I love your attitude towards failing and the way the video shows your actual process and doesn’t fake any progress nor hides any problems. Thank you, I am a sub now!
This was a pretty great video. The editing was good and I was surprised by your 3d animations to explain everything. I've thought about building one of these myself, but now I may hesitate a little more. You're right about learning how the real world works vs theory. You might have made things harder on yourself in some ways, but like you said, you learned a lot from your failures.
Happy to hear that you enjoyed the video! Spent a lot of time on it and also had a lot of fun doing it too! For me, when I decided to start this project, I was hesitant as well, because the only flying thing I ever made was probably a paper airplane. But after some thinking, I realized that if I have some failures, then it's OK! It is my first time after all. Also, the point of this project was, as I said, to learn about how drones worked, so I did many things from scratch. If you want to make a drone that just works, there are off the shelf electronics with software that would make a fully functional drone, with RC capabilities and such. And if the project is meant for deeper learning purposes, then failure is OK! In fact, it's normal. Only issue may be the cost due to iterations, so for that, you can learn from my mistakes which are 1: don't use supercapacitors because you're too curious (use lipos or some other conventional power source), and 2: brushless motors (with an off the shelf ESC controller) would create much better drone response/control compared to brushed motors. If you want to do brushed motors like I did (for cost reasons), then MOSFETs are one of the best and simplest ways to drive brushed motors. You may need RC control for stable flight, cause it won't stay in one spot for long. Hope this helps!
Really Appreciate man I really likes journeys because whenever you start a project you just also excited to do it and you spend time on it and you learn through the mistakes and maybe after a shorter period of time you start to give up and then you think that let's give it a one more try and it starts to give you some results this is so fascinating and even if you shut shut down the project you learn a lot out of it And that's what engineering is Being an engineer I am very happy to see your journey
The most interesting drone I ever built was a tri-copter. I used an old fishing pole as the frame and some cheap brushless motors with 9x7 propellers. And a single metal geat digital servo to steer the tail motor. It took a good bit of tuning to make work well but it was amazing to see it fly around. Too bad so few tricopters exist, they can do some really neat stunts. Although not quite as stable as a quad, they can do things the quad can not do as well. Their flips and rolls are just so different looking. I wish more people would build them. You did a great job on the video and build. Keep it up
You should add a capacitor in parallel to the lipo battery, that should solve the arduino powering off caused by voltage drop of the battery without using a boost converter
Very interesting, I am on a similar project now. Before custom drone I built an FPV with existing components so that I can compare my experiment and I see they all use a capacitor against fluctuations. With supercapacitors they are probably affecting the center of mass and causing the flip, I would try to face them downwards so that they add to stability. Also you can account for those things in your code. For hovering in place probably you will never get there :) as of my current understanding is that you also need a way to measure distance from ground and an optical flow sensor. I am still looking for a good way to have a test setup similar to tripod. Great video
This was fucking great man! I recently decided to take the plunge into small electronics and learning how to do some small projects myself. Some very helpful information in this video on running dc motors! :) Defenetly earned a subscriber here :) Looking forward to seeing more of your content! Glad to see you have some videos posted allready, i was afraid this would be your only one sincly you "only" have 1.7k subs. Allmost at 2k! :D Again, thanks for the helpful information i didnt know i came for, and great job on the project, and GREAT job on the video! :)
The reason the drone was flipping over with the heavy super caps, is that the center-of-mass was not centered within the 4 propellers, so the lighter side was being lifted faster, thus flipping the drone.
I think you should revisit the super capacitor design, but try attaching them horizontally to the spokes/arms of the drone to distribute the weight and maintain a lower center mass, also redesigning the PCB where the IMU chip is located at the center of the design, with the other support components arranged to equalise the weight/center mass of the board
Side-note: Betaflight is very common in FPV communities and all of these kinds of flying problems have been solved. I've been building with it for years, and you can put together a drone like this in a couple hours.
The issue is not the brushed morors, the issue is either your software or the weight of the build. I have made a lot of brushed drones in the past, the recent ones we made for a university event were ran in a swarm configuration so plenty precise and each was only 40ish grams which included a small GPS. Your drone seems too heavy, use a smaller pcb, use a mcu with a dedicated Radio like the esp32 it is also much lighter and faster. Use a smaller frame and try using a barometer or TOF sensor to stabilize the drone mid air.
