Electronics > Projects, Designs, and Technical Stuff
Multi layer, dual sided PCB Printer.
Siwastaja:
--- Quote from: SiliconWizard on July 23, 2019, 07:44:42 pm ---
--- Quote from: Siwastaja on July 23, 2019, 04:03:00 pm ---
--- Quote from: SiliconWizard on July 23, 2019, 01:21:45 pm ---I suppose you know Voltera: https://www.voltera.io/
and have taken a look at their approach.
--- End quote ---
They specify resistivity as 12mOhm/sq at 70 um. Equivalent copper is about 0.25mOhm/sq. This is 50x worse, meaning completely useless for most purposes, at least without extra precautions during design.
This is something you need to look at with silver ink devices. A huge trap, because such a number is not advertised clearly.
--- End quote ---
Yes, this is pretty bad. But "useless for most purposes" is a bit of a stretch...
--- End quote ---
It really isn't a stretch at all. 5x worse than copper would be a "maybe-ish?" corner case. 50x is totally useless.
In most PCB designs, we really rely on a decades of experience of "good practice" of our own or that of others. So we are handwaving our stuff and adding a good safety margin. Do you analyze each and every trace for current, heating, etc? No, no one does such perfect job unless designing a trivial sub-unit for a space station with unlimited budget.
Yet we tend to mostly succeed and only encounter a few gotchas every now and then. We fail whenever our assumptions are wrong, but they are not wrong too often. This is because we have a lot of margin in our handwaving. This isn't by coincidence - the components are engineered to mostly work in typical conditions, and datasheet examples, appnotes and generic wisdom communicate this idea of "typical". They are supposed to be mounted on a copper PCB with a certain conductivity. The exact value usually isn't important.
But when your track resistivity suddenly goes up, not by 2x, or 3x, not even by 10x, but by 50x, all of the mental rules you follow suddenly change - you are two orders of magnitude away.
Suddenly your puny simple microcontroller circuit's Vcc is fluctuating, not by meaningless 5mV, but by 250mV.
You wire the trendy ESP8266 module up like everybody else to prototype your newest IoT coffeemaker breakthrough? Tough luck, it's pulling spikes of 350mA, and it's already brown-outing due to trace resistance of your 50mm long power trace, even though you used a generous 20-mil trace. dV=420mV for this magical silver ink; dV=9mV for copper.
Power anything just stops working. Including regulators and supply rails to anything bigger than the smallest AVR or PIC circuit blinking an LED at 10mA.
Suddenly anything using more than half a watt becomes "power electronics", with associated difficulties.
Only the least demanding corner cases are left. But as a customer and an engineer, genericity would be exactly what you wanted from such a device.
Sure, you can analyze the shit out of every circuit, learn and accept the new rules, and get working circuits out of it, but really, the device's supposed to exist to enable quick prototyping. Which of course isn't possible if you need to design with completely different rules and knowledge than for production circuits. Or limit yourself to a very small subset of possible circuits. And need to rule all power electronics out completely.
I'm not saying OP's solution would be as crappy as this, they are clearly aware of the issue. Quite the opposite, it seems that "prior art" is so crappy that the market is totally there, without serious competition. We just shouldn't make excuses for crappy products, but concentrate on finding and fixing the showstoppers (ink resistivity clearly still being one of them. Hint: if you can do it properly, that's a great selling argument).
MagicSmoker:
--- Quote from: Siwastaja on July 23, 2019, 09:17:29 pm ---...
Suddenly anything using more than half a watt becomes "power electronics", with associated difficulties.
...
--- End quote ---
On the plus side, you get excellent damping of any stray resonant networks... >:D
***
I think the OP's idea is very interesting, so mainly just replying here to be subscribed to the thread. Carry on.
excitedbox:
When 69% copper is already being done I don´t know why you are suddenly talking about 50X the resistivity. We are talking less than 1.5X the resistivity with current technology.
Pretty much all displays these days are printed using inkjet technology. I am trying to bring the cost of the printers down from the 200k+ point to something that is affordable for hobbyists and smaller companies.
Voltera is very different. They are mixing conductive things with plastic to make conductive paste like those conductive 3d printer filaments. Then they squeeze it out of a syringe. Way too slow and not conductive enough. You can light up a LED but battery life will be crap. They have been working on their ink for years and I already had better conductivity on my first test.
