FulanoDetailFriday, 16/08/2024, 16:19
I feel like my passion for this project is just fading away. I liked spending time researching and theorizing and all that stuff, but I feel like I never make any progress.
I mean… I don’t know. Pretty funny how I’m feeling down so close. And even funnier how everything seems so close and so distant at same time.
I talk, I talk and talk, but I really don’t know what to do…
I feel like I don’t know what to do, I feel totally lost
I’m really undecided on what to do next? Or am I just trying to procrastinate?
You know what? I just promised to myself I will only post this project log when I actually finish 3D modeling something.
It is funny to me that every day I’m either “this project is stupid, I’m a joke” or “f*ck yeah, this project is awesome”
Also, have you guys seen the teaser for “Secret Level”? There will be a f*cking animation from Armored Core!
It seems that this project log is so long it won’t fit in a single post, sorry for that.
The one time I actually do something in this project and the posts are completely overwhelmed.
Also, for some reason I decided to make a wargame tabletop with a friend of mine. I’m definitely not good with my life choices… Because it will also take years to just study the already existing wargames systems and rules.
About Electromechanical Rheostat Servos:
In the previous project log I talked about how the electrical actuator mech with electromechanical rheostat-switches wouldn’t be practical because I would need a lot of brushless motors.
Obviously I was wrong, although I don’t have the precision to make tiny brushless motors on mass, I can actually make something closer.
- I can make linear brushless motors that can rotate crankshafts.
- I can make solenoid that also can rotate crankshafts.
- I can make bigger, less efficient DIY brushless motors.
For some reason the obvious never crosses my mind on the first try.
I will list here some videos I found about making DIY brushless motors, there are many, so I will just put the links:
- https://youtu.be/0j2epmD4MYs
- https://youtu.be/VQu7KKK3qXo
- https://youtu.be/OZarwftUh8w
- https://www.youtube.com/shorts/Vtm_HCDdENM?feature=share
- https://youtu.be/FxmJE8zcoZg
- https://youtu.be/HaURynfTqqo
- https://youtu.be/bE5CMq7bSZk
- https://youtu.be/TFQrXbMy84c
- https://youtu.be/1K8LQPfMmJk
- https://youtu.be/w25OAqQk_9g
Anyway, I need to start 3D modeling this piece of crap that I call mech, and then I will need to actually build something.
And my first doubt is: how should I make the direct drive brushless motors for the mech?
I mean, I’m thinking of just going “frick it, we ball” and 3D modeling every option of actuator for the mech. The linear brushed motor, the linear brushless motor and then the rotary brushless motor.
… And maybe the exoskeleton I talked about before.
High Torque Density motor:
Before I finally go with the hydraulic/hydrostatic route, let me at least try to find a super high torque to weight ratio actuator. Maybe this way I can still keep it fully electric.
Being honest, I doubt I will be able to pull it off, but even if I reach a pretty close value to those ultra light high power motors, it is still going to make the electro-hydraulic mech lighter.
Anyway, what number of poles and slots would I need for a direct drive brushless torque motor?
I always use the REB-90 as the reference and I also posted that excel document on which shows the best ratio of slots and poles, but none of them talks about the number of slots/poles based on the power output you want.
The “best” approach I have for now is to “just” increase the AWG of the wires, since changing the wires of a brushless motor for a thicker one also increases its torque while decreasing its output speed.
I already 3D modeled the REB-90 copy ages ago, so maybe I just need to cut the number of poles and slots by half and significantly increase the thickness of the aluminum wires.
You know the drill by now, no?
I will talk about direct drive motors, then I will talk about torque motors and how none of the manufacturers tell how many poles and slots they have, yara yara yara…
Some relevant links for this task:
- Before anything, the previous Project Logs about DIY electric motors are still valid, but I believe that what I talk about here wasn’t discussed on those logs. On those I was more concerned with practical construction techniques instead of the high density to weight ratio part.
- https://electronics.stackexchange.com/questions/590217/quasi-direct-drive-actuator-slot-pole-combinations-for-high-torque-motor They show the spreadsheet I talked about talking about the number of slots and poles. They said that for a torque motor, the more poles/slots, the better. So, I’m assuming that it should be 108 slots and 74 poles based on the spreadsheet. But you could have some combinations even reaching 132 slots. Unfortunately, the Bavaria Winding Scheme calculator doesn’t allow for the number of slots above 99, so I have no idea of which winding factor (ABCabc) you need for this combination. The closest one would be 99 slots and 66 poles.
- https://www.moog.com/products/motors-servomotors/brushless-motors/high-performance-direct-drive-brushless-dc-motors.html https://ph.parker.com/br/pt/product-list/frameless-direct-drive-torque-motors-tk-series https://catalog.wholesalesupply.us/brand-siemens/torque-motor-1000nm170rpmaxial-cable/sku-V4319-1fw61600wb072jc2 https://emrax.com/e-motors/ Lists of available Direct Drive Brushless DC Frameless Torque Motors that these manufacturers produce, their output torque, size and weight. For reference.