Some things to do, step away from PID, look into LQR or MPC Karmin filter or the steady state equivalent Inteoduce some system identification, all your motors will be different so ttying to tune the system just wont work out well unless you tune each individual motor and then tune the system Lower your center of gravity, put the controller on the underside
Really great video but is it possible to replace the Arduino iot with codecell , it provides below advantage 1) it have 9 axis imu fusion 2) proximity sensor to do jump on hand trick 3) its really tiny and lightweight 4) ble can be used to control insted of wifi to save energy 5) esp32 c3 seems good soc instead of 2 chip design of Arduino Few mechanical design considerations: I have seen dji inspire flying dead stable because of its hanging Center of mass so if you solder your caps upside down it will help it to stabilize rather then tip over ( what is currently happening) 4 caps at 4 board corners will also work as stands to eliminate ( for lack of proper words ) stickey effect of drones without landing gear Imu fusion and pid works for drone stabilization but nothing can beat the physically waight balanced drone , you should try to center its cg because currently its tipping to its side really bad
I don't think its a hardware limitation I've seen people making functioning ESP32 drone with brushed coreless motors (I've tried this myself with little success). You got some incredible results you should keep going!
It's definitely software. Have a look at open source controllers like Crazy bee, ESPCopter, Ardupilot and see the diagram required to make your own flight controller
Look into implementing a kalman filter for sensor fusion, that might improve results. Additionally, look into the applying lowpass filtering to the accelerometer and highpass filtering to the gyro. All of this might improve the estimation of the attitude.
Good suggestion, however, I actually at some point had a simple low pass filter, but when changing the gains, it didn't reduce noise to the point of improving drone performance. I did look at Kalman filters on youtube, but was a bit intimidated, and heard that complementary filters would essentially get the same job done, with less computational power. My guess is the lack of motor response, as my motors are cheap brushed ones. At some point in the future when I try to make a better drone, I might use a kalman filter and brushless motors.
@@TheTinkeringTechie kalman filter is both heavy on computation side and also intimidating to apply. But it is much better and considering vibrations from motors, kalman filter is w than simple complementary filter. If computation becomes a problem(I think it will be for arduino nano) you can use other simplified ver. which are much better than complementary filter. But is the project still on?
@nikaX2000 Project would probably be continued some time in the future, and it will have new hardware and software solutions (ones that I am reading about in the comments section, like yours!)
Hi, I'm looking to build something similar to this with the MOSFET based motor driver using those same exact DC motors (though they are brushless) but with a pi pico instead, but it looks like you discovered all these issues by actually building it. Also, have you considered using a 5V battery with voltage regulators to the motors to deal with power spikes? Would it be possible to integrate the IMU readings to correct for drift? Also, did you ever try experimenting with how far the motors are from the center? Currently, I've been trying to connect distance sensors to my pico through i2c, but it's taken embarrassingly long because they keep not working. But seeing you work through this process on a similar project is really inspiring, thanks for the video!
Glad the video inspired you! Generally, I won't want to use a voltage regulator if it is to power a motor, because the current that those motors need to draw, multiplied with the voltage drop across the regulator (which is how the voltage is regulated), would most likely go above a typical voltage regulators power rating. Some large capacitors could help with power spikes, filling in the current needs of the motors in a short time period. For the IMU, I tried to get velocity by integrating accelerometer values, but the value itself drifts over time (after a few seconds, a completely still IMU would think it's moving 3 m/s). There may be a way to that I am not aware of though. I could try experimenting with motor distances in a future drone project. Could help.
Thanks, I appreciate it a lot! The self balancing gimbal camera idea does sound cool, and practical too. I'll have to add that to my list of ideas for ranking 😁
your drone is facing the ground effect in the last few clips of it flying. if its higher in the air it will preform better as the ground effect no longer takes place.
Most likely, but it could be heat as well for the 3rd iteration. I did add protection diodes for the 4th PCB design. I didn't mention it, but I did show it when doing SMD soldering (the black rectangular components next to the AOD424 MOSFETs).
You are correct. However, my thought process was that it's good to have the capability and not need it rather than the reverse. Some cases could be that I soldered the wires backwards, but all I need to do is change a thing or two in code and its all fixed. Or, perhaps I could improve my motor response by powering the motor in the reverse direction that it was moving in to stop it faster (braking mode I believe it was called for those H-bridge ICs).
I was thinking the problem with the first design (large super caps ) was it looked top heavy. Might be why it kept flipping over. For the FETS driving your motors you might need protection diodes across the motor s to prevent back emf from blowing them. Also is the FETS you picked good for logic level drive? VGSon voltage should be less than you arduino voltage output high so that the FET turns fully on. This might be why they were getting hot. Cool video though, nice editing!
Glad you enjoyed the video! And yeah, the top heaviness was pretty obvious after a few tries. After frying many MOSFETs on the third design, I did put protection diodes across the motor for the fourth PCB design. Both MOSFET types had max Rdson and min Rdson drive voltage of 2.5V and 4.5V respectively, placing the Arduino's 3.3V digital output a bit below the middle of the two. Not as ideal, but at least when I was searching on Digikey, I couldn't find any MOSFET with a high enough drain current rating with a sub 3.3V Vgs on voltage. The next best thing I did for the 4th design was to choose a MOSFET with a low maximum Rds on of 6.4 mOhms at 2.5V Vgs on, because in the end, it's the resistance that matters. However, the MOSFET still got hot when supercapacitors were used, and this was because of the voltage of the supercapacitors in series, which could get up to about ~13V. After switching to Lipo batteries, with a max voltage of 4.2V the MOSFETs "magically" never get hot! Thanks for pitching in your suggestions, I appreciate the consideration!
