The holy grail of ferrofluid displays (Carl Bugeja Collab!)
ฝัง
- เผยแพร่เมื่อ 6 พ.ย. 2021
- This 12-Layer PCB coil made by Carl Bugeja creates a magnetic field strong enough to lift Ferrofluid!
Video Sponsor: www.PCBWay.com
Watch Carl Bugeja's part of this collaboration: • PCB Electromagnet vs F...
--------------
Twitter: / appliedproc
Facebook: / appliedproc
Instagram: / appliedprocrastination
--------------
Credits:
Intro/Theme song:
- Nicklas Myhre: / myhre.nicklas
Music:
- Same Time - Spence (TH-cam Audio Library)
- Patience - Dyalla (TH-cam Audio Library)
- Summer in NY - Dyalla (TH-cam Audio Library)
- All Stars - Patrick Patrikios (TH-cam Audio Library)
- Distance - Anno Domini Beats (TH-cam Audio Library) - วิทยาศาสตร์และเทคโนโลยี
Good job Simen! It was so satisfying watching my coils working with your ferrofluid tank!! 😁
We totally agree! It was super satisfying to see your PCB do some seriously heavy lifting :D
@ __
Applied Procrastination It feels so great to see you both in this way 🍻🍻 More plz
@Applied Procrastination Same here
Why not make a large board with multiplexed 12 layer coils and than fabricate a CNC shaped steel pixel plate which will form each pixel around the coils and stick that behind the display?
Would it be possible to bring the 12 layers to 6 pins so the magnetic strength can be adjusted by powering each coil pair individually?
Great work Simen & Carl!! I see the beginning of the Ferrofluid display revolution starting here ;)
So close to solving the rats nest of wires!
The magnetic field lines in the EMs are shorter, they go from the core to the housing. While in the PCB EMs, they go from the steel puck to the front of the PCB.
Longer filed lines -> weaker magnetic field (for the same flux). Also having a steel puck with less area (let's say, same diameter as the coils) would make the lines tighter and help a bit.
I think you could make it work if you were able to place each PCB EM within a steel pot core, like the off the shelf EMs (en.wikipedia.org/wiki/Magnetic_core#Pot_core )
you can easily solve the ratsnest by just not having each pixel connected via wire but make a large backplane PCB.
I think the wire option only makes sense for prototyping but in a final design you'd solder the (traditional) magnets on a PCB with for example IDE cables going to the controllers or having them on board or even as plug in module
Seriously UNDERRATED channel 😇
Couldn’t you work with Carl to print a MATRIX of MULTIPLE coils on a single large PCB (ie: the size of your display)? That eliminates 99% of the external wiring and 99% of the labor to build the display. Also, the PCB might be able to absorb/conduct away some of the heat generated by the coils - especially if a copper layer was added to the PCB for the purpose of being a heat sink layer.
Further cooling, if still needed, could be added by placing some Peltier solid state cooling blocks in physical contact with the PCB and/or heat sink layer of the PCB.
The Peltier coolers could then “pull” the heat energy from the PCB/copper heat sink layer and transfer the hear to heat sinks mounted to the hot side of the Peltiers.
(Just some thoughts).
Before I forget, GREAT content guys.
This channel and carl Bugeja are underrated
Did you thought about getting rid of the transparent back panel of the tank and use white PCB as a back panel instead ?
The electromagnets would be much closer to the ferrofluid and you would be able to drive the coil much harder as an the PCB would be directly sandwiched between the liquid and the iron plate.
I imagine that would be difficult to seal completely and also hard to prevent the PCB from getting stained by the ferrofluid.
@@AppliedProcrastination What if you coated the PCB with a hydrophobic waterproofing spray to keep it from staining? Or laminate the PCB with a really thin sheet of acrylic? Get a really thin sheet of acrylic and glue it to the PCB using something like UV resin. Put a big glob of resin in the middle of the PCB and slowly lower the acrylic sheet down. It’ll spread the resin, or whatever adhesive, out evenly without air bubbles. Use a roller or squeegee to push out any excess resin, then cure it with UV. You’ll have the stain-proof properties of the acrylic and the PCB will give the thin acrylic the same/similar rigidity to the thicker acrylic without the extra space between the ferrofluid and magnets.