- https://electronics.stackexchange.com/questions/562402/how-to-select-bldc-motor-for-high-torque-applications/562459#562459 Torque motors perform the best the wider it is, because the torque is generated between the electromagnet tips and the permanent magnets. And so, the smaller the distance between the edge of these two and the lever/arm that will apply work, the better. But you can’t realistically make it infinitely wider. Although, you could take a smaller motor that has an output of 100 kw or more and make it wider. (by the way, I took the 3D model of the REB-60/90 I’ve made and basically, if it had 2.5cm of thickness, it would be 1 meter wide. The most compact that I could make it without these absurd sizes was giving it 10cm of thickness and 50cm of diameter)
- It could help to make the wires as flat as possible (or have as many parallel wires as possible) to make the coil as small as possible.
- https://youtu.be/PMma3OJUHhs I think it is interesting, it shows that I can’t “just” add a random piece of iron to block magnetic fields, the magnetic field must be either fully absorbed by a really thick piece of iron or redirected to the other magnetic poles in order to close the loop.
- One thing that I noticed: some motors have peak torque of the same value, bigger or lower even though they are of the same or bigger wattage. I would guess that a way of actually getting the power/force you want from the torque besides the number of poles, slots, diameter and wire thickness is to take the input watts of an electromagnet, solenoid and the like and multiply its weight in wire between the electromagnets of a motor. For example, the linear solenoid motor that I was thinking of using has an output force up to 10 tons. What if you took the input, 200,000 watts and divided it by the number of electromagnets (slots) in a brushless motor? Every active slot would output a smaller version of that power, but together would be 200,000 watts. 200,000 divided by 20 active slots = 10,000 watts. How do you want to divide that amount of watts? 500 amps x 20 volts? 500 volts x 20 amps? So on and so forth.
With those parameters established, the first thing that comes to mind when talking about high torque density are axial flux brushless motors, but I don’t like one aspect of them. The bigger the electromagnets, the more distant they are from the edge of the motor, which means that not all electromagnetic fields are being used to its full extent. So I was thinking of making a wobbly axial flux style axial flux motor.
In fact, you could actually make multiple stators and rotors with the axial flux brushless motors in order to increase their power output without increasing the weight too much, no?
Found this article: https://ntnuopen.ntnu.no/ntnu-xmlui/bitstream/handle/11250/258043/759964_FULLTEXT01.pdf?sequence=2&isAllowed=y
“By stacking more discs in series, higher torque can be obtained. Doing this will not increase the efficiency of the machine, because the arrangement is made up of several machines with the same efficiency. The losses will increase proportionally with the torque increase. The end coil volume for each stator disc will also remain the same.”
I've seen an axial flux motor called “D500 1x3”, which has around 350kw of output power while only weighing 30kg.
I couldn't find any other motor with this absurd output, what is their secret?
I see that a lot of these high power low weight brushless motors normally need water cooling… I would guess that they are sacrificing efficiency by using lighter windings and using the cooler to keep it from melting itself…
Being honest, with all this headache over motor weight and power output, I’m starting to not worry about 90%+ efficiency.
It seems I’m mistaken, they actually have high efficiency and the reason they need liquid cooling is because 10% to 5% of energy converted into heat starts being a problem when you have 200,000 watts, which would be around 2000 to 1000 watts of energy converted into heat. That is enough to melt metals by the way.
About how they are so light, I received an answer:
“Brushless motors may be powered by DC but they turn that DC into high frequency (kHz range) PWM (Pulse Width Modulation) energy that drives the coils in the motor, effectively creating a multi-phase rotating field. This allows the field coils to have lower inductance than a conventional 50 or 60Hz motor. We're talking 1 to 10% of the copper required.
Because of the small motor size, the gaps between the coils and the armature can be very small, enhancing efficiency.
There are many more reasons but these are the big ones: high frequency PWM drive and efficiency.”
I asked chatgpt:
“PWM, or Pulse Width Modulation, is a technique used to control the speed and power of electric motors. Here's a more detailed explanation:
PWM refers to both the input signal and the method used by motor controllers to control motor speed. The basic principle is that the controller varies the perceived input voltage to the motor by rapidly turning the power on and off in a series of pulses. [1]
The width (duration) of these pulses is modulated to control the average power delivered to the motor. A wider pulse means more power, while a narrower pulse means less power. [1][2]
This works because motors have inductance, which allows them to "smooth out" the pulsed power and respond to the average voltage rather than the instantaneous voltage. As long as the pulse frequency is high enough, the motor will run at a speed proportional to the average voltage. [2]“
Well, I don’t see that being easy to implement in a fricking 300 kw motor using mechanical switches like I was planning.
The best way I could think of doing such an absurd thing right now would be to: spread the transmitting wire as thin as possible (like a foil), divide it in two, submerge it into a dielectric solution and then force the connection of both thin plates using electromagnets.
That would work exactly like a relay, but as small as possible.
A 100 amps 80 volts (8 kilowatt) transistor costs around 15 reais (3 dollars), but in order to deal with 300,000 watts I would need 38 of these, so 570 reais (around 100 dollars) in total.
But the idea of this project is to be as “DIYable” as possible and cheap as possible, if I simply go the route of buying high power components, I will easily surpass the 5000 reais (1000 dollars) limit.
