Making a Coilgun - Part 4: A New Revelation
Вставка
- Опубліковано 17 тра 2020
- An explanation of how my old ideas of an altered coil geometry would likely be inefficient and my new ideas on how to design the ideal coilgun.
Support my projects - / levijanssen
Get more frequent project updates - / leviajanssen - Наука та технологія
Now this is interesting, its refreshing to see someone focusing on a different perspective. Nearly everyone just focuses on brute force to achieve power when in reality they should be focusing on increasing efficiency so that they can achieve more power with the same amount of energy.
think smarter not harder :)
both.
both is good.
imagine a coilgun with primary AC power coming from a tractor powered generator, 40,000 watts/300 volts, step up with rectification to DC, throw in 5-10 VM, probably need home made capacitors, optimal coil geometry and projectile weight/length/switching
hard work and talent
Apologies, I didn't see this follow up when I posted my comment on your last video. You've realised a few of the things I mentioned already. I have a few answers for your coil questions.
A little tid bit that might be useful, you wont want to use identical coils the whole length down the barrel. You want the current in the coil to turn on and rise when the projectile front reaches the back of the coil, and then switch off when the projectile is in the centre. The time this takes (projectile enters until projectile centre) gets shorter as the projectile speeds up (i.e. is further down the barrel), so you'll want coils with lower inductances (fewer windings) further down the barrel so that the current can rise to its maximum in the shorter amount of time.
Your pcb seems to take up a lot of space relative to the coil which is what you said you wanted to avoid with IR sensors. To hold the coil together, thicker wire will hold it's shape fairly well on its own, but otherwise you can apply a bit of nail polish or super glue to hold the coils together.
As far as detection goes, one idea I've heard people mention is inductance, the coils ahead of the projectile should have a detectable voltage induced in them from the magnetised projectile approaching them. Never seen this implemented but I've seen a few people saying it should be possible. Also, as far as efficiency goes, have a look at "Brook's coil" it's the most efficient (highest inductance, most force) way to wind a coil for a given length of wire (i.e. for a given resistance, what gives you the highest inductance or strength of electromagnet).
For how many turns you want/what inductance it should have, it should be just enough so that the current in the coil can rise to its maximum before it needs to get switched off. This depends on how fast the projectile is moving through that particular coil and your voltage. Have a look at inductor time constants.
Finally, I'd avoid thyristors, once you trigger them, they continue conducting until the current drops off on its own (i.e. your capacitors discharges, or in the case of a battery it just keeps conducting) which means you can't switch them off when the projectile reaches the center of the coil and you're going to get suck back of the projectile (i.e. it gets pulled backwards, back into the coil as it tries to exit out the front).
Trial and error to find the most efficient coil set up a test where the coil is at a certain height and the bullet is face up on the surface below, then measure how much voltage and current it takes to lift the bullet off the table or make it fall over. This is the best method I have found
Because the metal projectile would become magnetized. Maybe you could use the coils ahead of it to also detect the approaching magnetic field of the projectile (or rather lack of, as I'll explain) and switch when appropriate. That should be pretty fast and accurate. So each coil except the very first would be a sensor initially, and then when each detects a certain magnetic field strength behavior they swap via a transistor to become a powered coil. There might need to be some software fanagling to get it to know what it's looking for instead of just triggering from the activated coil before it, but the way this would be done would probably be to detect when the field strength is DECREASING or has finished decreasing and started increasing, as that means that the projectile has moved to the optimal position to "soak up" the magnetic field due to reluctance and is likely in the region of greatest magnetic strength. If you measure the magnetic field in front of the coil of a coil gun that never shuts down power you can probably get a good idea of what the magnetic field looks like after the projectile has passed the midpoint. Then just scale this to several thin coils, each with there own transistor. To save on the number of analog ports on a microcontroller needed to read all those coils, you could probably have the analog data line of the coil that the projectile has passed switch to the next coil down the line. This switching could even be done as soon as the coil goes from a sensor to a powered coil. I want to make this now.
to make it even thinner, the coil doesn't need a pcb backing -- just make a hole in the pcb such at the coil can be surrounded by it. This way everything is flush!
That's a good idea. Individual coils could be separated by a very thin slice of plastic.
Resin inbeded coil. Similar to "Potting" on sensors and encased electronics for use in aviation and similar uses. Would require multi stage production process. PCB mounted I would guess you'd have a track amperage limit even at a thick copper layer for PCB.
Thanks for the update, couple of ideas, with your winding issue, start with the max voltage you want work with then work back from there to get your optimum coil. could you combine the bus bar as the heat sink for the MOSFET, maybe even have the bus as a square tube to run air/coolant through for cooling. Keep up the good work
I was considering using the busbars as heat sinks. Even if they aren't the intended primary cooling system, they will necessarily have a high thermal mass and inherently act as heat sinks.