@@AppliedProcrastination conformal coat the PCB? Just a little UV discoloration would be worth the simplification. Plus might look cool 🤩
@@stormbreaker932 Having failed on a similar idea, anything other than glass will get stained by ferrofluid, no coating i tried will permanently resolve this.
@@AppliedProcrastination Have your boy check this out. en.wikipedia.org/wiki/Halbach_array . A circular coil like he is creating is the least ideal setup. He could easily make a coil many many times more powerful than this thing but simply arranging a bunch of smaller coils properly and considering he is making 12 layers that should be more than enough room to get some unusual magnetic fields going. It makes a huge... HUGE difference and allows focusing of the magnetic field all onto a single point.
You might want to consider using small PCBs for the inductors to get rid off the wire jungle.
You should add current measurement to the inductor drivers (only one current probe to the whole machine could work too!). This way you can automatically measure if there is the correct current, and the inductors is connected.
Or a single LARGE PCB
One idea would be to use a steel insert going through the pcb (like at 1:49) with a magnet at the back and adding one or two other 4 layer pcb on top to get 2/3 time the number of turn.
You could stack the PCBs with through hole header or if there is to much interconnexions by reflowing them together.
Also I think having the steel in contact of the glass will make the magnetic field inside the tank much stronger.
The problem with adding more layers is that you also add more resistance. The magnetization strength of an electromagnet (the H field) is proportional to n * I where n is turn count and I is current. When you add more layers, you get more turns, but proportionally you also get more resistance, which drops the current. It ends up cancelling out. My team recently made a 40 layer prototype PCB coil, and going from 20 to 40 layers basically did nothing to change the strength.
@@billfan7528 Yes adding more layer and turn count will increase resistance, but it all comes down to copper mass and cooling. If the resistance go UP you can still increase voltage to keep current the same or paralleling some layers as long as you can pull the heat out.
Was going from 20 to 40 coils done keeping the copper mass the same or cooling interface the same ?
@@MrThecoco974 our pcb coils were kinda fundamentally flawed in that we used 1oz copper, so the 40layer stack with 550 turns had 160 ohms of resistance. By contrast, a hand wound coil with 550 turns and magnet wire had 20ohms, lol. So to hit 0.35A (what we found was needed to float the ferrofluid) we needed something like 56v... Lol
@@MrThecoco974 we tried going up to 36v and I think it desoldered some pads between some of our boards
You can increase strength by using Halbach arrays for the configuration. When used in matrices as 2D arrays it would increase the strength with symmetry and maybe easier for your project application than single e-magnets. Additionally, use SMD Ferrites as surface mounted cores, which you can set over the surfaces as arrangements; they will increase the effective permeability, thus increase the field strength. Aluminium as a back sheet will act as mirror to magnetic field and help with cooling.
This channel's gonna blow up.
if you give reverse polarity to the pcb coil the it will weaken the effect of the behind installed magnet and relase the fluid much efficent, also you can give it a pulse of high current to do that, also you can use a pulse of high current to lift the fluid initially, then you can decrease the current when the fluid raises to the desired ( or even close ) level, and you can do that using simple microcontroller.
The "correct" solution to the rat's nest is probably to get some sort of off-the-shelf push connectors (maybe even just regular pin headers), solder one type to the coils (small pcbs would be useful here, just use some sort of glue or something to mount them to the coil), solder the other type to a rigid backing (or you could solder them to small boards, and mount those boards to the rigid backing), and use a grid frame mounted to the rigid backing to hold the coils in place. Replacing a coil would just be a matter of pulling the old one out, and slipping a new one in.
Very cool and small electromagnets are in old cash registers with needle printer.. They are, usually, 8 or 9 in row, soldered to small pcb and with common metalic insert..
Other option - disassemble car relays from scrap yard, leaving only coil part and maybe base. There are small ones too..
For removing wire nest, solder these in X Y matrix (don't forgetting protection diodes), like in keyboards, LED displays, with axis connections near coils.. 😉
You could probably get a good improvement just by making the glass thinner. Perhaps using something made for LCDs. Or even a thin sheet of white plastic, to combine the background and container in once piece.
I have seen PCB mounted coils used for brushless motors on floppy disk drives, although they were surface mount and I don't know if they are widely available still.
pcb electromagnets are the future
Make a stack of PCBs with holes in the middle and put a piece of iron down the middle. You have effectively made a conventional electromagnet.
