7 questions on laser scanner design (DIY?)

[withlasers]

Many thanks, mate.
Much clearer now what the signal limits should be, for now I will limit the firmware to never request more than 9.5 volts differential from the DAC, in X and Y, this global downscale will keep me out of the edges without messing with the center/offset.
I already did and it worked yesterday, actually, but I copied the implementation from a brightness setting that existed on the config webpage, and that was kinda broken, so my setting was also broken.
So from yesterday to today I'm rebuilding the firmware with various fixes and new stuff.
I gotta say if his plan was to get rich, then he should look for a better plan , this isn't exactly a "disruptive" market where you can design a simple product and become rich just like that.. but maybe you know him better than I do
Anyway, I'm finishing up my firmware soon with basic features just enough to enable a testing workflow (load ilda test pattern) and then I will move back to my application, thanks for the tip on focusing on the end goal, in between all the other ways to project an image, the different purpose galvo sets, the protocols, the laser knowledge, the electronics and pcb making for the laser, the lens knowledge, the system design of the whole projector, the data manipulation from CAD assembly to individual parts .dxf's, this project has me all over the place, it's easy to lose track hehe but this is the final stretch, I honestly think the only challenge ahead is the dxf to ild conversion, which I'm banking on openlase forks to help me with, after all I will only be projecting the outlines of the parts and some alphanumeric codes next to it....
So yea, wish me luck! and thanks for all the support, and cool jethro pic! it took me a few times opening this page to put a name to the face but I knew that I knew it from somewhere!

[withlasers]

Nah. I've only watched some of his earlier YT videos, trying to use steppers, instead of galvos.

: to be honest I have been thinking about pushing steppers to the limit for a summer project, I'm wondering how interesting the workflow could become if switched to digital, will be slower, sure, but it might allow a different way of drawing paths, directly with vectors instead of "timed points", just because it's cheap to try and I'm curious with it. Also analog control loops breaks my brains, left and right..

Sounds like you're making good progress.

: yea, once I'm done I'll check with the company if they let me share parts of the report here for your curiosity but also for future noobs reference. I took too much from this forum not too contribute hehe.

By the way, question about the sounds, I got rid of the clipping by the global downscaling the X-Y coords at the end of the pipeline, right before the DAC's, but the circle is still making a noise, less, but still making clicking noises, I looked into his code and I'm assuming it has to be with how he generates the points for the circle vs for the square/cross/heart, I'm assuming the circle has a crazy high point resolution and that's causing the audible noise, even at lower scales.
However when loading more complex frames, like test patterns, it gets a little noisy too, from soft coil whirring up to audible ticks and clicks, not as bad the test circle from ildawavev2 firmware, but still noisy. And now I'm left wondering: what level/type of noise is considered acceptable? and at what point should I get concerned with the lifespan of the galvo's/driver board? Because when I do a simple square from the firmware, or a circle.ild from the library here, I hear 0 sounds coming from it, but then again those are very simple shapes with smooth acceleration curves... I just don't know what to expected from the complex frames...

Jethro Tull was the band. Aqualung was the 'poor old salt' on the album cover, looking alot like TheHermit

: oh I know, I was lucky enough to catch Ian Anderson in a swampy tent concert during a week-long music festival in the south of Brazil, a bit over a decade ago, those were some magic flute solos! And you mean it looks like yourself? Or like the hermit archetype in esoteric practices like tarot?

[withlasers]

"timed points", as in **kpps?

yes, I meant I was wondering what if instead of spitting out two points " : " at a set period, with the caveat that we don't turn off the laser in between them, and get a line, what if we could spit out a vector " l " that has its own duration for travel? sure it's not interesting with lines, but I wonder what the workflow would look like for splines, anyway, just a cheap experiment to run.

What does his sine wave look like at the X/Y output connector on the scope?