To hold the coil in place / prevent it from unwinding, you could use some sort of flexible resin. If I recall correctly, I was considering epoxy, however, the radial induced force would have cracked it. Keep in mind that with high currents, the coil itself will expand radially (another reason why super-electro-magnets are solid metal).
As for the switch, there is no reason to place it within the barrel structure. Just have leads that leave the barrel and connect to the external switching device (besides, I don't think you can get IGBTs small enough to meet the size your are looking for).
For cooling, there's really no solution there, and the coils will get very hot (and potentially melt). Your choices would basically be to submerge the assembly in a coolant (like what's done with super-electro-magnets), or restrict the current such that you can fire it once, and then must wait some set time T before it can be fired again. You do not want to use an aluminum heatsink though, because even though it's not ferromagnetic, you will still induce eddy currents in it with a high enough magnetic field.
For the barrel, a carbon fiber tube would probably lead to issues with splintering, PTFE might be a better solution as it also provides some level of reduced friction. The nice thing about the coil geometry is that it is partially self centering, which could correct for any flex in the tube (you would have to rotate your stages though to correct for the slight geometry imbalance from the way the coil is wound).
Just to correct what you said about the issue with turns... if you have more turns, you have higher resistance, but you simply increase the voltage to keep the current the same. That does have an upper limit, but that's dependent on the maximum current per wire, as well as the heat dissipation, and the breakdown voltage of the enamel coating.
And you are correct about it being sort of like a bell curve. Though it looks more like a reverse Poisson distribution, where it slowly ramps up and then quickly falls off (because of the issues mentioned above).
Încercă și rotirea inițială a proiectilul ui. Vei învinge doar cuplul giroscopic inițial și va avea direcție bună.
Hello, Janssen, Thanks a lot for a detailed theory of ops. I do believe this is like an electric motor "unrolled" into an linear device. commutation is very important in motors and so you will need some kind of a sensor to commutate at the right point. Also, as the projectile speed increases, and there is an inherent lag between the sensor reporting the position of the projectile and the current building up to max in the inductor at every stage, I think the coils towards the end of the barrel should be spaced further apart to compensate for this lag than at the beginning where the projectile is moving slower.
I love this approach. Systems like this are awesome to precision control, but this could be a great application of a genetic machine learning algorithm to control the coil timings based on a muzzle exit velocity.
I was thinking the same thing. Basic machine learning algorithms coupled with sensory data such as projectile mass and velocity (muzzle and intermediate velocities using internal lidar) could optimize coil timings for various projectiles.
@@InfuriatedHawk I'm sure there's some mathematical model that could be built, but it's amazing how close a learning algorithm can get to the ideal state and continue to adapt to changing conditions.
Brian Exactly, a machine learning algorithm could rapidly adapt to different projectile masses and shapes as well as the atmospherics affecting drag on the projectile within the system.
Regarding the position sensing problem, since the gun needs a barrel, (even if it's super thin) it could have sliding contacts on the inside, of course, they would wear out if made of a softer metal than the projectile, like copper. The only other things I can think of are sensing changes in coil inductance due to the presence of the projectile acting as a moving iron core, or using the old method of optic sensors but instead of simply jamming them between coils, a thin fiber optic cable could extend the IR emitters and occupy a lot less space, maybe running inside the coil's PCB. Plus I think not extending the sensors with fiber optics could lead to induced currents on those circuits and eventual destruction.
Carbon fiber is a good choice, but I'm afraid it's not oblivious to eddy currents. I've seen so-called "indetectable carbon fiber knives" been detected by metal detectors because depending on the type of carbon fiber, it is more or less conductive, but much less than lead so it shouldn't be a problem.
I have not made a coil gun but have mulling a few ideas, one is to use fiber optic strands. the IR diode is glued to a photo optic strand, and inserted into a disc used to hold the coil the disc could be cut in three parts end doughnut, two semicircles glued to the former and the fiber optic of the IR transmitter and the fiber optic to the IR receiver, then the other end doughnut. I have a few other ideas floating in my head if you like.
Really appreciate the convo. Definitely makes sense.
ive been super looking forward to an update on your coil gun design
If you are familiar with calculus, and can set up a formula for the parameters you want to (maximize/minimize), and use derivatives to find the minima and maxima.
I still believe in SCR's and capacitors though. Have you looked into the large IGBT's? They are especially good at driving inductive loads. But I have to say I have learned a great deal about strange magnetic fields by calculating the magnetic fields at a point using the Biot-Savart law. It's an enormous amount of number crunching, but well worth it. I wish I had time for more of this stuff.