Correct me if I am wrong, but isn't 4V*0.15A = 0.6W? We can calculate power dissipation from the MOSFET with P = IV = I^2*R. The highest on resistance of the AOD424 MOSFET is 6.2 mOhms = 0.0062 Ohms. Let's assume a very bad scenario, and the motor that is driven by the MOSFET pulls 10A constantly. P = 10^2(A) * 0.0062 (Ohms) = 0.62 W, a good amount less than the rated 2.5 watts. You may have put the numbers into the calculator wrong 😉
@@TheTinkeringTechie 4 motors at 5V is 4x0.15A or 0.5A x5 = 2.5W. Peak draw at power up will be much higher. Also keep in mind the power drop across the transistor is also dissipated into the mosfet. Mosfets power rating is often listed assuming a heat sink, you need to check the datasheet.
I now understand where the numbers are coming from, but is that assuming that all 4 motors are being driven by 1 MOSFET with the motors in parallel? It seems like you are taking nominal current draw of 1 motor (0.15A), multiplying by 4 (4 motors), and then multiplying by source voltage (~5V, though lipo max voltage is 4.2V) which is about 2.5W power dissipation. However, each motor has its own MOSFET to power it, which is necessary to allow for individual control for each motor for active stability. So instead of 0.15A*4 amps going through a single MOSFET, shouldn't it be only 0.15A? Then it would be 0.15A*5V which would be 0.75W. Looking at the AOD424's datasheet, the 2.5W power rating is for when the device is "mounted on 1in2FR-4 board with 2oz. Copper, in a still air environment with TA=25°C", while the power rating for if a heatsink is used is between 50-100W for case temperatures 100C and 25C respectively. As for a sanity check of my calculations, I touched the MOSFETs after running the motors for a long time, and they were essentially room temperature, maybe warm to the touch. Lastly, when I switched over to using Lipo batteries, all the thermal issues seemed to just disappear, probably because the supercapacitors would be charged up to ~13V, and at that point 0.15A*13V would be getting close to 2.5W power rating, and would definitely go above the power rating of the first MOSFET I used. MOSFETs were definitely hot when using supercaps, as I remember touching them after motors ran, and recoiled from the heat. I am not writing this to invalidate your responses, as it seems like you have some experience, but I just want to make clarifications and also as a sanity check on my experience.
You are wrong that brushed motor can not be used in self hovering drones, I personally have tried brushed motores in a DIY drone that can fly very well. I agree that they are not as responsive as BLDC but there are tons of self hovering drones out there which used brushed dc motor. But great attempt.
your mosfets are dying because the mcu cant drive the gates effectively, the gate of a mosfet is a little capacitor, the mcu gpio cant supply enough current to charge this capacitor in a reasonable amount of time (i mean, to us its almost instant, but its actually very slow), this means your mosfet is spending a lot of time not fully conducting, and acting as more of a resistor than a switch, you need a gate driver ic, itll turn the low current from the gpio into a nearly instant change in mosfet state(10s of ns, as opposed to tens of uS) unfortunately if you're using a mosfet with PWM you cant just take the mosfet iDS rating, you need to calculate the power dissipation of the mosfet, or you can just use a slightly overkill mosfet and worst case scenario slap a heatsink on it, but again a gate driver is a must also the reason your motor randomly turned on when flashing code is pretty normal, the gpio's are unpredictable when flashing, maybe using a swich that turns off the gate driver would be a good solution for that
Good suggestions! For this project, I wanted to keep the components to a minimum to get the general ideas of drone design. When I try again some time in the future, I'll most likely go more in depth with more components. As for the GPIO output current, I did test on a circuit simulator (Circuitlab I believe) with the 0.007A output current and square wave input into a capacitor, with the same capacitance as the gate capacitance, and I saw fairly square output voltage, indicating that the capacitor charges fully and fairly quickly (no saw tooth, no obvious rounding of edges). I recreated it here: www.circuitlab.com/circuit/32abgzvfa52q/aod424-simulation/. Actually, zooming in, I can see the transition time of about 10 uS, so you're right! Gate drivers would be a good bet. If I wanted to keep it simple, would a BJT work well enough for that? As for not being able to use the iDS rating for PWM, is it mainly because the mosfet spends more time in a higher Rds on state because it turns on and off quickly, and thus has to go through the higher Rds on repeatedly, generating more heat? In that case, would it be ok to eyeball by looking at the highest Rds on possible, then calculate power dissipation with the highest estimated current that would go through the mosfet? That's what I did, and the "worst case" was pretty far below the 2.5W (for the 4th design). That conclusion was consistent (when using Lipos), probably because the supercaps would need to be charged at a higher voltage (~13V) as opposed to 4.2V for the Lipos. Thanks!