I have code to produce arrays of pcb coils in kicad that could be easily have the centre holes added if you are interested.
Quick Idea: If you only want to display time you in theory only need four 7-Segment displays. What if you didn't use electromagnets at all, but rather move long Cuboid permanent magnets closer to/away from the tank to draw up/let go of liquid
That's a good design for a clock indeed, but we opted for a more general purpose display when we started our Fetch project :)
@@AppliedProcrastination And what if you would use pcb electromagnets to move permanent magnets to and from the glass, which in turn attract or release the ferro fluid? A bit similar as using a leaver... you could even use weak springs (non-ferro) to make sure the magnets stay in a specific position when not in use.
Definitely depends on what you want to create in the first place
what about making a super thin tank? like... 2mm or something. Would that have an effect?
Wow that tank is 2 years old here? Awesome!
Any updates??! It's been a year!!
Nice video, keep it up, thank you :)
Great video! I suspect you have already thought of this, but have you tried using a stronger permanent magnet instead of an iron “core” and then partially neutralising it by driving it in reverse?
Yeah, that does work, but it’s an approach that’s full of trade offs for small coils like ours. Basically you have to use quite weak permanent magnets to be able to cancel them out/turn them off with the electromagnet - all while also consuming more current than a normal electromagnet (for the same reason). Plus, we have more negative pixels at all times than positive pixels, meaning inverting the driver circuitry would increase the overall load. It’s a great concept for things that basically never need to be turned off though! One very common application is holding fire doors open until the alarm goes off, or holding doors closed until someone beeps their access card, etc. :)
@@AppliedProcrastination maybe it would be possible to balance such a setup so that to lift blobs up you would have to drive the coils; the permanent field being *just* strong enough to retain a small blob, and reversing the coil field to release? That would reduce current consumption, but I imagine getting the field strength just right would take some work...
And of course a tri-state driver circuit is more complex - hmm... addressable LED strip-like PWM setup might work - the ICs for that are readily available I believe, though drive current might be too low unless you find something meant for giant billboard-type displays. I/O would be really simple though.
Watching this video back right after bitlunis Flip Dot Oscilloscope video.... These Flip Dots are basically PCB mounted electro magnets... Not sure where to source these though but maybe a hint ;) Great colab and video :)
I'm wondering if you've thought of using two smaller coils wrapped in opposite directions wired in together with an iron backer. I think that a configuration like that could be more effective since the ferro fluid could bridge better the two poles.
Why not a thinner container for the ferrofluid? That way there's no need to elevate a big amount so you don't need a big magnet, and ferrofluid is opaque enough to show itself as dark if it's displayed in a layer.
Mainly because it looks cooler with spiky/3D ferrofluid than something pressed flat against glass, but the idea would probably help a bit!
You could use a stronger permanent magnet, and a magnetic switch to toggle the magnet on and off.
e.g. A strong north facing magnet, and a south facing electromagnet would cancel out magnetic fields. When you flip the direction of the electromagnet, the magnetic field will be twice as strong.
If you haven't tried already perhaps parallel wound series connected bifilar pcb coils might be helpful.
nice project! did you think about moving pernament magnets using servos? it may be too bulky but 90g servos cost arround 1-1.50 USD it would be probabbly slow but maybe it is worth to try, you can adrust the magnetic field by distance and in case of clock it would consume more power only once per minute while rearanging
Could you drill holes length ways into the iron backs and have internal arterial water cooling to compensate for higher voltage? Too bad the ferrous material doesn't conduct heat better
Haha, that's probably possible, but it would also probably stop making sense from a complexity and cost reduction perspective.
Damn I would love to try and run a ferrofluid reacting to some of my music. I am curious how that would go.
you need to put the driver transistor and a driver chip on a small pcb on the electromagnets, and then drive them like neopixels are driven. maybe the same kind of control chip thats used for leds and daisychainable would be enough to switch a transistor and drive the EM. then only need like 5 wires daisychained
Now just have a single pcb matrix of them made :)
Put iron pole pieces in front of the PCB coil as well as the disc behind. One disc the size of the coil, the other a ring inside diameter same as the disc plus tiny airgap, outside diameter same as the back disc.
That would give a display of circles, not dots...