One idea that I had would be to use a cheap low power high resolution pulse modulator to take the sensor input and then control the high power electromechanical system.
… I should’ve thought of that before…
Also, I found something interesting.
This article shows how to reduce the resistivity of aluminum metal using “simple” tricks, but nothing that will make aluminium super duper mega conductive. It was intended to make it more competitive compared to copper.
Essentially, to make the aluminum less resistive, you need to make it as pure as possible and reduce surface/internal defects as much as possible. You would do that by heat treating the aluminium wire, probably with thermal shock followed by aging (exposing it to around 200ºC to 400ºC for a couple of hours).
A way that ChatGPT suggested to achieve that by recycling soda cans:
“In summary, the overall aluminum alloy composition in soda cans is approximately 97% aluminum, 0.65% manganese, 2.75% magnesium, and up to 0.5% other impurities.
The amount of fluxes and refining agents needed to remove impurities from aluminum soda cans would depend on the specific impurities present and their concentrations. However, I can provide some general guidelines based on typical refining practices.
Fluxes: For removing oxides and other surface impurities, a flux mixture containing 75-85% chloride salts (such as NaCl and KCl) and 15-25% cryolite (Na3AlF6) is commonly used. The amount of flux needed would be around 1-2% of the weight of the aluminum.
Refining Agents:
Sodium (Na): To remove magnesium and iron impurities, the amount of sodium needed would typically be around 0.1-0.2% of the weight of the aluminum. However, this can vary depending on the concentration of impurities present.
Magnesium (Mg): To remove silicon impurities, the amount of magnesium needed would typically be around 0.2-0.5% of the weight of the aluminum. Again, this can vary depending on the concentration of impurities present.
Calcium (Ca): To remove hydrogen gas, the amount of calcium needed would typically be around 0.02-0.05% of the weight of the aluminum.
Flux Reactions: The chloride salts and cryolite in the flux will react with the aluminum oxide and other surface impurities to form a molten slag layer on top of the molten aluminum. This slag layer will float to the top of the melt and can be skimmed off to remove the impurities.
Refining Agent Reactions:
Sodium (Na): When sodium is added to the molten aluminum, it will react with magnesium and iron impurities to form sodium magnesium aluminate (NaAlMg4) and sodium iron aluminate (NaFeAl4), respectively. These compounds will float to the top of the melt and can be skimmed off along with the slag layer.
Magnesium (Mg): When magnesium is added to the molten aluminum, it will react with silicon impurities to form magnesium silicide (Mg2Si). This compound will sink to the bottom of the melt and can be drained away to remove the silicon impurities.
Calcium (Ca): When calcium is added to the molten aluminum, it will react with hydrogen gas to form calcium hydride (CaH2). This compound will float to the top of the melt and can be skimmed off along with the slag layer.”
It would be easier to “just” at ⅓ of recycled soda cans to a high purity aluminum crucible… But high purity aluminum is expensive.
I asked again, but using aluminium alloys to make the slag instead of flux:
“New Mixing Proportions:
Soda Can Alloy (100g): Aluminum: 97g Manganese: 0.65g Magnesium: 2.75g
Addition of Aluminum-Copper Alloy (e.g., 2024):
Add 10g of 2024 alloy (which contains around 0.44g of Cu).
This may reduce the Mn content by forming Cu-Mn intermetallics.
Addition of Aluminum-Silicon Alloy (e.g., 4043):
Add 15g of 4043 alloy (to react with magnesium and some manganese).
Expected Outcome:
Manganese: Partially removed through Cu-Mn compounds and silicide formation.
Magnesium: Reacts with silicon to form Mg2Si, which can be removed.
Silicon: Should be controlled by not adding too much 4043 alloy.”
And the most obvious way of all to reduce resistivity: reduce its temperature.
To make it have the same conductivity/resistivity of copper, you need to reduce its temperature to -65ºC (according to ChatGPT). According to the actual paper, it seems like you would need to reduce the temperature to between 100 to 150 kelvin, or -173ºC to -123ºC. Unless the motor is encased in a liquid nitrogen, I don’t see this being possible. Besides, if I were to use liquid nitrogen at all, it would still be better to use copper.
According to this article (that is actually accessible), ChatGPT was correct. However, it depends on the type of aluminum alloy, the 1100 is the purest aluminum. Which also proves the resistivity of materials based on their purity.
Well, you could “just” add a little bit of copper, reducing the resistivity and in turn also increasing the cooling from -65ºC to something that even a freezer could handle (-17ºC).
Of course, that would slightly increase the density (and weight) of the aluminium.
“To determine the amount of copper that should be added to pure aluminum to achieve the same resistivity as copper at -17°C, we need to consider the following factors:
Electrical resistivity of copper and aluminum at -17°C:
The electrical resistivity of copper at -17°C is approximately 0.0174 μΩ·m [1].
The electrical resistivity of pure aluminum at -17°C is approximately 0.0265 μΩ·m [2].