Increase the efficiency of your coil gun design by using a chemical propellant for the initial acceleration of the projectile, and increase its speed further by the coil gun design.
A sensorless system was my initial idea as well, the most efficient way for a part to work is to not have that part in the first place.
great work, keep chipping away at it man
The best way to cool the transistors would be to move them out of those cramped spaces for easier access. If you are dead set on doing these with modules you could rotate each module 45 or 90 degrees so they won't be cramped and you could cool them more easily, but honestly I'd just have an array of those transistors shoved somewhere else with wires running off to the coils.
In just two videos you teached me a lot of good things. Thank you!
Teach is an irregular verb. So past tense of 'teach' would be 'taught'.
you only need a pinhole width of space between coils to implement a diode position sensor, the LED and photoresistor/comparator circuit(s) are off to opposite sides of barrel(outside of), just need to keep ambient light from photoresistor
you can have a sensor under the coil.. equation, laser data point interpolation. for a projectile that is the same size and will always be that specific size the blind method would do fine if designed around that specific projectile.
Keep up the good work. Love seeing the problem solving process that you’re going through
So reality in a way helps to simplify this problem somewhat since you don't have to model the relationship between the number of turns to the magnetic field given the current and voltage limits of your power supply and the available volume for the coil. Magnet wire comes in a finite number of gauges so instead you can figure out how much magenetic field you get from using the maximum number of turns for each gauge of wire available to you and just look for the biggest number.
I think heatpipes or vapor chambers and some copper shims might be a good way to deal with heat dissipation. Look for what's available and tailor your design to that.
I literally starting drawing and planning for a laser "range finder" (idk how to say i dont really speak english) for my coil gun just 2 videos after this one hahahaha, so this makes me confident that my idea doesnt suck
I propose a hybrid idea of nested geometry with the original exponential geometry except with the inefficiency nested or truncated and nested inside the next coil and so on all the way down combined that with high coil density as mentioned in this video and you would have something really interesting- in terms of cooling I would suggest fitting the entire coil geometry section inside a High pressure steel tube with liquid nitrogen coolant.
And since you’ve got that for you might as well make the entire thing superconducting…
Just retro fit your coils... say you build 8 coils, you could put a sensor at the end of 1 coil, measure the speed/time, use that speed/time (avg over x amount of firings), remove sensor, add second coil and repeat until desired length/speed is reached.
Arrange the coil distance like guitar frets . Ascending the distance, means ascending the velocity .. use small coils in the rear side, and then use bigger coils in the frontside.. the front bigger coils will have more magnetic power because it is wide like guitar frets. Well, this is just an idea.
IDK if you will ever read this, but anyway.
Disclamer: i am not english, sorry for the grammar.
Your problem with the resistence, you can more or less ignore that, since you do not want to go for big coils anyway.
Copperwire has so low resistance that you need to have a very big coil for it to kick in.
The bigger issue with big coils is the inductivity. Basecly the more coils, the longer the magnetic field will linger around and pull back the projecktile.
Sadly i havent jet found a mathematical solution for this, but i am working on it to at least beeing able to guess more accuratly.
Thank you for making such a great video!:)
Use effect of the induced current in the projectile passing through the coils to turn on and off the next coil.
Iar proiectilul trebuie să nu fie mai lung decât distanța dintre bobine. Cea din față să tragă cea din urmă să împingă. Depinde în cât timp proiectilul î_și păstrează magnețizarea și ce polaritate are. Eu aș face și un proiectil bobină tip capcană magnetică, cu intensitatea câmpului magnetic variabil pe lungime. Forță la start trebuie să fie minimă iar la finiș maximă. La fiecare bobină mai adaugi un conductor așa că intensitatea va avea libertate mai mare. Distanță dintre bobine din ce în ce mai mică la ieșire pentru a imprima accelerație.
This seems like a brushless motor / ESC problem. A sensorless motor provides the feedback to inform the progressive pulse. The question about wire diameter and number of turns and resistance is a question of ampacity. The wire size you choose determines the current you can push V=I.R. If you sequence the firing of coils as you describe but do them in series you put more voltage across the coils.
it depends on the wire you use, I've made a cylinder biphlar coil with only 5 turns and it's pretty fast and pretty strong.
I'd say somewhere between 5 and 15 turns depending on the wire size and wire type.
The moment i´am realized you uploaded the video on my Birthday....
There is an other problem to spent attention to. The coils have inductivity. They switch on and off slow. When the current is high it is hard to switch them off without protecting the mosfet. I don´t know if this is long enough to effect the timing. If the power comes from a capacitor you don´t need to switch off if the cap is already empty but there could be a resonat effect.
Oops, I see this is already in a later video.
The longer the coil the more uniform the field so you actually have peak field quite a bit of distance away from the center.