I think it would be an amazing project if you could make a display that has a single pcb. pcb electro magnets, voltage regulator, processor and all.
Would a heavier clear fluid like gliseran work to slow the movent of the ferrofluid ?
I'd have the ports on the magnets line up with the PCB, them you don't need long wires
^ oversimplified
You could also have a cutout whether magnets or sit in place, then another contact board behind them to route all the connections to the appropriate place, and then another board holding all the controllers.
It's quite the redesign, but it could also make it a lot easier to manufacture and sell them, once perfected
Love the idea ;3
If you print a 3D jig, you can add magnets in front of your electromagnets so then can attract or repel them with the existing electromagnets. This way you can improve efficiency.
Another idea for your Clock: use just 1 electromagnet and a x and y axes motors to move around repealing or attacking magnets in your 3D jig.
I wonder if the magnetic force required lessens, if you put the glass at an angle, so the ink is pulled up a slope?
also, maybe stick the pcb directly into the tank with the ink? or make the glass thinner? magnetism was inverse-square-distance-thinggie?
Try using ferrite as a backing, might be a good alternative to iron.
what about using a pump to lift the ferrofluid up to the top and "drip" it down across the display like rain. The stream of ferrofluid should stop on the magnet locations, and then release to fall, creating a visible number, while also creating a very entertaining and calming effect.
Hey, why stop at iron? Why not use something with a higher magnetic permeability like mumetal?
Hello, why don't you make your coil control/connections similar to addressable LEDs?
You guys are creating the PCB magnet wrong. You need to not make a circle, you need to make a bunch, its called a Halbac array or something like that where you basically sacrifice one magnets strength at one point to massively increase the strength at another point. It focuses the Magnetic field onto a single point while spreading out on the other pole. Perfect for your use case honestly as it makes a super strong point while eliminating interference on the useless pole
Any reason why making one large 12 layer PCB covered in electromagnets won’t work? Also perhaps modifying solenoid inductors to have more open coils would be worth doing. I know some SMD inductors aren’t too expensive and have a somewhat squarish shape, making them easy to tile. On the other hand, SMDs would be difficult to solder without a big hot plate, and be even harder to desolder.
What about soldering and glueing the inductors to small square PCBs? Each PCB with its own connector, connecting to a large PCB. The large and small PCBs are cheap 2 layer ones, used only for wiring. This would not be terribly expensive. The advantage here would be that they can used their old inductors, and still get rid off the wires.
Could the boards be stacked to increase the field streangth?
Hey, Carl *meant* to start that fire.
I know, we just wanted to show what can happen if these aren’t cooled down sufficiently
Hi you could make a pcb with though holes bigger than the wires to pull them through then have pads to solder then to on the back maybe in 12x12 segments should tidy up the wiring
(Duh, I just edited out my brilliant idea I’d stacking coils to get more power, after watching Carl’s video where he explained that heat dissipation is the problem :-/)
I do think though that your specific use case is quite different than Carl’s general interests, so there might be some different directions you could go. I need to think about it a bit more…
I do think it would be much more effective to go ahead and leave a small hole in the middle for an iron core (or even a small “bias” magnet used as a core). That far in, I doubt you’re getting a lot of effect from the tiny little turns on the PCB, but the increased permittivity of the core would concentrate the rest of the field greatly.
(Hmm - I wonder: I think you might get a much more concentrated, locally-intense field if you went with a steel slug significantly smaller than the coil diameter. I don’t recall enough of my college electromagnetics to say for sure, but I think the field lines will be “drawn into” the core, vs spread out it it’s larger in diameter. What you want though, is to maximize the flux density in a small area, vs spreading it out with a larger “pole piece” - you’ll note that high-flux magnets for laboratory use often have tapered pole pieces, to concentrate the available flux into a smaller area.
==> How about trying a small piece cut from a steel bolt or rod instead of the large-diameter slug and see what happens? I think that the length of the rod might have some effect as well; AFAIK, a section say an inch long should create a field that stretches out further beyond the end of it, making it easier for it to attract the ferrofluid.
Don’t give up on this yet! 👍😁
If you would do a 12 payer pcb coil with a hole in the middle and insert into a iron pot core the magbet would be drastically stronger
What if you tried the inverse approach. Having natural magnets that are strong enough to hold the fluid on their on, and reduce their force with the electromagnet by having them placed on and inverted polarity? So you'd have coils turned on on "pixels" that are off, and coils turned off on "pixels" that are on. Would that work?