Calculating the required copper content:
The resistivity of a copper-aluminum alloy can be approximated using the rule of mixtures:
ρ_alloy = x * ρ_Cu + (1-x) * ρ_Al
Where x is the weight fraction of copper in the alloy.
To achieve the same resistivity as copper at -17°C, we need to solve for x:
0.0174 = x * 0.0174 + (1-x) * 0.0265
Solving for x, we get x ≈ 0.34 or 34% copper by weight.
Therefore, to make a copper-aluminum alloy with the same resistivity as pure copper at -17°C, you should add approximately 34% copper by weight to the pure aluminum.
Adding 34% copper by weight to aluminium would increase the density of the aluminium in the following ways:
Copper has a higher density than aluminium. The density of pure copper is around 8.96 g/cm³, while the density of pure aluminium is around 2.70 g/cm³. [1][2]
When copper is added to aluminium, it forms an aluminium-copper alloy. The addition of the higher density copper increases the overall density of the aluminium alloy.
The density of an aluminium-copper alloy increases approximately linearly with the amount of copper added. With 34% copper by weight, the density of the aluminium alloy would be significantly higher than pure aluminium, likely around 3.8-4.0 g/cm³.”
On Quora answers however, some people pointed out that aluminum’s resistivity increases with the addition of copper. Source: https://www.sciencedirect.com/science/article/abs/pii/S0925838817328852
Either way, I will just let these two links here:
List of additives that reduces the freezing point of water: https://www.researchgate.net/figure/Chemical-materials-able-to-lower-the-freezing-temperature-of-water-solutions-if_tbl1_273362118
Accordingly to this patent, potassium acetate only reduces the freezing point of water to -95ºC when it is mixed with potassium formate: https://patents.google.com/patent/US5104562A/en
Hyperspace pirate’s pulsetube cryocooler: https://youtu.be/GjRoThMyNGA https://youtu.be/WOmjJFk8rl0 https://youtu.be/cy8aGMH8Tz4 https://youtu.be/vC2it8LHKSQ
The idea is to use the R410A gas (or Helium, depending on how cheap it is) coolant in the pulse tube cryocooler to reach -70ºC and keep the protected electric parts submerged on the water+additives at -70ºC. So the whole system will be as efficient as possible.
Observation: Although the system doesn’t use flammable products, it is still extremely dangerous to touch a liquid at -70ºC (imagine being splashed by it).
This, of course, is just to make Aluminum as conductive/resistive as copper while being as light a possible.
If -70ºC is too much (and I do believe it is too much), you should increase the size/weight by 1.54 times in order to compensate for the resistivity at ambient temperature.
And yes, this also introduces more complexities into the system, and unfortunately, you would need contact seals to maintain the cryo liquid inside the motor.
But you would need to do something similar anyway, we are talking about 2 to 20 kilowatts of heat generated by the entire system.
You need a robust cooling system.
By the way, HDPE starts cracking at -120ºC. I got worried about the integrity of the whole system.
But it does mess up PVA, so the graphene polymer composites would need to also be mixed with HDPE.
I’m even considering making a DIY magnetizer for DIY permanent magnets, but although there are many tutorials on the internet, I’m so sure what I can use to make the material that will be turned into a magnet…
Well, now I need to figure out how to make the primer material for a permanent magnet.
For this, I even thought of buying neodymium magnets in bulk (from scrap or normal ways) , smash them up, mold into the way I want and then redo the magnetization process.
But… A 600 gram Neodymium magnet costs fricking 500 reais (100 dollars), even when I try looking for smaller pieces in high quantities it is still 100 reais (20 dollars) for every 400 to 500 grams. I think I would need around 10kg of this stuff for the motor…
“Neodymium magnets are not made of pure neodymium metal; instead, they are composed of an alloy known as NdFeB, which stands for Neodymium (Nd), Iron (Fe), and Boron (B). The typical composition of a neodymium magnet is about 65-75% iron, 15-20% neodymium, and 1-2% boron, with small amounts of other elements like dysprosium or praseodymium added to improve specific properties, such as resistance to demagnetization and high-temperature performance.”
But Some time ago I heard about iron nitride permanent magnets and how they would outpace neodymium magnets:
“Based on the search results, here is a detailed response about iron nitride magnets:
Iron nitride (specifically the Fe16N2 phase) is a promising non-rare earth permanent magnet material that could potentially outperform neodymium-based magnets. The key points about iron nitride magnets are:
“Iron nitride magnets do not contain any rare earth elements, making them less susceptible to supply chain constraints and geopolitical issues compared to neodymium magnets [1].
Iron nitride has a theoretical maximum magnetic energy product (a measure of magnetic strength) of 130 MGOe, which is more than twice the maximum for commercial neodymium magnets [1].
Iron nitride has a very high saturation magnetic flux density of 2.9 Tesla, significantly higher than neodymium magnets [1].
Iron nitride has a much lower temperature coefficient of coercivity (0.4 Oe/°C) compared to neodymium magnets (-81.9 Oe/°C), meaning it maintains its magnetic properties better at higher temperatures [1].
However, current iron nitride magnets produced by companies like Niron Magnetics can only reach around 10 MGOe, well below the theoretical maximum. More research and optimization is needed to unlock the full potential of iron nitride [2].
