the image, you probably don't want to use that because you'd get inconsistent curing as the beam scans across. It'd be like those old cheap ink-jet printers that you could see banding between each successive row.

Why don't you stick with the DLP arrays instead? They deflect the light to produce a single image at once, instead of a point-beam.

 

[Murilolana]

So many people interested in this resin curing stuff...

Galvos are the wrong technology for this type of work. Use them for actual vector work, like drawing lines, not raster-type operations. Even if you control your galvos to scan across the image, you probably don't want to use that because you'd get inconsistent curing as the beam scans across. It'd be like those old cheap ink-jet printers that you could see banding between each successive row.

Why don't you stick with the DLP arrays instead? They deflect the light to produce a single image at once, instead of a point-beam.

There are several dlp printer projects on the web, but I have my doubts about the dimensional accuracy (not resolution) of it. There are optical distortions related to that.*
Laser systems are being used since the 80's for this application, but the machines were very expensive, until this:*Formlabs FORM 1 high-resolution 3D printer spotted in the wild - YouTube

This printer uses XY galvos, apparently 20kpps. So that's what I'm trying to figure out.

So, if the galvos could be driven by CAM path it would be great. The beam diameter would be the "tool" diameter and so on... In this case the tool path doesn't necessarily need to be a raster kind, but also offset, spiral or any other.

Mach3 uses TTL signals to send step and direction commands which are converted to analog by the motors drivers.

Do you think it is possible to do that?*

 

[Things]

Galvos and steppers operate in a completely different way. You would have to come up with some way to electronically convert the step and dir signals to an analog voltage, usually between -10 and +10V I believe if you want full deflection.

You could probably do this with a microcontroller, however you'd have to be very efficient in your code, as the pulses are quite fast.

I think what you're referring to is raster scanning as Cyparagon said, like this:

 

[Bionic-Badger]

There are optical distortions related to using galvos too. You're essentially deflecting from a point source which will introduce path-length variations, affecting the spot size.

Also I'd trust the DLP system more than the galvo system with respect to accuracy, as the pixel are at least spaced at fixed increments, rather than relying on the mechanical motion of the galvos that can overshoot or cause other problems. Also the DLP images the layer all at once, rather than relying on scanning, leading to more consistent curing.

You are, of course, welcome to try making a resin curing system with galvos and such.

 

[Murilolana]

You are, of course, welcome to try making a resin curing system with galvos and such.

Thank you. I'm junmping in this "adventure" for sure.

Well, the base spot size would be something very tiny, like 0.25mm diameter, so, some variation is acceptable. If this variation gets to unacceptable levels, I could make use of f-theta lens in order to make the spot size constant.
Good DLP projector with good optics could be too expensive. For larger parts with high resolution, only full HD projectors would do fine. The minimum focal distance is kind of a problem too, so the optical elements need to be modified. DLP lamps could be much more expensive than a laser module too, and they doesn’t last as much. You got to break the color wheel inside the projector in order to let the UV spectrum pass through… All these factors make it not so simple and maybe not so reliable.

If the galvos could be controlled by a gcode planner, that would solve most of the job. It would be possible to control everything; laser intensity, blanking, projection type (raster, offset…), exposure time and so on…

Any tips on how to do that?

 

[LSRFAQ]

I spent years using a commercial galvo based SLA system with a UV HECAD, a pair of G120s, and a long focal length F-Theta lens. Writes rather fast. The galvos were held at a constant temperature to make the position sensors stable enough to make a .0001" horizontal resolution. Galvos with capacitive feedback sensors are used for this application, which is why Cambridge and Scanlab still make galvos with that kind of feedback sensor. Four sensors with pinholes in the corner of the scan field provided registration.

I find trying to hack the TV MEMs juvenile and un-productive compared to a direct write laser head. Only reason to do it is the in-ability to source a F-theta or do active focus by a bunch of hobbyists who never seem to bother to look how the pros do it.

G-code is a waste of your time. For the simple reason the polygon render code output for the 3D structure is already a vector path. Adding G-code adds a slow intermediate math process that will cost you time and resolution.

Hint, there is a spatial filter and beam expanding telescope upstream of the galvo head on commercial units. You'll need a single mode diode.

Steve

 

[Murilolana]

I spent years using a commercial galvo based SLA system with a UV HECAD, a pair of G120s, and a long focal length F-Theta lens. Writes rather fast. The galvos were held at a constant temperature to make the position sensors stable enough to make a .0001" horizontal resolution. Galvos with capacitive feedback sensors are used for this application, which is why Cambridge and Scanlab still make galvos with that kind of feedback sensor. Four sensors with pinholes in the corner of the scan field provided registration.

I find trying to hack the TV MEMs juvenile and un-productive compared to a direct write laser head. Only reason to do it is the in-ability to source a F-theta or do active focus by a bunch of hobbyists who never seem to bother to look how the pros do it.

G-code is a waste of your time. For the simple reason the polygon render code output for the 3D structure is already a vector path. Adding G-code adds a slow intermediate math process that will cost you time and resolution.

Hint, there is a spatial filter and beam expanding telescope upstream of the galvo head on commercial units. You'll need a single mode diode.

Steve

The idea of using Gcode to drive the galvos was due to the fact it could provide better setup possibilities for this application; things like G50 (scale factor), G68/G69 (rotate coordinate system) and some other. Another great aspect of using Gcode is that many other path strategies could be used, like offset and spiral. For this particular application it would be much better (in my opinion) than the traditional raster strategy. Step over control would give great options for path overlay in relation to the “tool diameter” (beam diameter). There is already an open source software called Skeinforge which slices STL files and generates gcodes for each layer. I think this is what RepRap uses. But, in order to make this possible, some code and hardware would be required to translate step/direction pulses precisely into -+5V.
The problem with this is that it would require another device to generate step/direction pulses at high frequency; a board like smooth stepper, which costs around 170,00 dollars. Another downside of this solution is that it needs to be fed constantly by the control software, like Mach3. So, if I wanted to add a “print from SD card” feature on my 3d printer, it would be a pain in the…

So, the solution I found is to write a Gcode interpreter to send integer values to the control board .
I’m no programmer, so I’ll need some help doing this. Some programmer friends of mine will assist.

I’m waiting for a galvo kit I bought to arrive, so I’ll begin to experiment…

 

[kingston250]

I think your second option is better for you so you can convert stencil images to full shaded vectors. You can use the Illustrator editor to edit your image and convert it into vector graphic. You can the graphic examples here. Headphones Vector