ฝัง
- เผยแพร่เมื่อ 12 พ.ค. 2024
- In this video I design and build myself a Spin Coater... a tool which is used to deposit thin films as thin as a few dozen nanometers! Truly nanotechnology at home! Some design files, stl files, schematics, code and the rest can be found on my GitHub and Printables pages
GitHub: github.com/BirdbrainEngineer/...
Printables: www.printables.com/model/6589...
Some extra points to note about this spin coater... It's not the greatest spin coater out there... the biggest problem I found is that the PC fan I used is kind of bad. When hitting the minimum duty cycle (which is 10% rather than the advertized 5%), the speed always spikes up to nearly 2000rpm... so it's unfortunately impossible for me to do a slow acceleration. I think a better way to do it would be to use a 3-pin fan and control the speed by directly using the PWM on the power of the fan... I'm just not sure what that would do to the tachometer signal.
Music:
(0:00 - 0:47) The Plan's Working - Cooper Cannell: TH-cam Audio Library
(0:48 - 1:57) Smooth Waters - SergePavkinMusic: • Smooth Waters - Serge ...
(2:00 - 4:00) Drift Ashore the Dreaming Island, Link's Awakening
(4:01 - 6:07) Gerudo Valley, Legend of Zelda - (Cover) Luke Pickman -- / instrumentmaniac: • Legend of Zelda - Geru...
(6:10 - 7:32) Kokiri Forest, Ocarina of Time - (Cover) The Noble Demon: • Ocarina of Time: Kokir...
(7:46 - 8:22) Sci-Fi-Dramatic-Theme - Twisterium
(8:34 - 9:47) Tittle Screen - Sawsquarenoise
(9:51 - 10:45) Investigation, Phoenix Wright Ace Attorney: Justice for All
(10:47 - 11:53) Main Theme, Final Fantasy III
(11:56 - 13:55) Wind Scene, Chrono Trigger - (Cover) Malcolm Robinson Music: • Chrono Trigger - Wind ...
=|Chapters|=
0:00 - What is a spin coater?
0:47 - Spinner assembly
1:57 - Microcontroller and speed control
6:09 - Tachometer
6:46 - Peripherals
7:10 - Finalizing hardware design
7:54 - Software
8:15 - Assembly
9:49 - Usage
10:45 - Testing
14:00 - Files and Thank You - วิทยาศาสตร์และเทคโนโลยี
![[TH]FFWS SEA Spring 2024 Knockout Stage - Day 17](http://i.ytimg.com/vi/f6Zd-OGk278/mqdefault.jpg)
Some tips on using the 7805:
-the precise values of the electrolytic capacitors dont really matter at all, they just smooth out the voltages.
-you accidentally did a really good thing by using an electrolytic capacitor AND a ceramic capacitor. The electrolytic capacitor is great for smoothing while the ceramic capacitor is better at filtering higher frequency noise
-consider putting a diode from the output to the input pin. When you remove the input voltage and the output still has its capacitor charged, you could get a higher voltage on the output than the input. They dont really like that.
-the arduino has a voltage regulator included, you can input 12V on the "VIN" pin. Be aware that you shouldnt draw big currents from the 5V rail in that setup.
The ceramic-electrolytic capacitor combo was not really an accident, figured exactly what you said.
So, the regulators need a flyback diode? That's interesting, hadn't heard about that before.
Most arduinos (including the nano versions) do have a voltage regulator on them that can use 12v as input, yeah... but it's not always guaranteed (chinese clones can have some wrong regulator on them) and I realized I would want to use the Pico pretty early on, so the regulator was going to be there regardless. Just worked out a bit better for the "story" of the video to talk about it earlier rather than later ;)
@@BirdbrainEngineer its not really a flyback diode i think? You can find it if you scroll down the datasheet to the recomended circuits. Lets say you plug your pico in to upload code, then you have 5V on the output side and 0V on the input side of the chip. Its not really nessesary, but its good practice.
The rest of the circuitry (apart from SD card and manual pwm control potentiometers) is prevented from being powered over usb through the Pico thanks to diode D2 (this is a required diode). In-fact that single diode means that it's even safe to plug the Pico over usb even if the rest of the circuit is powered off 12V... Infact I regularly had that happen, and nothing broke...
@@BirdbrainEngineeryup just checked, you are right, i am sorry😅 Its genuinely a cool project and also very refreshing to see someone actually know their electronics!
You earned a sub for that.