17:30 is that just me or is the animation reversed? The previous animations only showed the specific PID term discussed instead of a fusion of all of them, yet here the D term appears to cause the rotation instead of counteract it as it should. Minor nitpick, otherwise this is truly a great video, I expected to see way more subscribers when I scrolled down to type this!
Thanks for the compliment! What I was going for in the animation was to show the value that the D term would be putting out as a result of the speed of rotation. The D term is meant to "go in the opposite" direction as the P and I terms, and increases in value as the drone rotates faster. That is why it looks reversed to P and I, as it is meant to prevent overshooting. It's like barreling down the street in your car to try and reach the red light as fast as possible (P and I), but you resist going too fast (D) as you don't want to drive through a red light. Unrealistic example, but that's what I got at the moment.
@@TheTinkeringTechie I do understand that. I've worked on PID controllers for a good while myself. But unless I'm missing something here: Yes, your vectors indeed get longer with respect to rotation velocity. But usually the D term damps velocity, so the force output of the motors would have to counter the rotation. I believe what you have pictured would be a negative D term.
i might have an idea as to why your mosfets failed repeatedly. you may have been reading absolute maximum current this should never be used under normal circumstances, generally speaking this is what you factor in as worst scenario, never reached under normal operation. to add confusion to noobs it is usually at the top of the datasheet.
Yeah, in fact, the last mosfet, AOD424's datasheet ONLY has the absolute maximum ratings, no recommended. Though I tried to find a sort of overkill mosfet, atleast what I think is a bit overkill, for the last iteration.
@@TheTinkeringTechie actually i just checked and it provides the other info outside of absolute max but you have to calculate it yourself to get a rating in amps. i think it's good practice to calculate it yourself anyways because the current ratings are temperature dependant and not everyone has the exact conditions as the tests used in the datasheet for their particular use case.
@@TheTinkeringTechie oh and for overkill solutions i typically use 30n06le with through hole packaging, not sure if they're still in production but i keep a box full of them for whenever i just need something to work and can't be bothered to calculate the load requirements. datasheet says absolute maximum ratings for current is 30a but using a reasonable heatsink (the one reason to use a through hole component in 2025) ive regularly pushed 75+ in 5s bursts and 60 continuous without issue. the gate voltage threshold says 1 minimum and 2 max but through my testing and their own graphs 5 - 8v is what you should actually use.
![Blox Fruits Dragon Rework Update [Full Stream]](http://i.ytimg.com/vi/EqDAp8udhm0/mqdefault.jpg)
You should mention in the title that it's a super capacitor drone!! Incredible
Indeed i discovered it at 0:15 and i got surprised as hell !
Yes preety catchy title compared to current one
Not sure if you finished the video, but it's not. He had to ditch the super capacitors since they were too heavy and unbalanced
The mosfet will fail if you don't add a flyback diode to protect them. This is something you must do since you are using brushed motors.
Much respect for trying the capacitors, just to see why they they weren't more common. I don't see a mistake, i see a brave pursuit of knowledge.
I love your attitude towards failing and the way the video shows your actual process and doesn’t fake any progress nor hides any problems.
Thank you, I am a sub now!
In science you do not fail if you learn and move forward. By sharing, we all learn. Well done and I am now a subscriber.
This was a pretty great video. The editing was good and I was surprised by your 3d animations to explain everything. I've thought about building one of these myself, but now I may hesitate a little more. You're right about learning how the real world works vs theory. You might have made things harder on yourself in some ways, but like you said, you learned a lot from your failures.
Happy to hear that you enjoyed the video! Spent a lot of time on it and also had a lot of fun doing it too! For me, when I decided to start this project, I was hesitant as well, because the only flying thing I ever made was probably a paper airplane. But after some thinking, I realized that if I have some failures, then it's OK! It is my first time after all. Also, the point of this project was, as I said, to learn about how drones worked, so I did many things from scratch. If you want to make a drone that just works, there are off the shelf electronics with software that would make a fully functional drone, with RC capabilities and such. And if the project is meant for deeper learning purposes, then failure is OK! In fact, it's normal. Only issue may be the cost due to iterations, so for that, you can learn from my mistakes which are 1: don't use supercapacitors because you're too curious (use lipos or some other conventional power source), and 2: brushless motors (with an off the shelf ESC controller) would create much better drone response/control compared to brushed motors. If you want to do brushed motors like I did (for cost reasons), then MOSFETs are one of the best and simplest ways to drive brushed motors. You may need RC control for stable flight, cause it won't stay in one spot for long. Hope this helps!