Niron Magnetics aims to commercialize iron nitride magnets with an energy product above 10 MGOe in 2024, and above 30 MGOe in 2025, which could make them competitive with neodymium magnets for applications like electric vehicle motors [2][3].”
That is cool and all, but how do I make one of these?
https://hackaday.com/2022/09/01/iron-nitrides-powerful-magnets-without-the-rare-earth-elements
“Low-temperature nitridation is also possible, using iron oxide nanoparticles as a starting material. In this method, the particles are treated with ammonia gas to get the nitrogen into the crystal structure. Alternatively, iron oxide can be mixed with ammonium nitrate in a planetary ball mill; after a few days of milling at 600 rpm, the stainless steel balls decompose the ammonium nitrate into elemental nitrogen, which diffuses into the iron nanoparticles. The resulting α”-Fe16N2 is then separated by a magnet and can be formed into solid shapes. This method seems like it would easily scale up to an industrial process.
High-temperature nitridation of iron foils and wires is also possible. This method uses ribbons of an iron-copper-boron alloy and exposes it to an atmosphere of ammonia and hydrogen at 550°C for 28 hours, followed by a rapid treatment at 700°C and an ice-water quench. A variation on this method is the strained-wire approach, where high-purity iron is melted in a crucible with urea. The nitrogen that decomposes from the urea diffuses into the iron, and the mixture goes through heat treatment and quenching steps before being hammered flat and cut into strips. The strips are put into a straining device and stretched during an annealing step, which serves to elongate the crystal structure and trap the diffused nitrogen.”
Well, time to look for 32892389328 scientific articles about iron nitridation synthesis just to find a better alternative too late, like every time that I research these subjects.
Aaaaand I was right, if it was that easy, some random science youtuber would’ve done it. I mean, there are literal fricking particle accelerator tutorial videos on youtube:
With that put aside, I don’t think that permanent magnets would make that big of a difference, that 500 grams neodymium magnet I talked about before. It could “only” lift 300kg. While a holding electromagnet’s coil wouldn’t weigh nearly as much (the metal core definitely would weigh more), its holding force would be around 15 tons.
Wouldn’t that mean that if you want absolute performance instead of maximum efficiency, you would want electromagnets instead of permanent magnets? I mean, the magnets would always save some weight because they are already magnetized…
To Be honest with you, I think I will stick with the cheap and weak permanent magnets just because I don’t want to make an induction coil to power up the rotor’s electromagnets.
Ferrite is fairly cheap and I can buy kilograms of the stuff.
Now I need to find out how the hell ferrite permanent magnets are made, because the information I find is always mixed up. For example, the ferrite core that is made by mixing a polymer/resin with ferrite powder is used for transformer cores, but then, there are injection molding ferrite permanent magnets that are made in the same exact manner, but somehow they are permanent magnets. Then there are sintered ferrite transformer cores and sintered ferrite permanent magnets, somehow? The sources keep saying “magnetic field” this, “magnetic field” that, but the word is always used in both cases of a magnetic core and as permanent magnet. For some fricking reason…
I keep asking and explaining while giving links to WebGPT (it can access and read links you send it) and this stupid goddang piece of overhyped vaporwave keeps telling me the same exact thing without explaining the difference between a ferrite transformer core and a ferrite permanent magnet. Bonded by resin/polymer or sintered.
It keeps saying Barium ferrite and Strontium ferrite, but it keeps using these two to talk about both ferrite permanent magnets and ferrite transformer cores.
I can only suppose that the difference is that in the resin/polymer bonded transformer was made by binding ferrite powder without an external magnetic field and the resin/polymer bonded ferrite permanent magnet was done by injection molding under a stronger magnetic field to align the ferrite powder and after it, make it go through a magnetizer.
You can kinda turn a common piece of steel or iron into a weak permanent magnet by rubbing it with a permanent magnet, which aligns its crystalline structure with the magnetic field direction. But I’m not very confident it could survive the varying magnetic field of an electric motor.
(it is actually called the “stroke method”, either way it makes me chuckle, lol)
I just know that ferrite permanent magnet made by bonded powder works because there are companies that sell drone motors that use them:
I saw this video on my youtube feed, but never bat an eye on it, and ironically enough, he explains all of my questions (although the bonded magnet he shows is made out of neodymium).
However, there is a really important piece of information, there are “soft” and “hard” ferrite powders. The “soft” ferrite powder is for transformer cores, the “hard” ferrite powder is for permanent magnets.
Unfortunately, I can’t find ferrite powder of the latter.
The only way I could think of was to do the process I was thinking of doing for the Neodymium magnets: buy kilograms of the stuff, smash them up, mix with a bonding material and magnetize them.
I just found out you can’t use Ferrite magnets at temperatures below -40, but as a core, it loses a lot of magnetic flux at -180ºC.
“The exception is that ferrite magnets experience permanent magnetisation loss at temperatures below -40 degrees Celsius, and magnetic tapes and sheets can lose magnetic force at below -20 degrees Celsius, and should not be exposed to significant cold temperatures.”