(That and for our shared hate for FreeCAD)
Well done! No fluff and all stuff, just what I'm looking for. This video surely deserves more heat from the algorithm.
@09:16 Yep... Had a look at it for potential employment at a company that was working with it which was baffling to me how someone was willing to suffer through such an over bloated but undercooked piece of shit while at the same time having like dozen Altium Designer Seats that cost a fortune. It still blows my mind thinking about that conundrum.
Except that is free
Cool stuff, I love open source hardware like this.
It inspires me for my projects.
Finally, spin coaster made not by just glueing some magnets at lead of the fan :D Keep up the good work)
As someone who learned cad more or less in free cad, I always find it entertaining how it drives everyone else using it for the first time insane.
I assume that you were fighting with the classic face naming and rebuilding issues that free cad has.
It's much better than it used to be (like a decade ago when I started using it) though it's still a little raw, give it another few years to cook and it'll probably be usable for first time users coming from other programs(especially if they fix the face naming, which will solve a lot of rebuild issues too).
*freecad users have the same cope as the linux desktop users, next year will always be the year of freecad going mainstream like blender*
At first I ran into the topology naming problem or whatever it was called... but the workaround with datum planes is... fine-ish, though realistically, it really shouldn't be a problem in the first place. But FreeCAD is also just very clunky to use in general... While I say I'm not going to use FreeCAD again... I'm also not sure whether there is a better solution that doesn't cost an arm and a leg to use (I can't use a student license like many other maker youtubers seem to be)
Heard good stuff about alibre, might be worth a shot, since it only costs a couple fingers rather than a full limb. Free 30 day trial and no forced updates/subscriptions.
I haven't had the time to try it, so idk what it's really like unfortunately. Well I would probably be fine with it since my standards have been 'well it isn't crashing' for a real long time.
@@BirdbrainEngineer
@@BirdbrainEngineerwe use Fusion at work and are probably switching to OnShape. Both are pretty good and free if you're willing to put up with the usual corporate bullshit. I love the open source ethic of FreeCAD, but the software just isn't usable compared to commercial CAD. Hopefully soon it will be.
@@evanbarnes9984 OnShape is great, but the moment you make a single penny from what you make in it you have to shell out for a license, which really sucks.
@@evanbarnes9984 IIRC, OnShape was cloud-only? Kind of don't like that.
insane work
Amazing ! Keep pushing these designs
LOL, so this is what you were building. Very nice result !
Indeed :D
Thanks!
Well, I don't know with who to share this, but I was wondering if the spin coater would allow for a nanoscale 3D printer?
The idea would be to spin the resin in a transparent plate, apply the UV light, then use the laser to take out the undesired parts and repeat the process.
However, this idea sounds kinda wasteful, and the only different manner I could think of would be using a coating and laser removal on every layer, leaving behind space for the resin.
This would avoid the waste of resin, but it would still need for a proper planning for optimal coating removal (if you want to remove it at all).
There also the problem of align everything up after the thing was spun (at the nanoscale). But _maybe_ it could be solved by marking the plate and aligning it with some kind of sensor?
What you explained is a process called photolithography, and is one of the most important processes in enabling our modern world - Integrated circuits are made in a similar way!
Spin coating is geared more towards uniformity rather than thinness in coating. However I'm not sure why you need a laser ablation step, in an sla printer the substrate is dipped into the resin bath, in this case you'd be spinning the resin then dipping the part down in to it before exposing it with uv, however your z axis resolution would really destroy the practicality. You can't really flip it or the features you're printing would interfere with the spin coat leaving you with uneven garbage. Overall spin coating isn't really applicable to resin printing as far as I can see.
Incredible (also +1 for Chrono Trigger music and the way all the music is laid out in the description)
love your channel keep up the awesome. designs
Nice work!
The ATmega328P is capable of outputting a 25KHz, width-modulated signal (up to 31+KHz in fast PWM mode)... Just don't use the Arduino's so called "PWM" functions... Those are intended for low-pass analog control.
Prefer it a decent timer library, or better yet, use an online AVR timer calculator and set the registers values yourself.
Arduino's dedicated pwm stuff is not able to do exactly 25kHz, but you can get the 25kHz pwm if you use a timer interrupt, which I did do in the video. But an Arduino is just generally a worse choice in this case due to the speed control resolution. If I ever make a more advanced version of a spin coater then it would probably use an ESP32 as the microcontroller.