Congrats on the project, loved it!
Really Appreciate man
I really likes journeys because whenever you start a project you just also excited to do it and you spend time on it and you learn through the mistakes and maybe after a shorter period of time you start to give up and then you think that let's give it a one more try and it starts to give you some results this is so fascinating and even if you shut shut down the project you learn a lot out of it
And that's what engineering is
Being an engineer I am very happy to see your journey
Dude this video is amazing.
Dude this video was awesome!!!!
Very inspritational! And well done for getting it flying!
Great work! And a good quality video as well.
Glad you enjoyed the vid!
The most interesting drone I ever built was a tri-copter. I used an old fishing pole as the frame and some cheap brushless motors with 9x7 propellers. And a single metal geat digital servo to steer the tail motor. It took a good bit of tuning to make work well but it was amazing to see it fly around. Too bad so few tricopters exist, they can do some really neat stunts. Although not quite as stable as a quad, they can do things the quad can not do as well. Their flips and rolls are just so different looking. I wish more people would build them. You did a great job on the video and build. Keep it up
You should add a capacitor in parallel to the lipo battery, that should solve the arduino powering off caused by voltage drop of the battery without using a boost converter
That could work for the initial start up, and a cheaper solution too!
Welcome to the club!
Insane work man, keep it up. Love seeing the integration hell 🎉
What a great project maybe you can fit a flyback diode after the mosfet to protect the circuit?
bro you are incredible !🙇♂
Great video ! You made me want to attempt it as well. Thank You :)
Very interesting, I am on a similar project now. Before custom drone I built an FPV with existing components so that I can compare my experiment and I see they all use a capacitor against fluctuations. With supercapacitors they are probably affecting the center of mass and causing the flip, I would try to face them downwards so that they add to stability. Also you can account for those things in your code. For hovering in place probably you will never get there :) as of my current understanding is that you also need a way to measure distance from ground and an optical flow sensor. I am still looking for a good way to have a test setup similar to tripod. Great video
Dont understate how impressive 3 seconds of flight time is. That is really cool!
this is very cool good job man
Those 3D models. I love them!
This was fucking great man!
I recently decided to take the plunge into small electronics and learning how to do some small projects myself.
Some very helpful information in this video on running dc motors! :)
Defenetly earned a subscriber here :) Looking forward to seeing more of your content! Glad to see you have some videos posted allready, i was afraid this would be your only one sincly you "only" have 1.7k subs. Allmost at 2k! :D
Again, thanks for the helpful information i didnt know i came for, and great job on the project, and GREAT job on the video! :)
Thanks a lot! Very happy that I was able to get you inspired and to give you some knowledge on the subject!
The reason the drone was flipping over with the heavy super caps, is that the center-of-mass was not centered within the 4 propellers, so the lighter side was being lifted faster, thus flipping the drone.
Great work man, keep going...
great informative and entertaining video bro. keep it up
Good work❤
I think you should revisit the super capacitor design, but try attaching them horizontally to the spokes/arms of the drone to distribute the weight and maintain a lower center mass, also redesigning the PCB where the IMU chip is located at the center of the design, with the other support components arranged to equalise the weight/center mass of the board
Side-note: Betaflight is very common in FPV communities and all of these kinds of flying problems have been solved. I've been building with it for years, and you can put together a drone like this in a couple hours.
Also, yes please come back to it when ur ready
Love super capacitors for drones. I would have used a small lipo to keep the capacitors charges, just a small 100mah battery
The issue is not the brushed morors, the issue is either your software or the weight of the build. I have made a lot of brushed drones in the past, the recent ones we made for a university event were ran in a swarm configuration so plenty precise and each was only 40ish grams which included a small GPS. Your drone seems too heavy, use a smaller pcb, use a mcu with a dedicated Radio like the esp32 it is also much lighter and faster. Use a smaller frame and try using a barometer or TOF sensor to stabilize the drone mid air.
Some things to do, step away from PID, look into LQR or MPC
Karmin filter or the steady state equivalent
Inteoduce some system identification, all your motors will be different so ttying to tune the system just wont work out well unless you tune each individual motor and then tune the system
Lower your center of gravity, put the controller on the underside
what about modeling the drone? With LQR you need an almost perfect model
@brianblasius they're out there done
Really great video but is it possible to replace the Arduino iot with codecell , it provides below advantage
1) it have 9 axis imu fusion
2) proximity sensor to do jump on hand trick
3) its really tiny and lightweight
4) ble can be used to control insted of wifi to save energy
5) esp32 c3 seems good soc instead of 2 chip design of Arduino
Few mechanical design considerations:
I have seen dji inspire flying dead stable because of its hanging Center of mass so if you solder your caps upside down it will help it to stabilize rather then tip over ( what is currently happening) 4 caps at 4 board corners will also work as stands to eliminate ( for lack of proper words ) stickey effect of drones without landing gear
Imu fusion and pid works for drone stabilization but nothing can beat the physically waight balanced drone , you should try to center its cg because currently its tipping to its side really bad
I don't think its a hardware limitation I've seen people making functioning ESP32 drone with brushed coreless motors (I've tried this myself with little success). You got some incredible results you should keep going!