Source: https://e-magnetsuk.com/how-does-temperature-affect-magnets/
However, Alnico magnets can survive until -100ºC.
However², even though Alnico magnets aren’t rare-earth magnets, they are even more expensive than Neodymium magnets (in my country).
I researched so much, and in the end I will need to work with neodymium anyway.
Just… Bruh.
Also, almost forgot about this:
https://www.youtube.com/watch?v=NX7GHqq28uU&list=PLzP6_DqZoxc3qNnCqWT7R58rQC3nV30PN&index=1
I obviously won’t build PCB motors specifically because of the resistivity of the copper deposited in their surface not being the best, but they share relevant information about axial motors.
This one about electromagnets.
I’m still in doubt on how to proceed, by the way.
I’m still checking other super high power density motors, and maybe the idea of using as many poles as possible for super high torque is not a good idea…?
The D250 1x3 from evolito is on itself an example, it has 220 kilowatts of power, but it only weighs 8 kilograms because it is a high speed, low torque motor.
While the other high torque motors are 3 times heavier.
… But that pattern doesn’t repeat itself on other motors, like the Freerchobby MP240150 that outputs 100 kilowatts, half of that and weighing double of that (16kg).
I can’t make heads of why, I ask on forums and no one answers.
I even searched for the datasheet for the D250: https://evolito.aero/media/2024/07/D250_210724.pdf
It doesn’t show much tho, it just shows the information already provided and the efficiency graph of the motor.
Calculating it - Dubious result:
Sorry for procrastinating so much, I don’t know why, I always feel a wall on my mind every single step of the way.
Also, these last 2 to 3 weeks I’ve been very sick. First I had a high fever, then I started having a lot of coughing and now one of my ears is so inflamed that I can’t hear anything and my jaw can’t completely close. 😐
Anyway, I will attempt to calculate the weight of the coils even though I didn’t make any 3D model.
I’m also thinking of changing the electric motor from a super high torque direct drive motor to a high energy dense motor, like I talked about the D250 Evolito brushless axial motor.
So, assuming 300 kilowatts of nominal power and maybe a possible 600 kilowatts of nominal power, I would need to divide the 300 kilowatts between 4 stators with 99 slots and 5 rotors with 66 poles. So 396 slots in total with 330 poles. I want to make this with 5 rotors and 4 stators because Axial Flux motors are known for having bad heat dissipation due to its unique structure, so I want to spread the motor’s surface area as much as possible for maximum heat dissipation. Of course, this will come at the cost of efficiency.
Since it is a 3 phase BLDC motor, when one of the three phases is deactivated, other two are reaching peak, so it would be safe to take that ⅓ of the total power and add to the full system, so all the slots are at full power at any moment.
So, 300 kw / 3 = 100 kw + 300 kw = 400kw.
400,000 watts/396 slots = 1010.1010101 watts per pole.
(yes, I’m assuming all the phase windings are divided in parallel)
I’m also assuming that whatever is the actual voltage and amperage that you use, the coils will always have the same weight when taking into consideration the thickness of the wire and the number of turns (which, if it was true, the D250 Evolito wouldn’t be a thing).
What should be the thickness of an aluminum wire with the same conductivity/resistivity of copper in order to transmit 1010.1010101 watts through a given distance?
I asked ChatGPT to calculate the required thickness of the copper wire to transmit 1010.1010101 amps and 1 volt across 10 cm (I’m just guessing this length for the single-turn wire, but it could be less and it could be more), with energy losses of 5%, 3%, 1% and less than 1%.
… And obviously, it was useless because it kept giving completely different values every time. 😐
So, based on copper awg charts based on amperage, the diameter of copper wires aren’t universal. Essentially, each wire thickness can have a maximum amperage and frequency before the wires heats up to 60ºC, 75ºC, 90ºC and so on. So I’d assume that if you double the thickness for a given amperage based on the 60ºC, you would have a wire that barely heats up above ambient temperature.
So that could be the 5%, 3%, 1% and >1% of efficiency loss on wire thickness.
Based on various different types of charts (they always give a different values for some reason), for copper wire with AWG of 1000, having 11.684 mm of thickness and 107 mm² of area, it allows 300 to 455 amps at 60ºC, 545 amps to 75ºC and 615 amps to 90ºC
Like I said, in some charts you have up to 900 amps to heat a 1000 AWG copper wire to 60ºC, so this list isn’t universal. You would need to actually test yourself, but until then, here we go.
So, for each amount of efficiency loss that would turn the energy into heat, the modified aluminum wire would need to be:
- 5% = 107mm² of area for every 615 amps = 14,96 mm of thickness.
- 3% = 107mm² of area for every 545 amps = 15,89 mm of thickness.
- 1% = 107mm² of area for every 450 amps = 17,487 mm of thickness.
- >1% = 107mm² of area for every 300 amps = 21,417 mm of wire thickness.
Using a cylinder volume calculator and adding the length of 10cm for every thickness, adding the volume to a Density to Weight calculator, I would get:
- >1% = 22mm diameter = 38,01 cm³ of volume = 0,10263 kg of weight per slot 0,10263kg x 396 slots = 40.64148 kilograms in total for the 300kw axial brushless motor.