You could put the ir leds/diodes flush with the coil within the same slice. Though I wonder if the added distance between the radius of the coil and the projectile would effect the efficiency. I'd also worry about said components being damaged.
Also consider the inductance and hysteresis of the bullet itself. That way you make sure that Maxwell's Third and Forth Equation is working for you than against you.
Pre-programed firing sequence and it automatically strobes power through the coil connections from the start to end ...got the idea from seeing the landing strobes at an airport ....
Excellent video!
Good logic, grats on the conclusion.
Just a couple of thoughts.
In terms of coil control, it is similar to microstepping stepper motors.
About detecting the position of the projectile, the projectile will change the inductance of the coil. There could be a voltage change on the coil telling you where the projectile is. Thanks for your video.
On the topic of knowing where the projectile is, you could easily use VERY small diameter fiber optic material with a bright LED on the PCB, and another strand opposite the projectile opening on the other side leading to a photoreceptor. This will let you time things, and I feel that the blind timing is a dead loser. Can't address the coil geometry, but the PCB is more than thick enough to accommodate grooves that will hold the fiber optic strands.
Coil
B = 4 π µ₀ (N I) / ι
B is magnetic flux density in Teslas
µ₀ is magnetic permeability of a vacuum _(air)_ 1.25663706212 × 10−6
N is number of turns of the coil
I is current _‡_
ι is length of coil in meters _(i.e. 1 inch = 0.0254m)_
_‡ current is use here, usually identified as amps, but magnetic field is generated by moving electrons which suggests watts may offer more flexibility in comparing coil configuration.._
So on your mosfet (or whatever you use), if you put your modules 1 up, 1 down, at 180 degree difference you could have aluminum fins to absorb some of the heat, where each circuit board slides in between a fin. l.l.l.l.l.l scenario, may be one way to keep the size down and absorb that heat.
It doesn`t work because of the butterfly effect. If you place the projectile just a fraction of a mm different in front of the first coil the timing will not be apropiate.
Carl Bugeja has been working on PCB coils and motors, may be worth looking into his videos!
Tinkering with coilgun ideas for a few years.
In my oppinion the initial stage is most of the time unefficient. The projectile has te be pulled to to coil from a big distance. Exactly what you are telling. Don't think there is much we can do about it.
A nice way is to use a solenoid to push a projectile from a magazine into the barrel and place it at the optimal position for the first stage.
Usually a multi stage uses a ir gate to trigger the next stage. But the ir gate is in front of the coil. So when the capacitor discharges it has to pull from a long distance and have a low efficiency. I'm thinking of two ideas.
When winding the coils embed a glass fiber rod though the coil at the optimal trigger location. And hook up the IR gate to the fiber rods. This way you can measure and trigger from inside the coil.
Second idea with the ir gate in front of the coil is a trigger delay that can be set with a potentiometer. Ns-ms range.
Most of the time the velocity after the fist stage is stable at a given speed.
When the projectile passes the ir gate the trigger is delayed to the moment the projectile reaches the optimal discharge location.
By Modifying the delay you can tweak the maximum exit velocity. Each stage should have less delay since the projectile speeds up.
How about using a conductive projectile where the base (or end) of the projectile is used to "make" the connection which gives power to the coil. The instant it accelerates and reaches the second coil (by kinetic energy of the first coil), it completes the power supply to the second coil which gives it thrust to move to the third coil and so on.
Only issue is the froction heat which could destroy the rails eventually (after firing a few rounds)
From what i understand that would essentially make it into a rail gun, defeating the purpose
What if you offset the rotation of the coil pcbs, to free up space to allow for smaller coils if necessary or larger MOSFETs, the only problem would be is that the power rails would have to wrap around the the barrel. Also, you should get a discord server, it would help with communication.
You make a good point, that's also something I thought of. if you offset every other transistor position by 180 degrees, you could just have four busbars instead of two, so power distribution wouldn't be much of an issue.
Hey,
I guess i may help,
U asked u want more turns, without increasing resistance,
Why don't u go for thicker wires,
In that way , you get a longer wire to increase the number of turns, without increasing the resistance,
Like if u increase the diameter by 2 times,
Then u can get 2 times tge length of previous wire at the same resistance,
Also , ur 2nd question, u want to kniw how many turns to go for a single coil, well
First u should know how much each coil should provide force,
Based on that u can easily calculate the no. Of turns u need, using simple formula.
Everyone is talking about optical sensors to sense where the projectile is located. Why not use small single-wire coils between every large coils to sense with?
You could make them ignore the magnetic fields from the discharge coils and only sense the projectile, right?