Interesting. I might use ESP32 for the next time I make a drone! More processing power and more features too.
It's definitely software.
Have a look at open source controllers like Crazy bee, ESPCopter, Ardupilot and see the diagram required to make your own flight controller
@@TheTinkeringTechie sounds good! Let me know if you are interested in collaborating on a version two! Definitely a worthwhile project.
I guess you should make more thrust on side where more mass located due or make central of mass
Look into implementing a kalman filter for sensor fusion, that might improve results. Additionally, look into the applying lowpass filtering to the accelerometer and highpass filtering to the gyro. All of this might improve the estimation of the attitude.
Good suggestion, however, I actually at some point had a simple low pass filter, but when changing the gains, it didn't reduce noise to the point of improving drone performance. I did look at Kalman filters on youtube, but was a bit intimidated, and heard that complementary filters would essentially get the same job done, with less computational power. My guess is the lack of motor response, as my motors are cheap brushed ones. At some point in the future when I try to make a better drone, I might use a kalman filter and brushless motors.
@ ah okay, best of luck! I enjoyed the video!
@@TheTinkeringTechie kalman filter is both heavy on computation side and also intimidating to apply. But it is much better and considering vibrations from motors, kalman filter is w than simple complementary filter. If computation becomes a problem(I think it will be for arduino nano) you can use other simplified ver. which are much better than complementary filter. But is the project still on?
@nikaX2000 there is also the possible of a steady state karmin filter where you pre computebthe karmin gains. This reduces complexity and computation
@nikaX2000 Project would probably be continued some time in the future, and it will have new hardware and software solutions (ones that I am reading about in the comments section, like yours!)
Hi, I'm looking to build something similar to this with the MOSFET based motor driver using those same exact DC motors (though they are brushless) but with a pi pico instead, but it looks like you discovered all these issues by actually building it. Also, have you considered using a 5V battery with voltage regulators to the motors to deal with power spikes? Would it be possible to integrate the IMU readings to correct for drift? Also, did you ever try experimenting with how far the motors are from the center? Currently, I've been trying to connect distance sensors to my pico through i2c, but it's taken embarrassingly long because they keep not working. But seeing you work through this process on a similar project is really inspiring, thanks for the video!
Glad the video inspired you! Generally, I won't want to use a voltage regulator if it is to power a motor, because the current that those motors need to draw, multiplied with the voltage drop across the regulator (which is how the voltage is regulated), would most likely go above a typical voltage regulators power rating. Some large capacitors could help with power spikes, filling in the current needs of the motors in a short time period. For the IMU, I tried to get velocity by integrating accelerometer values, but the value itself drifts over time (after a few seconds, a completely still IMU would think it's moving 3 m/s). There may be a way to that I am not aware of though. I could try experimenting with motor distances in a future drone project. Could help.
Bro keep it up. I aspire to be like u!
Quality keeps going up and very impressive with story telling! Self balancing gimbal camera next?
Thanks, I appreciate it a lot! The self balancing gimbal camera idea does sound cool, and practical too. I'll have to add that to my list of ideas for ranking 😁
Boy, do not give up!😢
This video actually thought me something thx
0:50 "crutial for flight stability" is a AI-ah phrase
Better attempt than I could had tried. I get lost with math. My brain just doesn't like it.
your drone is facing the ground effect in the last few clips of it flying. if its higher in the air it will preform better as the ground effect no longer takes place.
Next time try using esc and brushless motor and two PCB
this is really cool! im also building my own drone. haven't gotten to pid controlling yet, but hopefully ill have a bit better luck than you :)
Back emf is probably whats killing the mosfets. Maybe add like diodes ?
Most likely, but it could be heat as well for the 3rd iteration. I did add protection diodes for the 4th PCB design. I didn't mention it, but I did show it when doing SMD soldering (the black rectangular components next to the AOD424 MOSFETs).
Why use the H-bridge chip? You don't need to rotate the propellers on the opposite direction right?
You are correct. However, my thought process was that it's good to have the capability and not need it rather than the reverse. Some cases could be that I soldered the wires backwards, but all I need to do is change a thing or two in code and its all fixed. Or, perhaps I could improve my motor response by powering the motor in the reverse direction that it was moving in to stop it faster (braking mode I believe it was called for those H-bridge ICs).
Why not He might need to do crazy acrobatics
I’m sorry that you only have 8000 visuals. With all the work you did
This is a great video, i wonder how you make the animations?