There are some electromagnets like that (called de-gauss EMs), and the main issue is that you need the electromagnet part of it to be equally strong as the permanent magnet in order to cancel out the magnetic field. So as a result, their permanent magnet is typically fairly weak (at least in small ones) while the electromagnet draws tons of current. It’s great for applications where they basically never need to be used (like holding fire doors open), but not too great for a display where we want most pixels “off” at all times.
Source of a small THT electromagnet?
Have you tried vertical THT inductors?
Also, it's possible to make your own using ferromagnetic cores, plastics and drill's.
Don't know about strength.
Oh, btw, go aggressive on glass thickness.
0.5 mm or less sounds good to me.
I'm new here, I wonder if 'spherification' (like in molecular gastronomy) of the ferrofluid would work. Maybe smaller beads are easier to manipulate than big blobs :)
Smaller beads are easier to manipulate, but big blobs are cooler to look at 😄
@@AppliedProcrastination fair! haha
Do you still use simple salt water, I wonder if some denser liquid would make ferrofluid easier to lift, but Im guessing you already gave it a lot of thought
Salt water is the easiest and best solution we’ve found that prevents the ferrofluid from staining the glass (which has been our number 1 priority). The high buoyancy has been a nice side effect
Бессмысленный комментарий для продвижения видео) очень интересно, надеюсь что Карл что-то придумает!
Try to use a transparent liquid with a similar density.
Have you tried sandwiching two 12-layer PCB coils together yet?
Could probably even flow the solder through the holes for the wires, to connect the coils in parallel.
Have you considered using the electromagnets array of a flip-dot display?
We have not. I’ve assumed they are multiplexed, clunky and fairly weak, but now that you mention it I’ll have a closer look :)
@@AppliedProcrastination Look at the latest Bitluni video. There is a closeup on his flip-dot display, the electro-magnets look very small. th-cam.com/video/8DvH6FiS3sg/w-d-xo.html
Have you tried using through hole inductors?
Yes, the ones we found at prices that were cheaper or as expensive as electromagnets were not efficient enough
have you tried paramagnetic fluids?
No, I had never heard of that before now. It seems quite cool! Maybe we’ll look more into it at some point!
What happened to these guys? I was interested to see further progress since this video since it was made two years ago, but it's still their latest video?
They all died of ligma😔
Lower the viscosity of the fluid and use magnetic nanowires.
Dumb question, but has the iron to be behind the coil? couldn't it be in front as well? That way one could wire the coil from behind and the wires would not interfere
I think the iron would shunt most of the magnetic field lines across itself rather than through as you imagine. Hard disk drives use strong magnets and must shield the magnetic fields from leaking out of the device.
Uhm..nobody has mentioned making this times 3 -- placed on the x y and z axis.. Micro-controllers to act as a catalyst, pulsed to lift for ex.; compensate for z being "gravity" constricted with it being 2-3x the size/power of field. X and Y steer, or pin-point the field. You should be able to 'drag' the fluid into the center of the field, then with modulation you could burst the fluid glob, or whatever. How do ferrofluid speakers do all this? Have you considered getting a ferrofluid speaker and breaking it down to advance your ideas?
why dont you put 2pcb's instead of one ?
why dont you angle the whole thing at 45 degrees?
Hmm wonder how it acts mixed with mercury
Can i contact you guys about something i think might be interesting ??
Sure, but I can’t guarantee that we will follow up on it unfortunately. Our viewers send us lots of interesting ideas :)
Just use your current electromagnets and cut the wires shorter to weld them to a pcb.
What about a magnetic sandwich? PCB magnet | Iron core or magnet or both | PCB magnet
why not just make a pcb backplane onto which you solder the normal tiny electromagnets using very short wires as if they were through-hole components?
you could even build into this board the control circuitry on the other side right behind each electromagnet.
We've considered that and the main issue is that the mechanical support would be minimal (also scary to do such a large non-reversible cable cut by hand 😅). I think there's no way around using screws to hold the magnets in place on the PCB, which would make it really difficult to route all the controller circuitry.
could have a small I2C pwm driver next to the mostfet of each coil, making it only 4 common conductors for all coils, or some other digital protocol that can use more addresses perhaps.