Optic fibre between the coils keeps the detectors outside the coils and can be made as a remote detector and control module. You can wind the optic fibres into the actual coils. On-time is important, but also, inductors take time for the field to collapse and will slow the projectile if the coil is not collapsed in time. Bundle one end of the optic fibres to your IR source and detectors for each coil optic fibre. One or more feed fibres and a detector fibre for each feed fibre. You need to understand inductance more in your coil designs.
This is a really cool design. I love how you went from 'block' modules to the 'exponential' coil because of inefficiencies and then back again, also because of inefficiencies. Just now you end up having virtual 'block' modules composed of many slices. My first though is "how much force can a single slice produce"? The more coils you have the more force you get, so having loads of smaller, thinner coils means that each one individually produces less force, no?
Another thing. Regarding the blind system you intend to implement. That's a really cool concept. Could you maybe use capacitors, which are charging up, to trigger the mofsets in the optimum sequence? By varying the RC value of each individual capacitor in your barrel assembly you should be able to precisely tune how long it is before a cap reaches the required voltage on you mosfet. Thus you can get a really acurate timing, independent of some microprocessor. Anyhow, i think your timing circuitry must be without a microprocessor. The timing are too tight imo (especially with thin modules which require more precision). Im really looking forward to the next episode!
Potentially. An individual coil would likely be able to produce less force, but who says you have to turn on just one coil? You could turn on multiple adjacent coils at a time and it would effectively be the same.
@@LeviJanssen thats the virtual block that i was talking about. At that point you have essentially reinvented the classic 'block' setup, except that you can move the 'virtual block' down the barrel alongside the projectile, which is more efficient as you said.
What do you thing of using charging capacitors for timing? I don't really do electronic engineering so i wonder how plausible it would be, but that's what popped into my head immediately.
Using RC circuits as a delay is certainly a thing, but capacitors tend to vary in capacitance by plus or minus ten percent and resistors can also be somewhat imprecise. Not to mention that every stage would require very specific values and none would be the same and it would be impossible to quickly adjust values for tuning. It wouldn't work. I also had some concerns with using a microcontroller for timing, but assuming a conservative on/off speed of 300 kHz, at a very optimistic 1,000 ft/s the projectile would travel 1mm per "tick". And that could only be the case at the end of the barrel. Everywhere else you would have much finer control.
@@LeviJanssen fair enough. Can't say that im not surprised that my solution wasn't optimal. I would have said that the caps would be the best way but i didnt have a clue about the error on the cap value.
You seem to have definitely put a lot of thought into this and im looking forward to future installments. Keep up the awesome project my dude.
@@LeviJanssen I know you were working on this a year ago, but I believe 1000 ft/sec is going to be unobtainable for a small coil gun, due to the fact that a smaller projectile is going to reach magnetic saturation fairly quickly, and once the projectile is saturated, adding more magnetism isn't going to make it move any faster, because the extra magnetism doesn't have anything to hold onto.
Try making the coils like a spool. Use an aluminum sleeve that'll go around the carbon fiber barrel so u can slide the spools on and off
13:52 what about putting some kind of a TOF sensor on each coil module PCB making up the barrel? TDK makes an ultrasonic one, but you could use an optical TOF sensor, too. You might also be able to use a hall effect sensor.... they're cheap enough to include on each board and would offer a more dynamic solution
it's always fun watching people rediscover UK Professor Eric Laithwaite's inventions (inventor of the linear motor, 300+ mph, aka the bullet train). see all his video's including 'Professor Eric Laithwaite: The Circle of Magnetism - 1968'
So you're thinking about it all wrong with the turns of copper wire. Think about it more in terms of resistance per foot and length of copper wire. The gauge of copper wire you choose will determine the length you use because like you said there comes a point where amperage decreases due to Resistance. So what you want to do is choose your gauge, then determine how long that wire can be. You want the biggest coil possible so if the resistance gets too high you will compensate with voltage. I believe that if you just choose a starting point it will make it easier and thickness of wire is in my opinion the best place to start because from there you can calculate everything. I like where you're going with this idea and I believe I have a coil design that will work better. You're design is good except that the outer diameter of the coil is too far from the projectile so it won't be adding much to the equation except for losses. This is getting too long haha. I've been putting a lot of thought into the same thing if you couldn't tell.
There is no ideal number of Turns in a coil. It all depends on the voltage you use. If your voltage is high, Your current can get lower because you can use a coil with more resistance and thus more turns. I observed this in different coils and voltages from 20v up to 1kv
I have Played around with Different coils and Different voltages. The thing i learned is that you can pretty much calculate the number of turns in a coil Based on your voltage and the resistance of your wire to get the Maximum current at the given voltage. You have to conclude every component in the circiut because the total resistance is important.
I would suggest that you measure a given length (1m) of your wire. Based on this you can calculate the Maximum current at your operating voltage per 1m of wire. Then you just multiply the length and calculate again until the Maximum operating current is within the Limit of your Power source.