I used Blender!
I was thinking the problem with the first design (large super caps ) was it looked top heavy. Might be why it kept flipping over.
For the FETS driving your motors you might need protection diodes across the motor s to prevent back emf from blowing them. Also is the FETS you picked good for logic level drive? VGSon voltage should be less than you arduino voltage output high so that the FET turns fully on. This might be why they were getting hot. Cool video though, nice editing!
Glad you enjoyed the video! And yeah, the top heaviness was pretty obvious after a few tries. After frying many MOSFETs on the third design, I did put protection diodes across the motor for the fourth PCB design. Both MOSFET types had max Rdson and min Rdson drive voltage of 2.5V and 4.5V respectively, placing the Arduino's 3.3V digital output a bit below the middle of the two. Not as ideal, but at least when I was searching on Digikey, I couldn't find any MOSFET with a high enough drain current rating with a sub 3.3V Vgs on voltage. The next best thing I did for the 4th design was to choose a MOSFET with a low maximum Rds on of 6.4 mOhms at 2.5V Vgs on, because in the end, it's the resistance that matters. However, the MOSFET still got hot when supercapacitors were used, and this was because of the voltage of the supercapacitors in series, which could get up to about ~13V. After switching to Lipo batteries, with a max voltage of 4.2V the MOSFETs "magically" never get hot! Thanks for pitching in your suggestions, I appreciate the consideration!
4x0.15 is 2.5W so you are MAXING that mofet and without a heat sink you are probably burning it up. ALWAYS build 2x safety margin.
Correct me if I am wrong, but isn't 4V*0.15A = 0.6W? We can calculate power dissipation from the MOSFET with P = IV = I^2*R. The highest on resistance of the AOD424 MOSFET is 6.2 mOhms = 0.0062 Ohms. Let's assume a very bad scenario, and the motor that is driven by the MOSFET pulls 10A constantly. P = 10^2(A) * 0.0062 (Ohms) = 0.62 W, a good amount less than the rated 2.5 watts. You may have put the numbers into the calculator wrong 😉
@@TheTinkeringTechie 4 motors at 5V is 4x0.15A or 0.5A x5 = 2.5W. Peak draw at power up will be much higher. Also keep in mind the power drop across the transistor is also dissipated into the mosfet.
Mosfets power rating is often listed assuming a heat sink, you need to check the datasheet.
I now understand where the numbers are coming from, but is that assuming that all 4 motors are being driven by 1 MOSFET with the motors in parallel? It seems like you are taking nominal current draw of 1 motor (0.15A), multiplying by 4 (4 motors), and then multiplying by source voltage (~5V, though lipo max voltage is 4.2V) which is about 2.5W power dissipation. However, each motor has its own MOSFET to power it, which is necessary to allow for individual control for each motor for active stability. So instead of 0.15A*4 amps going through a single MOSFET, shouldn't it be only 0.15A? Then it would be 0.15A*5V which would be 0.75W. Looking at the AOD424's datasheet, the 2.5W power rating is for when the device is "mounted on 1in2FR-4 board with 2oz. Copper, in a still air environment with TA=25°C", while the power rating for if a heatsink is used is between 50-100W for case temperatures 100C and 25C respectively. As for a sanity check of my calculations, I touched the MOSFETs after running the motors for a long time, and they were essentially room temperature, maybe warm to the touch. Lastly, when I switched over to using Lipo batteries, all the thermal issues seemed to just disappear, probably because the supercapacitors would be charged up to ~13V, and at that point 0.15A*13V would be getting close to 2.5W power rating, and would definitely go above the power rating of the first MOSFET I used. MOSFETs were definitely hot when using supercaps, as I remember touching them after motors ran, and recoiled from the heat. I am not writing this to invalidate your responses, as it seems like you have some experience, but I just want to make clarifications and also as a sanity check on my experience.
Pcbway is the best ahaha
How are you managing soldering fumes? Especially since you are indoors? Please be safe.
Well, at the moment, I just breath out when doing hand soldering, and have a fan for SMD soldering. So, not much 😐
You are wrong that brushed motor can not be used in self hovering drones, I personally have tried brushed motores in a DIY drone that can fly very well. I agree that they are not as responsive as BLDC but there are tons of self hovering drones out there which used brushed dc motor. But great attempt.
You will achieve :D
may be nano response time is slow
what did u use for the animation it looks so clean
I used Blender! And thanks for the compliment, spent a lot of time on it!
We chose to build a drone not because it is easy, but because we thought it would be easy
LOL, exactly!
the fact that arduino is not in the center wouldn't that affect the imu reading
What name the motor and how much cost?
I believe it was these: 8520 Coreless Brushed Motor Set 53000rpm 8.5x20mm + 75mm CW CCW Propeller. About $12 on Amazon.