So the weird one I mentioned before was this:
https://drive.google.com/file/d/1nhZ...usp=drive_link
But I'm at this:
https://drive.google.com/file/d/1mRk...usp=drive_link
https://drive.google.com/file/d/1D7Y...usp=drive_link

I was busy trying to reduce the breaks between the end of a frame and the start of the next one, but it seems that this is the most I can shrink it.
Now, even at 10kpps and a moderate angle (0.7*30deg), I hear something, I wouldn't call it a click, more like a tick, a quick tap, if I lower the angle absurdly (0.1) I only hear the high frequency whirring that I would expected anything to do at this speed, unless in a vacuum hehehe. so I assume that's normal sounding, but not sure if healthy.
But then if I load the 12k ilda pattern, it's ticking like crazy, which I'm not sure if safe or not, and that's at 10k which makes the image barely readable, just strong dwell points with light lines in between them...

Shouldn't have clicking noises with a sine wave.

I thought the clicking was from the end of a circle frame to the start of the next, and tried to reduce that, it improved a little but not much, I guess I would need to have a constant sine+cos to get the cleanest sounding circle, but then we're not talking about an ild file being interpreted and passing thru a DAC anymore, right? then we're talking about a signal generator.. because of the blanked points at the start and end of a file/frame...

Is it clean or still have overflow inversions going on? If it's only a test circle for deflection/size measurements, then it doesn't make much sense to be high frequency.

In fact I don't think I ever got the inversion you're talking about specifically, just that distortion shown above, but maybe the audible noise is coming from the steps along each point, if you look closely you will see that the step response of each point along the circle shape is very fast and ripples around... atm I'm thinking that's where most of my clicking is coming from.

It's important to have the driver boards' power amps properly in contact with heatsinking usually screwed into the aluminum optical plate w/paste. They can get pretty hot, if being driven hard with pulse, square, sawtooth, or even ild frames with long leaps. The galvo's X/Y block is also a heat sink.

yep that's how this kit is made, you described it exactly, I'm happy that they insulated/heatsunk the tab properly, and then I bolted the heatsinks to the frame (3mm stainless steel sheet bent into a cube) but it is getting hot, I measured 43deg yesterday, so I will most definitely be installing a fan, when I'm done I think I will even interface the heatsink to the frame, either with pads or paste...

In terms of noises, you can tell by the image shape whenever the galvos are being driven to hard. Whenever sharp corners start rounding off, lower the frequency and/or image size.

I got this with the LMTESTF.ild (diamond inside square that became famous for coming in standard on a piece of hardware/software, you definitely know it) at 30k, the diagonals became ellipses, but at that point it sounds almost like my ultrasonic toothbrush lol do you mean with this phrase that if the image shape is pristine, the noises can be ignored as normal operating noise? If so then there's my threshold to feel safe playing with it. I really gotta move on to building a mapping() function so the plane transforms can happen to my drawings before they are converted from .dxf to .ild, but I don't feel safe with it just yet.

Yes. I had never heard R&R flute, like Ian's. Inspired me to buy a silver flute and ave a go. In '79, I was doing Laserium shows at the London Planetarium and was fortunate to see Tull at the Hammersmith Odeon, during their peak. Such an amazing showman.

Yes, when my hair is long and yes, but without the hooded cloak.
😎

damn you are living the dream ain't ya? hahaha let me know if you know other bands like that, because I find it hard to relate jethro's sound to anything else out there

I also pushed a video to google drive so I can share the noises it makes, unfortunately it seems the upload button here is broken...

https://drive.google.com/file/d/11nb...usp=drive_link

[withlasers]

#1: Nasty waveforms, as if it's only noise or the scope's ground isn't connected to the project's ground. Is the project actually grounded (aka Earthed, in EU speak)?

I just figured out what was the problem with those ugly waves, it's the generated circle from the firmware, if I load the circle.ild I get the long pauses, and if I load the circle from the firmware I get those ugly waves, I've seen the code and I think it's because it's generating points on a 90deg arc and multiplying coords according to the quadrant they are, was something overcomplicated like that, anyway not a problem, I won't be dealing with that.

#2: Don't understand what those spikes are about.

those are the V+ and V- of the X axis making a circle, and as you can see there is a looong time in between them... I thought I reduced all the firmware delays before and after frames/files, but now that you said it should be a constant wave, I might look into the firmware again.. those are being read from the sd card, let me try and stream points over wi-fi see if the delays are different then.