- 1% = 18mm = 25,45 cm³ = 0,06872 kg x 396 = 27.21312 kilograms.
- 3% = 16 mm = 20,11 cm³ = 0,0543 kg x 396 = 21.5028 kilograms.
- 5% = 15 mm = 17,67 cm³ = 0,04771 kg x 396 = 18.89316 kilograms.
I rounded the numbers up because I’m really not that interested in making it in microgram precision, but you can already see that unless you have a really chonky cooler for -70ºC, you would considerably run into more and more weight saving at the cost of efficiency.
Out of curiosity, I took the value of 10% with the density of copper just to see how heavy it would be if I were absolutely certain that I wanted to go with the full electric route: it would weigh 43.53624 kilograms in total per motor. If it was aluminum, it would weigh 13kg.
You could, in fact, reduce the efficiency until it reaches just 10 to 5 kilograms of weight per motor. Reaching a maximum 30% of energy converted into heat for the 5 kilogram heavy motor. However, this is assuming the coils are at -70ºC. And if you keep converting all that energy into heat, it will quickly heat up to ambient temperature and exponentially increase the resistivity until you have a pile of molten metal.
This one is a bit tempting to me, to be honest, having 5kg of weight per motor…
… But you would need a really powerful cooler to run through all these motors…
Now I need to figure out how much it would weigh if the motor in question had an average torque and a high rpm, just like the D250 Evolito’s brushless motor.
Again, the 40kg motor I calculated above is for high torque density, this one is for high power density.
Well, I literally asked WebGPT to search and calculate a high power density:
Based on the search results, here are the key points regarding the ideal number of slots and poles for a high power density motor instead of an axial flux motor:
“For fractional-slot motors with high power density, the selection of both the number of magnetic poles and the slot-pole combinations is crucial. This choice significantly determines the motor's parameters and operating capabilities. [1]
The best slot-pole combination in terms of power density is 24-pole, 27-slots, although this results in a very high operating frequency of 960 Hz. This combination achieved a power density factor of 6.0 kW/kg, which is much better than other combinations. However, the high frequency may make control and powering the motor difficult, especially at high current loads. [1]
The slot-pole combinations with q = 0.5 (where q is the number of slots per pole per phase) are characterized by the lowest rotor losses. The 20-pole, 30-slot combination (q = 0.5) had 2.6-2.8 times lower rotor losses compared to other combinations. [1]
Slot-pole combinations where the number of stator slots Qs is smaller than the number of magnetic poles 2p should be avoided, as they can have large amplitude subharmonics in the magnetomotive force (MMF) distribution, which negatively impacts the motor's performance. [1]
While q = 0.5 combinations have favorable winding factors, the selection of the number of poles must also consider the motor's overall dimensions and operating conditions, as this affects the output parameters and rotor losses. The target solution cannot be predicted solely based on the q value and winding factor. [1]
In summary, for a high power density motor, the ideal number of slots and poles appears to be in the range of 20-24 poles and 24-30 slots, with a slot-pole ratio q around 0.5. However, the final selection should consider the tradeoffs between power density, efficiency, rotor losses, and ease of control at the high operating frequencies involved.
Learn more:
1: https://www.mdpi.com/1996-1073/15/1/215
2: https://things-in-motion.blogspot.com/2019/01/selecting-best-pole-and-slot.html
3: https://www.anttilehikoinen.fi/technology/electrical-engineering/pole-count-good-bad-ugly/”
Well, I will at least find some documents to feed to a new AI bot on the Poe website.
I did it, but I don’t think it helped a lot… As expected for ChatGPT. But since this is the only free option to reach that conclusion, then we ride with it.
It suggested a motor with 10,000 rpm, 300 Nm, 30 slots to 26 poles configuration with 400 volts DC and 750 amps, suggesting 4 turns per pole for some reason?
Just like before, I will only assume the number of turns, its length and size.
Assuming each slot has 4 turns and 750 amps, it would probably be 40 cm of length with the 22mm of thickness, it would weight:
- 40cm of length 22mm of diameter = 0,4105kg per wire.
- 0,4105 kg x 30 slots = 12.315 kilograms in total.
Well, this is an interesting result, but being honest, I don’t quite feel satisfied.
I’m not really sure if this result is correct, and another detail is that the higher the rpm, the lower the torque. So it wouldn’t deliver 300 nm of torque at 10,000 rpm.
The torque, density and efficiency map of the D250 Evolito’s motor shows that. It also says it uses cobalt based core material and permanent magnet materials.
I changed the base AI that this trained bot is supposed to use and it gave me a completely different answer:
“Summary:
In summary, for your 300 kW brushless motor:
Slots: 24
Poles: 20
Voltage: 600 V
Amperage: Approximately 500 A
Summary:
In summary, for your 300 kW brushless motor:
Wire Thickness: Approximately 2 AWG (2.588 mm diameter) to handle 500 A effectively.
Number of Turns per Pole: Approximately 150 to 200 turns depending on the specific design and magnetic characteristics of the motor.