Bro, you did quit in the end. But the most important thing is the trip, not the destination
your mosfets are dying because the mcu cant drive the gates effectively, the gate of a mosfet is a little capacitor, the mcu gpio cant supply enough current to charge this capacitor in a reasonable amount of time (i mean, to us its almost instant, but its actually very slow), this means your mosfet is spending a lot of time not fully conducting, and acting as more of a resistor than a switch, you need a gate driver ic, itll turn the low current from the gpio into a nearly instant change in mosfet state(10s of ns, as opposed to tens of uS)
unfortunately if you're using a mosfet with PWM you cant just take the mosfet iDS rating, you need to calculate the power dissipation of the mosfet, or you can just use a slightly overkill mosfet and worst case scenario slap a heatsink on it, but again a gate driver is a must
also the reason your motor randomly turned on when flashing code is pretty normal, the gpio's are unpredictable when flashing, maybe using a swich that turns off the gate driver would be a good solution for that
Good suggestions! For this project, I wanted to keep the components to a minimum to get the general ideas of drone design. When I try again some time in the future, I'll most likely go more in depth with more components. As for the GPIO output current, I did test on a circuit simulator (Circuitlab I believe) with the 0.007A output current and square wave input into a capacitor, with the same capacitance as the gate capacitance, and I saw fairly square output voltage, indicating that the capacitor charges fully and fairly quickly (no saw tooth, no obvious rounding of edges). I recreated it here: www.circuitlab.com/circuit/32abgzvfa52q/aod424-simulation/. Actually, zooming in, I can see the transition time of about 10 uS, so you're right! Gate drivers would be a good bet. If I wanted to keep it simple, would a BJT work well enough for that? As for not being able to use the iDS rating for PWM, is it mainly because the mosfet spends more time in a higher Rds on state because it turns on and off quickly, and thus has to go through the higher Rds on repeatedly, generating more heat? In that case, would it be ok to eyeball by looking at the highest Rds on possible, then calculate power dissipation with the highest estimated current that would go through the mosfet? That's what I did, and the "worst case" was pretty far below the 2.5W (for the 4th design). That conclusion was consistent (when using Lipos), probably because the supercaps would need to be charged at a higher voltage (~13V) as opposed to 4.2V for the Lipos. Thanks!
Great video , i am building one my self maybe we could colaborate and share our own experince and maybe we could make it fly what do u think?
17:30 is that just me or is the animation reversed? The previous animations only showed the specific PID term discussed instead of a fusion of all of them, yet here the D term appears to cause the rotation instead of counteract it as it should. Minor nitpick, otherwise this is truly a great video, I expected to see way more subscribers when I scrolled down to type this!
Thanks for the compliment! What I was going for in the animation was to show the value that the D term would be putting out as a result of the speed of rotation. The D term is meant to "go in the opposite" direction as the P and I terms, and increases in value as the drone rotates faster. That is why it looks reversed to P and I, as it is meant to prevent overshooting. It's like barreling down the street in your car to try and reach the red light as fast as possible (P and I), but you resist going too fast (D) as you don't want to drive through a red light. Unrealistic example, but that's what I got at the moment.
@@TheTinkeringTechie I do understand that. I've worked on PID controllers for a good while myself. But unless I'm missing something here: Yes, your vectors indeed get longer with respect to rotation velocity. But usually the D term damps velocity, so the force output of the motors would have to counter the rotation. I believe what you have pictured would be a negative D term.
i might have an idea as to why your mosfets failed repeatedly.
you may have been reading absolute maximum current this should never be used under normal circumstances, generally speaking this is what you factor in as worst scenario, never reached under normal operation. to add confusion to noobs it is usually at the top of the datasheet.
Yeah, in fact, the last mosfet, AOD424's datasheet ONLY has the absolute maximum ratings, no recommended. Though I tried to find a sort of overkill mosfet, atleast what I think is a bit overkill, for the last iteration.
@@TheTinkeringTechie actually i just checked and it provides the other info outside of absolute max but you have to calculate it yourself to get a rating in amps. i think it's good practice to calculate it yourself anyways because the current ratings are temperature dependant and not everyone has the exact conditions as the tests used in the datasheet for their particular use case.
@@TheTinkeringTechie oh and for overkill solutions i typically use 30n06le with through hole packaging, not sure if they're still in production but i keep a box full of them for whenever i just need something to work and can't be bothered to calculate the load requirements. datasheet says absolute maximum ratings for current is 30a but using a reasonable heatsink (the one reason to use a through hole component in 2025) ive regularly pushed 75+ in 5s bursts and 60 continuous without issue. the gate voltage threshold says 1 minimum and 2 max but through my testing and their own graphs 5 - 8v is what you should actually use.
I Love You🥹
If you were truly Korean or Japanese, you would have done better.
😔