Awesome project! I don't have a ton of electronics experience so I'd love to know why the push/pull setup you mentioned would be a bad idea, or what led you to your specific choice of transistors in the project. Cheers, looking forward to your next video :)
As far as I was told, the push-pull setup *could* work, but it could also damage or destroy some of the internal circuitry in the fan... and because I did not have a second fan to tear down to reverse engineer, I did not want to take the chance.
The only specific transistor I chose for the project was the old BC547A, because it had better switching characteristics in conjunction with the fan's pwm pull-up, but technically it also isn't needed and could simply be a 2N2222A or something. The rest of the transistors were just ones I had easy access to... they can be replaced with similar/equivalent ones.
@@BirdbrainEngineer right on! thanks for the reply!
Awesome project! And even better video presentation! I was engaged the entire time!
Would putting the whole thing in a box and putting that box in the freezer help with the acetone evaporation rate?
Maybe if the whole coating process was done in a fridge/freezer then the acetone would behave... don't have a spare fridge/freezer to try it though
@@BirdbrainEngineer if you simply pre-cool the acetone and the glass in the freezer, can you work fast enough to put everything together outside the freezer and turn the spinner on before it warms back up? The glass might provide enough thermal mass to keep the acetone cool for a few moments.
this is SOOO cool!!!
Cool stuff. PGMEA is most common solvent in spin coating photoresists, also ethyl lactate and cyclophenone is used. If you need your layers get measured properly let me know, I have both profiler and while light spectromter available
Excellent work! I'm not sure if it'll dissolve acrylic, but toluene might be an option.
I don't think toluene dissolves acrylic... it's also a fairly nasty solvent, so I am getting that anisole.
Loved the video! I'm sure you've considered it, but PETG shouldn't react with acetone, at least at the concentrations you're dealing with. And I'm sure this would add a huge additional layer of complexity, but it would be great to add a thermoelectric element to the spinning chuck. There are very small ones available. Configured correctly, you could achieve heating and cooling of your components.
The spray catcher is made of petg, and technically everything in the spinning chamber could be made of petg no problem for extra chemical resistance. As for a cooling/heating device like a peltier device, it would require building a more custom spinner assembly as opposed to using an off-the-shelf pc fan :P
Excellent build! Have you tried using DMSO, isopropanol or ethanol as a solvent?
For DMSO it appears that post processing of the sample would be needed (heating) to remove the solvent properly. Isopropanol and Ethanol do not dissolve acrylic.
@@BirdbrainEngineer Yes, heating (1-3min at 90-100°C) the sample after spincoating to drive out the residual solvent is quite standard, but this would probably give you a nice flat coating. Might be worth a try. Acrylic has a lower solubility in IPA and ethanol at room temperature, yes, but it might be enough for the spincoating. Good luck!
Any merit to dispensing the liquid only once the platen has reached the desired rpm?
That's how it's done in a fab, the wafer spins up, then an arm with jets turns out over the wafer. It then spins a while, sits a minute, and gets loaded in a drying oven before loading into the lithography machine.
It is also possible to do it that way, yes... But I'm pretty sure it wouldn't matter for the acetone solution. Once I get anisole, I will test that way out as well. The tiny problem with my setup is that I don't have an automatic liquid dispenser so the inconsistencies in the coatings could be greater.
This is awesome! Just had a quick question - what's the point of all the series diodes? Are they largely just for MCU protection?
That is the idea, yes.
What about temperature? Acetone should evaporate slower in lower temperatures. Try putting solution in the fridge or freezer
Would need to perform the spinning in a cold environment too in that case. The tiny amount of acetone that is used for a single coating would heat up extremely quickly. In addition, one of the reasons why the acetone evaporates so quickly is because as the solution is spun, it thins out to be only micrometers thin on the substrate - that amount of acetone has very little heat capacity and would still probably just evaporate very fast.
The good news is that soon I will be able to use anisole though! Might make a shorts video testing out the spinner again with the better solvent then.
an easy way to test would be to refrigerate the slide being spun coated and the acetone mix, that way evaporation takes longer and the plastic housing acts as an insulator anyways, keeping it colder longer. It'll affect the time required to spin coat though, however with the difference in vapor pressure between room temperature and freezing, only being 4x, it might not be enough of a difference given the massive surface area. It would however be easy to try.
Wouldn't it be better to apply to mix with the substrate already spinning ?
It most certainly would be if it was possible to deliver the solution in a repeatable manner. I left a hole in the top of the cd spindle case to do exactly that, but I found that with the acetone solution, the results were somewhat even worse if I applied it during spinning, probably because I did not deliver it quite exactly in the center and at a constant rate.