#3: Much cleaner wave, but why only one, instead of a stream?

it's the same wave as before, just zoomed in, like I said I improved the firmware to reduce the time between files, but I got to a point where the time before and after each wave match the blank points at start and end of the circle.ild that I have. I am about to do the math to see if the resolution of the DACs match what I'm getting here, maybe I have a firmware bug shrinking my actual DAC resolution...

Is that what those lumps at both ends of the wave are? Points on top of each other?

nevermind what I said in grey, you're right, even if they were blank points at the start and end of circle.ild, they would have positions, unless that's their position, but nah it doesn't make sense that circle.ild would have more time blanking than lasing.. I will double check the .ild see where are the blanks supposed to be.

Diodes are far faster than the galvos. If there's blanking between points, often there will be several points at the same position, with pre and post blanking settings within the imaging app to hide overshooting and bright corners. Otherwise, that needs to happen manually while plotting the ILD frame(s).

Ok, that's going to be for v2 of this prototype hahaha no way I can fiddle with those optimizations in the short time I have, but I would love to, like I said I don't have an imaging app, I'm getting step assemblies, flattening to .dxf's where each part in the assembly is in a separate layer, and upon request to the application, it will gather a group of layers (parts), transform to the geometry stored (projector at the ceiling, assembly table has wheels and likely move during a shift or day), then convert to ild using one of the forks from openlase on github, hopefully my single-frame .ild files with nothing but a few letter and rectangles (I might even program round corners if I have the chance) should be easy enough on the hardware..

I don't think so. Those steps are too small to generate much sound which would be more of an hf tone, instead of clicking. Noticed the stair stepping on all 3 videos, as if X & Y outputs are alternating in time, inside a loop that has a millis() delay in between to control kpps, instead of spitting out both X & Y together, followed by a single delay before the end of the loop.

I see, it makes sense, still I will check if those steps are a product of the DAC limits or if there's some bad math somewhere limiting their true potential, I have to double check because it could be that I downscaled the projection so much when I took the picture, that the DAC resolution was showing.. I don't know about that second part, I haven't scoped both X and Y simultaneously yet, so all you've seen is one single axis.

Yes, fan(s) definitely. Fully enclosed case to keep the optics dust free. Thin AR coated glass emission window. Filtered air inlet at one end + ~4020mm fan near the galvo drivers on the opposite side of the case should be sufficient.
Your 5mw diode won't generate much heat, but still needs to be mounted inside a surrounding heat sink block. Whenever laser diodes get too hot, their resistance will drop, sending them into self destructive thermal runaway.

oof, guess I'll have to jerry-rig a fan on the lid of this thing, unfortunately I got to designing the box after shit had hit the fan already, and did it with loads of mistakes like this, I'll make a note for v2 to keep optics enclosed, and electronics exposed to airflow!

Yes, generally speaking, as long as the galvos are tracking accurately, then all swell unless you're throwing a high level >2kHz sine at them. The circle will look ok, but will be small because of inertia. The drivers will be working hard and generating heat.

ok, good to know, that gives me that threshold of acceptable whirring/buzzing/ticking

Wow! You can make galvos talk? 13 secs of a wandering camera doesn't provide much time for analysis. Couldn't hear the clicking in the video.

sorry but the interns don't their own office too bad you can't hear, I can hear it in the video, let me record another video with more info

Those galvo mirrors look very close to each other. Are you sure the clicking noise isn't the mirrors hitting each other at high signal levels?

nope, it's chinese but not that chinese hehe

I don't recognize those galvos. Ali Express no name specials, I presume? How many kpps & @ what deflection angle are they touting? If that's were they came from, then you might want to half those claims in reality.

this is the one they say 30k and 25deg, but yea I can't get a readable 12k ILDA test pattern even in a tiny angle..

Check your firmware to try getting rid of the stepping. Then you can start working on the ILDA imagery.