At the Ende you will have the exact length of wire you have to use. The ammount of turns hardly make any difference from there on given that you work with a "long" coil wich has more then 3 layers.
Also the results with long coils where the diameter is smaller than the length have been netter in my Experiments.
Contact me if you need more Information about coils.
Would it be possible to use the coils to sense where the projectile is?
By applying a lot of constant current you also heat up the coil. So yes, of course it'll be less power efficient, but I think it would be more efficient for the copper. Power is quite cheap, though expendable here. copper wire especially for bigger coilguns gets a bit costly and heavy.
Blind systems need to be fine tuned, and I would think it's main enemies could be gravity or centrifugal force from movement of the gun, and change in friction of the projectile for various reasons, but these aren't significant for the most part. Although at the end when the acceleration per coil falls dramatically, to use them at all I believe you'd still need sensors even with those being a small factor, but at some point they'll be enough of a factor to make a coil either turn on before or after the projectile goes through it.
I saw someone helped with the turns/wire gauge debate you're having so I won't go into that.
Now to the last video.
On large AC VFDs the gates are liquid cooled because its much more efficient than air cooling.
Could u use the coils to "sense" te projectile inside by measuring the inductance?
For cooling you maybe can suspend the whole unit into an inert atmosphere and cool like by evaporation like a fridge does basically.
Mans tryna blend in
7/17/2023 4:00 PM
1. 1 BIFILER COIL
2. 9 CAPACITORS at 1F each
3. G7 Aluminum bullet
4. Rifled RESIN barrel
5. 30 3.7V Lithium batteries
6. A arduino and code to control it all. And that only took two hours to come up with!
Pull and push? as the projectile passes mid point the coil reverses to push? use opto couplers between each coil for determining position of projectile, this could control triggeringand coild polarity flip, then use a geometric progression of distance between coils, close at start and increasing in distance, also ramp up power based on distance between coils? Arduino for control?
Also do a hybrid of controler based and trigger based accelerator. I mean you have something like an LDR strapped to a mosfet and the mosfet triggers directly the coil but it can only trigger if the controller gives the ok. So the controller is like a director that decides which coil may fire and which on doesn't. Also the coils give information to the controller such that the coil decides on the fly which coil to activate next
I was thinking of an induction based detection system by adding a magnet to the projectile and make the electromagnets behind the projectile repel the magnet so it would be more efficient.
I think any induction based detector would be overwhelmed by the strength of the adjacent electromagnets. I agree that adding permanent magnets to the projectile would make high speeds easier to achieve, but one of the beauties of a coilgun is the dirt cheap, simplistic projectiles. Magnets are pretty expensive.
I don't have the technical knowledge regarding the electrics but maybe I can contribute a mechanical idea on triggering your coils to switch on and off.
Is it possible to have the projectile travel along two segmented conductors (one positive and one negative) where each segment is a contact switch.
The projectile will close the contact for the coil it is approaching and open the the contact of the coil it has reached the center of.
As the projectile gathers speed the distance between the coils and switches may need to increase to make the best use of the timing of the switching sequence.
Is this likely to reduce the reliance on some of the electronics you have in mind?
21:30 ah yes the CURRENT idea
I love the evolution of thought and realization as you gain more information. Science at it's best sir!
i like your mind and thought the electronic switching timing to be automated to adjust for materials weight and magnetic responsiveness selected. Ideally i am looking to have as an insert into a staff, a walking stick.with a laser pointer and multi-tool.
Yes a coil gun without sensors works best. The timing doesn't even need to be that precise because the maximum force generated by the coil stays quite constant for a short distance within the coil. I measured the forces applied to the projectile depending on its position while pushing constantly 10A through the coil.
However, I'm actually having constant speed with every shot without using sensors. Only the starting position needs to be kept precisely for every shot.
I think you shouldn't take such a short coil. Aim for a square winding ratio instead like 16 x 16 or 20 x 20 or whatever. The reason is the (simplified) equasion H = i * n / l where l gets minimum when using a square shape. Using a coil that is shorter than its height has two negative effects:
- H becomes lower
- Acceleration distance is shorter
Don't use IGBTs please. I was using a 100 € one for my gun which could handle 200 A constant and I thought it would be no problem for it handling 2000 A for 2 or 3 ms like Thyristors can with ease. Hell I was wrong! But you could still use IGBTs or Mosfets for low voltage/current per coil but you will probably end up with 50 coils to reach 140J kinetic Energy which I can achieve with Thyristors and 7 coils.
My most recent Idea was using multilayer coils (each layer has its own independent capacitors connected). This way you could release the stored energy per stage a lot faster which gets more important the faster the projectile becomes.