T3sl4co1l:
Have you considered electroless copper plating? This is catalyzed by an activated surface, or autocatalyzed by a metal surface (so it continues to build up while it soaks in solution). Any conductive ink could be used, or ink need not be used at all if an appropriate activation bath can be found. (Which, I think they do this with FR4 commercially, anyway, so this should be practical?)
I don't know how practical such a process is for a printer; you'd probably need a reservoir of starting reagents, a bath where they mix with the board (handling should be pretty easy, just teflon-lined grippers and stuff, if nothing else?), then a waste disposal tank where the spent materials can be stored for disposal (or processed for recovery, or for easier disposal e.g. evaporating solvent to concentrate it for reuse, ion exchange to remove/recover heavy metals (copper), evaporating to solid for dry disposal, etc.).
Hm, I suppose all those things are what the industrial proto printers do already, wouldn't it be? A few extra steps (tin resist, electroplate, etch, strip..), but those could be omitted for a somewhat lower quality process. In which case you'd still have basically the same thing, and consequently it wouldn't really be cheaper, huh? So I guess your intent is to avoid chemical processes if at all possible?
I wonder if PVD something or other would ever be competitive. Drawing a hard vacuum is a pretty high bar to start off with (the hardware alone would probably account for 1/3 your target cost..), but if you can manage that, repeatably and cheaply enough, it's not nearly as messy, I don't think? You'd feed in, either copper plate for sputtering, or wire (to a tungsten/molybdenum boat) for evaporation. Still need a resist and/or etch step though, hmm.
I wonder if a plasma process could be employed for the etching, too? It wouldn't be practical on thick layers though, so you'd have to start very thin (~um) regardless. Which isn't very helpful compared to, say, a conductive ink, which does the same basic thing with a lot less bother!
And that's to say nothing of multilayer, which none of this is any help with.
So that sounds like, working on inks is the right direction, which is encouraging. :)
Flash sintering sounds awfully cool (hot); how do you control the temp rise in the insulator/substrate and conductor? Would that be a matter of picking exposure wavelengths versus absorbance spectrum? Doping with susceptors (e.g. graphite) or reflectors (e.g. TiO2 white pigment)? Alternately, spreading out the power density through depth, by using a much more translucent insulator than the conductor?
If it's not so easy to control, would it be necessary to raise the insulator melting point instead? Flash sintered ceramic or glass would be awfully cool. Harder to handle (brittle), but another interesting product nonetheless -- consider the cost of existing ceramic hybrids. Smaller market I would suppose, but still a problem in need of a solution? Would also allow better consolidation (final step in a kiln; should be a simple enough physical process, at worst needing an inert or reducing gas fill, no liquid or solid chemicals and no additional waste?).
Huh, does flash sintering not have any problems with the substrate curling up like a potato chip?
Anyway, just thinking out loud. This sounds really cool, and I wish you the best of luck on it! :D
Tim
excitedbox:
Electroless plating is how PCBs are made right now. I don´t plan on using PVD so no vacuum or anything like that. The flash pulses are so short that the materials don´t heat up much. It is more about activation energy than actual heat. Between 7-12 J/cm^2 is about right in a 5ms pulse.
Making copper nanoparticles is extremely easy using almost any copper containing chemicals. I have chosen copper sulfate. Using surfactants such as the ones used in soap making, and glycerin or polypropylene glycol which is used in vape fluid and ascorbic acid you can make inks containing copper nanoparticles.
Copper is very reactive though so it oxidizes quickly but during the manufacturing process the copper particles are capped with the reducing agent which slows that process. After printing the ink onto the substrate you can use photonic sintering to quickly cause a reducing reaction to get rid of the oxygen atoms, break down the capping agent, and melt the now pure copper particles together into copper traces.
The above is a simplified description of how it works but that is the basic process with a few spices thrown in. I have capped copper nanoparticles that after being exposed to air for the last month still have not oxidized. I am now experimenting with different sizes of particles and setting up a better test environment to control and test the parameters and do size classification of the particles.
I also want to experiment with the flash wavelengths and pulse lenght/power to get good penetration so that when you print multiple layers they are sintered all the way through and not only on the surface or else you build up a layer cake with unsintered ink.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version