That's maybe possible in smaller horizontal displays, but in our large vertical one we need to drive each magnet individually to be able to lift the ferrofluid higher than the first row. You can check our current PCB design here, which is based on the PCA9685 PWM driver and ULN2803 darlington arrays: github.com/appliedprocrastination/FetchCADFiles/blob/master/pcb/V2R2/Schematics%20-%20FetchControllerV2R2.pdf
yes exactly, just use an I/O expander, control it via I2C or CANbus. Just need 2 communication wires that will be shared between all the I/O expanders, and each of them will control maybe 1 or a few coils. Then you just need to route power to all the mosfets and it should work with only 4 wires coming from the main controller: vcc, gnd and 2 for comm
@@AppliedProcrastination why not make a 3D printed frame that would hold the electromagnets and then connect to the PCB. That way the mechanical stress would be on the frame in the PCB would just be there to handle the electrical connections.
Probably missing something important
But if you ordered a pcb
Spun a bunch of coils yourself
And wire that up?
That’s definitively possible, but incredibly time consuming :) we bought pre-wound electromagnets for our display
You're saying normal electromagnets are quite expensive with 4 dollars a piece and that the alternative is relased at a reasonable price. Carls store lists his 12 layer magnets (coming soon) at more than 20 dollars, where is that a reasonable price and alternative if you want to have 256 pixels?
I was a bit surprised by the 20 dollar price point (it wasn’t announced when the scene you reference was recorded). I’m not sure if that’s a maximum price or will be the final price as the actual cost will depend on how many people buy one.
We did however NOT say that 12-layer pcb magnets are a reasonable alternative to electromagnets in a big display like ours. We said that they are different from EMs and could have some cool use cases that EMs won’t do as well (size being a big deal).
Try taking the ferrous off of this PCB mounted inductor: LDZ2817330K-10
That one is more expensive per unit than purpose built electromagnets. We’ve experimented with some inductors before but couldn’t make them work (for a fair price) unfortunately
Why don't you look at relays? PCB solderable, and they have a magnet in them ...
Eletromagnet compnent that can be mounted to a PCB?
Well thats just an inductor isent it?
why Does the music at 4:20 sound like something I know, I cant put my finger on it and its killing me!
Ok but given the PCB only design wont work why not just cut the leads shorter on the actual ones and solder them into a pcb? Seem's like a straightforward solution to the too many wires and connectors problem. Am I missing something?
That’s what we do in this version: th-cam.com/users/shortseKKamLLzslQ
It’s a really compact and neat solution, but incredibly time consuming and difficult to pull off (the actual soldering and mounting, that is), and redesigning the PCB becomes annoying too when you can’t just disconnect and reuse the magnets but have to desolder them. For small arrays it’s the best solution though, IMO!
@@AppliedProcrastination ahh very cool, I hadn't seen that short. And yeah fair enough it still seems tedious for a large board and like you said reusing them would be an even greater pain. It's too bad no one (appears to) make something akin to a relay socket system but for electromagnets.
I forget if this was mentioned in the video but maybe stacking multiple 12 layer boards is still viable.
Can you make a electric transformer?… plz?
Why not just attach the electromagnet to a pcb, that would reduce the wires.
Why does those shity PCB still don't have a hole in the middle like the original magnetic coils and aren't embedded in a iron plate yet?! The magic coils have this hole on purpose.
these guys got killed btw, thats why they stopped uplodading and being active on social media
💀🪦
pahahahaha smd inductors go brrrrrr
no, seriously. that is exactly what you've been trying to do with pcb coils. and you could make it a cheap 2 layer design with the coil controllers as plugin boards.
Know it's been a while, but for all these DIY ferrofluid displays, why doesn't anyone make the fluid channel very narrow in the 'y' direction, 'thin', glass very close together, so the amount of fluid that needs to be lifted against gravity is minimized. The weight of the fluid is going to grow with area, why make things so hard on yourself with these sloppy fat bottles, since magnetic force is the issue, why not let surface tension and small amounts of fluid be your friend, just like every other 'neato' display from the 80s does, sand suspension, bubbles etc.
Your content is interesting as-is. You don't need to keep over doing the cheesy youtuber humour.
I can't watch it because of the too exaggerated enthusiasm, sorry :(
wtf happened to this channel??