I feel like the real problem with a blind system would be air resistance varies at different altitudes and temperatures but i dont know how large that effect would be
If the bolt moving through a coil creates a small amount of electricity, it could power a transistor to then go to the positioning circuitry. And then immediately power that coil.
Never mind I don't think this will work
Crazy idea, but maybe you could use a large heat sink with large gaps between the fins and place the mosfets on the fins of the heatsink
have you thought about using tiny wires between the coils and use the projectile to make electrical contact similar to a normal switch for locating the projectile, but I don't know how long that would last considering the forces involved.
For the coils you could try potting them in resin.
For cooling the mosfets there are already good suggestions, but for cooling the coils, in the past, I made a small high power BLDC motor using small copper tube instead of copper wire to Watercool the coils from the inside, that allowed me to push the motor from 500 Watts to 3300 Watts.
500 to 3300 watts? I'd love to see that motor, it sounds remarkable.
@@LeviJanssen It was a Skateboard/Longboard hub motor, I don't have exact data on it, but the motor itself was most likely capable of more than 500 watts, but I had to limit it in the controller to 500 watts because after 10 minutes on more it started melting the plastic tire.
What if, first of all, you make a projectile about the size of three coils and put the coil contacts inside the barrel? in this way, when the projectile passes, it will activate each one of the coils, but putting the contacts of the first coil before in the barrel, of the second coil in the center of the first, the contacts of the third coil in the center of the second coil. and so on? In this way, when entering the cannon, the projectile activates the first, being in the center of the first activates the second and being in the center of the second turns off the first and activates the third.
I do not have much knowledge about it but it is something that occurred to me, sorry if it is misspelled, I used a Google translator
If you do that, you increase the chance of the bullet welding to the contacts. Plus even when it doesn’t weld it to the contacts, it greatly increases the resistance the projectile experiences.
Super brilliant Congratulations
Nice! Would it be possible to collect some kind of data from the coil to detect when you power it? Some kind of hall sensor? Each coil should "produce" power and each coil should have an optimal point when to dump its stored power into the coil. Each coil would have it's own capacitors with the correct amount of coulombs/joules... Fairly "static" with regards to barrel friction and projectile weight, but that seems hard to get around without rather sophisticated real-time (high throughput) measurements...
this seems awesome... but regarding your longer barrel comment... what if I want a longer barrel for cosmetic purposes? and not a stubby.... thing?
Random idea, if used with a vacuum tube in the middle, and an electron source, could this make a Linear Accelerator?
Yep, that's how you could build one of those. There was a Master's thesis (I think from 2010), which did exactly that.
@@hjups actually, having designed LINACs for Ion Implanters decades ago, I can tell you that they use RF cavities (resonators) rather than coils to accelerate the electron source. one multi-MeV prototype managed to burn a hole through a 6" steel platen in the blink of an eye when the brakes were hit to stop the spinning platen and the signal did not get to the LINAC to shut it down quickly enough. now there's a death ray. but the thing was the size of a walk-in freezer, so not exactly a hand gun.
Is it possible to rearrange the coils into a halbach array to redirect the magnetic fields like a lense to drastically increase the flux density and hence the strength of the coil.
I'm not sure but can the longer design even be called a coil gun or is it more of a rail gun. The Navy has a couple " rail guns " and they do produce great power and range or distance. Again I think if you could move the power from a large pack pack with a belt mountain controller for 6 or 8 coils in a spaced out configuration of and of a ever increasing power size using a plastic barrel you would reach your desired end result. This would both be a true coil gun as well as matching your wishes of high efficiantsy. Sorry for my lack of Spelling skills as Dyslexia is a real bother.
Which one is more efficient coil gun or rail gun? I mean with same power which would perform better?
I don't know about the possible bell curve of the force, but assuming that it is constant for any current and number of turns as long as their product remains the same, then twice the turns means half the current to achieve the same force. Therefore, since power losses of a resistor is R*Current^2, it also means that doubling the number of turns halves resistive losses on the coils
have you thought about solving the overheating problem by implementing liquid nitrogen?
awesome!
Isn't the magnetic field inside the coil mostly unifor? So when the projectile reaches the magnet it gets accelerated by an almost constant force towards the (offset) center. You also need distance over wich the acceleration occurs wich the cone shaped coil provides.
I came up with the same idea shortly after watching your previous video on the subject. It would indeed be very effective to vary the voltage/current input to a coil to create a 'virtual' center of electromagnetic force, in a similar fashion to microstepping with stepper motors. I am by no means an expert, but I think that it could be worth looking into using high power DC motor driver chips/circuits to allow you to use PWM to vary the strength of the coils to create a 'virtual' center of electromagnetic force, in a similar fashion to how a linear motor works.