Summary:
The estimated weight of the winding using 2 AWG wire with approximately 175 turns per pole and a total of 20 poles would be around 154.4 kg. This weight includes the wire used in the winding and contributes to the overall weight of the motor design [4][6].
Keep in mind that this is a rough estimate, and actual weight may vary based on additional factors such as insulation, end-windings, and the specific configuration of the motor. Further optimization may be needed during the design and prototyping stages.”
I just copy pasted the summaries or else this section would be needlessly long.
Literally one message after this, it suggested 12 to 24 turns per slot for a 24 slot configuration. 😐
So, it would reduce the weight from 154kg to 24kg if using copper, with a further reduction of weight using pure aluminum cooled to -70ºC. And as such, 7.45kg in total.
Or maybe not.
It incorrectly assumed that the thickness of the AWG 2 copper wire is around 2.588 mm of thickness, when in fact you would need a 1000 AWG wire for the 500 amps, which would be 22mm as I calculated before. Making it around 10 times thicker, increasing the weight by 10 times.
So, 74.5 kilograms for the 300 kilowatt axial brushless motor.
In any case, I feel like this was a dubious result. I don’t know for sure if the result ChatGPT reached is correct, if the result I reached is correct, so on and so on.
The obvious choice would be to go back to the REB-90 brushless motor and slightly modify it in order to make it reach 300 kilowatts of power instead of 80 kilowatts.
But it has the same problem when it comes to the number of turns of wire per stator.
Either way I need to make a 3D model and learn how to make those electromagnetic simulations in order to check the validity of what I calculated here. 😐
… Which I should’ve done ages ago…
… So all these calculations were useless…? Bruh.
3D modeling:
I’m still procrastinating… Still Procrastinating…
I’m still not in the best of health right now, but even so… I feel I should post this Project Log already.
I feel like I really reached a point where I need somebody’s help (I mean, the whole project was like that already).
I will just make a few simple 3D models of the electric motors that I proposed and then I will check how to make the electromagnetic simulations.
I still remember how I tried to make a mechanical simulation of 3D objects on FreeCAD and my potato computer could barely do anything.
The WebGPT says:
“The formula to calculate the number of turns of copper wire on slots from brushless motors is:
Turns per phase = (Voltage * 60) / (4.44 * Flux density * Stator area * Frequency * Kv)
Where:
Voltage is the desired motor voltage
Flux density is the maximum flux density in the stator, typically around 0.78-1.2 Wb/m^2
Stator area is the cross-sectional area of the stator
Frequency is the motor's electrical frequency
Kv is the desired motor Kv (rpm/V)
This formula allows you to determine the optimal number of turns per phase to achieve your desired motor specifications, such as voltage, Kv, and flux density. 1: https://endless-sphere.com/sphere/threads/motor-winding-calculations-cooling-design.122425/
2: https://www.instructables.com/Rewinding-a-Brushless-Motor/
3: https://fishpepper.de/2017/11/10/tutorial-brushless-motor-rewinding-based-on-a-br1103b/”
While the trained bot says:
“To calculate the number of turns per slot in brushless motors (BLDC), you need to consider several factors, including the slot area, the slot fill factor, and the desired torque characteristics of the motor. The formula used to calculate the number of turns per coil (T_C) is derived from the relationship between the slot area (A_slot), the slot fill factor (SF_g), and the current flowing in the coils.
The general formula can be expressed as:
Aslot⋅SFg=2⋅TC⋅ISP
- Aslot is the area of the slot,
- SF g is the slot fill factor (commonly around 0.4 for round copper wire),
- TC is the number of turns per coil,
- ISP is the current flow through the coils.
To break it down further, you first need to calculate the slot area, which is typically provided by design software or can be estimated based on the dimensions of the motor. Next, you multiply the slot area by the fill factor to determine the effective copper area available in the slot. After that, you can rearrange the formula to solve for T C :
Tc = Aslot*SFg/2*ISP
This calculation allows you to estimate how many turns of wire can fit within each slot while optimizing for maximum torque output. The number of turns impacts the back electromotive force (back EMF), which is crucial for the performance of the motor.
For example, if you have:
- Aslot=10mm2
- SFg=0.4
- ISP=10A
You would calculate:
You would calculate:
Calculate the effective copper area:
10 mm^2 ⋅ 0.4 = 4 mm^2
Solve for T C T C :
TC = 4mm²/2*10A = 4/20 = 0.2
Since the number of turns must be an integer, you would round this value based on practical considerations and further calculations on the motor’s torque requirements.
This formula and approach allow engineers to optimize the design of brushless motors effectively, balancing the number of turns with the physical constraints of the motor and the desired performance characteristics.”
Either way, I need to model it…
Well, the REB-90 is already modeled, so I will start with it.
Using an Area calculator, it seems it has around 24 cm² of area.
Well, in one of the calculations it gave me around 7 turns. But this is the calculation for flux density, not number of turns. The calculation should take the number of turns and then the magnetic flux to see if it has the appropriate force.
When I said the number of turns, it gave me around 1.5 to 2.5 teslas of magnetic flux.
… I have absolutely no idea the impact of this value 😐.
This project log got so full of text that I had to split it up to the next project log.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.