@@BirdbrainEngineer You should be able to do it with a peristaltic pump
@@ilovetechnology8436 Something for v2 if I get to it one day
Could we directly use the spin coater on windows instead of the homemade graphical interface ?
I am not sure what you mean... Do you mean if it'd be possible to interface the spin coater to a PC running Windows? If so, then the answer is that technically you could, but one would need to write quite a lot of extra code to facilitate some sort of protocol to talk to a program or webserver on the PC. If you mean that whether the spin coater itself could run Windows, then the answer is no; the hardware is far too weak for that, and even if you used a proper raspberry pi to run say Raspberry Pi OS (Linux), then you'd technically lose the RTOS capabilities of dedicated hardware, which may or may not work out.
Having used freecad in the past i feel your pain
I hate to be "that girl" but at 5:09 in the video the CMOS Buffer is incorrect, those won't behave as expected due to body diode conductance and how pMOS and nMOS devices function (Vgs must be negative in a pMOS to conduct and positive in an nMOS to conduct). A CMOS Buffer is traditionally just two back-to-back CMOS inverters due to this as there is not really another viable method for a simple one stage buffer.
Also your CMOS Inverting buffer, the top pMOS is pointed the wrong way, the pMOS and nMOS drains should be facing each other.
Other than that, very interesting vid!
Ah I suppose I drew the transistors the wrong direction... My mistake, I do not yet have a good intuition for how transistors are drawn, so can expect more of these mistakes unfortunately...
And sure, a cmos buffer is usually 2 inverters, I suppose I should have straight out said it is the actual normal way to do a "push-pull"... I was trying to just cheekily get across the point that as long as you keep in mind that your signal gets inverted, an inverter can also be used as a push-pull.
@@BirdbrainEngineerI'm curious why you didn't go with the push-pull actually.
@@xxportalxx. Mostly because I trusted the suggestion of a friend who has way more electrical engineering experience than me.
feels like your pursuit of 1 hz precision control was... misguided.
I totally agree, I can't imagine the control circuitry of the fan actually uses a resolution that fine.
Think what you will, but the spinner bounces about 3 or so RPM around the setpoint, as opposed to the 20 or so RPM with the Arduino...
I disagree. If you're trying to get flat and precise surfaces at the nano scale, you do need precise control over RPM. The fan speed controller may not be capable of that, but it's a worthwhile pursuit. Besides, why not do it? It's fucking awesome that he pulled that off, and very cool to see the process.
@@BirdbrainEngineerWell 20 rpm is really not much considering the 7000 rpm it spins up to. You'd get 20 rpm more just by simply turning the whole spin coater by hand once every 3 seconds lol
@@evanbarnes9984*she
Lab hardware prices are beyond anything reasonable.
I agree FreeCAD isn't great but I'm so "Stockholmed" I refuse to switch.
Calling a free cad package as powerfull as FreeCAD a heap o garbage because it's not easy to use(and i agree it's hard to learn and does weird shit form time to time) is wrong. And yes, syndrome is strong :P
I'd had used a HDD motor.
I thought about it, but it would have required a fair bit of extra work to control it (and moreover to control it to a good speed resolution)
@@BirdbrainEngineer you can use a field-oriented control (FOC) IC, or SimpleFOC with a fast Arduino board (e.g.Teensy), but yeah that would have taken more work than simply using a PC fan's built-in speed controller.
@@PixlRainbow Well, Teensy is not an Arduino of course (by that logic the Pico is also a faster Arduino :P). Using a proper BLDC with a high-precision FOC would of course be the superior choice to use, but it would make the cost a lot higher and would require more engineering effort.
@@BirdbrainEngineer I didn't actually know the teensy wasn't part of the Arduino family as I've always seen it sold with Arduino branding (in shop listings), my bad
@@PixlRainbow Hmm... maybe Arduino has made their own specific teensy dev board by now, but as far as I am aware, Teensy has been an established platform on its own... it does have an "Arduino-core" made for it, though... which is to say that you can program it with the Arduino library (as long as you set up your IDE to support it - many solutions like PlatformIO support doing so out of the box). In that same manner, I used the Arduino library (and other libraries that rely on the Arduino-core and Arduino library) to program the Pico for this project.
Very good project! But the video would be much more "enjoyable" without the really bugging "music"!🤪
I'm hoping AI tools will allow FreeCad to resolve their "Topological naming problem" in a timely manner now.
it's already being worked on and has somewhat been solved. I don't really know what relevance is bringing AI to this topic. This is not a big data problem