As far as holding the coils together, if you watch the Hacksmith's coilgun video, they wound the coils separately using a jig, and used a binder (glue) to hold the wound coil together. This could work very well, as it eliminates the need for a PCB on the back of the coil.
I'm no expert on numbers of coils in relation to power/force, but I know that DC motors with more turns generally spin slower at a given voltage, but with more torque. I don't know if that applies to this, but I think that maybe you should do some testing with some sort of force sensor, in order to trial-and-error a good number of turns. If you created a test model with only one coil, you could then use a pair of light gates, or an off-the-shelf speed sensor (whatever those are called) to see how the number of turns affects the projectile speed.
Coming back to the dc motor bit, I just remembered that a greater number of turns creates a higher back EMF. I am not sure if this applies to a coilgun design, however, it is good to be able to asses all of the factors that affect something like this. ElectroBOOM has a good video about DC motors, in which he talks about back EMF.
Keep up the good work, it is really good to see that someone is actually putting a decent amount of R&D into a coilgun.
Ya, you're right, you could totally use PWM to control the virtual electromagnetic center with an extremely fine precision. Assuming your square wave frequency in high enough that you'd get in more than a few pulses per stage, PWM could make a real difference. I am pretty set in the idea that keeping the electromagnetic center as close to the projectile center is the most important factor, and I'm surprised I didn't think to use PWM for fine tuning.
@@LeviJanssen I suppose that if you added a capacitor, you could smooth out the square wave, which would keep the virtual electromagnetic center from oscillating as much, similar to how a lot of DC motors have capacitors across their leads (although I don't know if that is for the same reason). I think the hardest part of using a blind system would be getting it precise enough to map the projectile's exact position at high speeds, and circuit latency might have to be accounted for.
@@quinnobi42 the capacitors across a DC motor are for noise suppression from the brushes in the motor affecting the rest of electronics or generating radio interference
@@andrewharpin6749 Interesting, I guess you learn something every day!
The POWER MOSFET wont be on long enough to overheat and blow out like a 22 caliber gun. If you are worried just use the high speed fans from PCs that cool the chip.
i think we should focus sound waves in to a tube forcing the ball out
1. You can't design the coil in a vacuum. The correct number of turns depends entirely on what voltage/current whatever power supply you are going to drive it with will use. If you have it hooked up to a capacitor where you can drive 10kA through the coil, the answer will be totally different than if you have a 60V lithium polymer power source with a much higher ISR. Also, you have to work with voltage/currents that you can actually switch feasibly.
2. Yes, there's not reason such a device won't work perfectly fine open loop, but you might be able to use hall effect sensors to get the position of the projectile. It will have some of it's own magnetic field/distort the magnetic field, and as it moves down the barrel, it will move past the sensor and create a signature different than just what happens when the coil turns on (though not sure if it would be discernible, signal to noise ratio might be really bad).
As far as detection goes, an optical rangefinder behind the breach of the gun seems most sensible to me: no need to space out the coils further and the information for timing all coils is obtained by a single sensor.
I agree with thin coils. I'm pretty sure the resistance, magnetic field etc gains of the number of turns in the coil all cancel out so long as the volume of copper in the coil remains the same. That means you should go by the current your power supply operates most efficiently at.
To cool the MOSFETS, i would just thermal cement them to the coils they control and use them as the heat sink.
I'm also wondering if you could direct the flux from the coils into the projectile better by installing ferromagnetic discs between each coil and a pipe containing the coil/barrel assembly in the centre.
To make it efficient, you'll also want to recharge your capacitors with energy stored in the coils when they're disabled.
Oh yeah and submerge the barrel assembly in liquid nitrogen =)
Use laminate metal sheets in between the windings. This may work the same as it does in solenoids and transformers. It concentrates the magnetic field and the laminate steel reduces loss from eddy current. Just a thought but best of luck
Would washers work?
This is genius.
layer upon layer held in place by a resin in a mould, fine layer of ceramic or glass to insulate ?
Have you worked out the most efficient windings ?
Why drive something like this with a transistor. Use the transistor to flip a relay that is of spec instead of worrying about rds and heat sinking. For your heat problem though, they make nonconductive oil you can submerge your fet in
You've got a 3d printer just print a bushing that presses through the coil, you could probably just slip it into the next bushing and ditch the carbon fiber all together. When its built, seal it in a tube and oil cool it.
His name is definitely getting on the list❤❤❤ but he's going to be expensive very expensive to hire
Hey Jannsen what if you take the infrared sensor and remove it from the direct area of the coils and use the space the diameter of the laser light takes up to detect when it passes? You could also potentially do something with a coloured part of the projectile to detect a specific point to time it better, 2 parameters is better than 1 i would assume