Will do, like I said it might have been that I was working with a stupid small angle, like 20% of the total angle, which would mean +-1v in the signal, and maybe at that scale of 0-2 volts the DAC resolution starts to show, not sure, will find out if it was a DAC limit or firmware bug.
But I gotta say I'm only looking into the steps and the gap between these circles right now because I think that might get my ILDA 12k performance to increase, and once I'm happy with the performance I'll move forward with the project.. still a lot left to do and plenty more to write in the report ^^

I'm going to record a new video right now with some more info on my build and show what I failed to show before

[withlasers]

Small update just to keep you guys ("guys" as in TheHermit and all his personalities) entertained:
•That day we closed up earlier and I couldn't record a proper video.
•Next week started slow on a Tuesday, but I did manage to solve:

1. The weird waveforms, was a byproduct of the Circle.ild that I had, the auto-generated circle from the firmware looked better but had its quirks as well, eventually I generated a "max" diameter circle with 1000 points and then the waveform would both: stitch smooth between each period, and have smoother steps along it.
2. Found the angle for scanning the 12k ILDA tp at 30kpps, it was 7% of max amplitude, which assuming a max of +-30deg is around 2deg, there you have it in case anyone is interested in buying that particular galvo head set.
3. Wrote a new firmware from scratch that focuses on reliability of the show, regardless of network situation/file size.

•However I also managed to:

1. Burn the laser, a zener diode and slightly damage the LM317 I'm using as current controller, all because I thought it was time to reduce the power to class 2, but wanted it adjustable, so I've put a potmeter along the set resistor, to allow me ranging the resistance between 20-40 Ohms giving me 30-60mA which is equivalent (when cool) to 1-20mW of optical output at the diode.
   Not sure what happened there, took me 1 day to get it running, and it was blinking, at the end of the day it died, took another day to figure out exactly what had died, the why is still unsure, definitely adding a potmeter in the current path was the problem, but it could be inductance from the resistance coil, could be that the pwm I introduced didn't bode well with the circuit, could be that adjusting the potmeter during operation caused the lm317 to freak out a little, due to the electromechanical design of potmeters (wiper losing contact as it turns would open and close the circuit, switching the setpoint of LM317 from 30-60mA wicked fast).
2. Burn the spare laser, decided to use wires to the leads this time, and the tin flowed down and shorted the pin to IC packaging, in my lead-smoked-brain I tried scraping it with the solder iron, which probably overheated the packaging, I realized it immediatelly and scraped the tin instead of using heat, but it was too late, multimeter says it's open with no voltage drop across :c we just ordered a new one and Monday I shall continue with testing the complete application.

So yea, all hands on deck now, I'm blasting the report faster than chatGPT, by Friday I deliver it as is and from Monday I carry on with final tests, production testing and presentations, until 10th of July.
When I get it up and running again next week I'll make sure to get a video and share with you guys my little projector, I was surprised with how well it handled some animations, it can get quite noisy at 30k even at a small angle, but luckily I don't need multiple frames anyway.

[TheHermit]

Small update just to keep you guys ("guys" as in TheHermit and all his personalities)

Thanks for the update. Hadn't forgotten about your June 9 deadline. Thought you might have needed to pull an all nighter and was giving you some time to recuperate, before PMing to ask how it went.
Glad to learn that you've managed to deal with all the technical burnouts to get your projector happ'nin and still on schedule with sufficient time to program your presentation, next month. Well done!
Unfortunately, your electronics mysteries are above my skill set, as a mere old school laserist, electronics guru isn't one of my personalities.
Perhaps Adam or Steve might like to chime in with some helpful electronics suggestions.
Looking forward to seeing your video.
BR
Roj

[mixedgas]

Potentiometers open and close the circuit as you spin them, especially when in the tail end of a LM317.

There is a driver circuit here that might help prevent such failures:

diliev.com/home/materiali/OLUP/0470402296_Electro-OpticalA.pdf

and here:

die4laser.com/dvd-rec/Die4Drive.htm

Steve

[withlasers]

My strongest guess is that the constant on-off of the potmeter is what messed up my circuit.
I got the new diode delivered, but turns out I didn't burn the #2 one while soldering, the multimeter accused open but I guess it can't handle the voltage drop of the laser diode... So now I have 2 diodes again
Just put everything back together now, adjusted the heatsink so the lens can thread in and out of the focal point (before it was just out of the focal point).
And I'm back to square one hehe where I was 2 weeks ago.
Built a straight dxf->ILDA converter with claude's help, now polishing some final aspects of my main program to test it and hopefully next week I'm at the factory with it, testing it in production.
Wish me luck! I'm making pictures and videos as I can and will share more as I have time, the last two nights I was up until 1am 3d printing lens seats...