I like so much a project called nematoduino. It uses the mapped neural connections of a nematode (a worm) to simulate the brains in the nematode inside an Arduino board. Maybe you can port that code to the board that you build?
i have some tiny bots and learning about pcb design to make the bots even smaller. they are currently 35mm x 40mm x 30mm but i can get them much smaller with different motors, an all in one pcb and a small 50mAh battery.
Nice work! Do you think you might sell your controller boards to other folks? I'm not doing Sumo, but it appears you've designed an excellent board that I might want to have around for an as-of-yet conceived project. Also as far as using the motor specs to select your cap values it seems that is mostly to save cost by selecting the smallest cap that will work. Do you think you could you build "generic" boards to have around for future applications by selecting the bulk capacitors based on the max specs for the motor driver instead? Matching the high-power current and voltage specs of the 8245 driver with the largest (tallest) electrolytic cap that will fit in the board footprint? These would be versatile boards suitable for many applications, though they would be over-kill for some. I looked at the TI video you recommended and it seems to say that using a larger bulk capacitance is (in general) is a desirable trait, though from what I saw, it might affect braking for some techniques. What do you think?
Yeah, you’re completely right that I selected only that much bulk capacitance to save cost and space. Actually running 64A (32 per motor) would need a ridiculous amount of capacitance to be stable, so I didn’t add that much in. My all-in-one controllers are meant to be ‘all-in-one’, that is, working on multiple platforms and robots while being a single board solutions. Since I’m 13 as of now, I’m in no position to sell my boards, but I can sell the designs and manufacturing rights to other people if they have the means to manufacture and distribute the boards. Thanks 😁
Hey, really enjoying watching your journey, that's a really nice board, especially considering you're just starting out! I've also been designing a custom drive and control PCB for my beetleweight, it was the only way to fit all the components I need in the space I have! I've just ordered rev2, using a Seeed Xiao RP2040 and 4x TI DRV8256 motor drivers (via a PCA9685 for PWM/port expansion), plus an IMU. Four independent drivers for the four motors mean I can do torque vectoring, and the IMU should allow me to close the loop on the control system. I have a couple points to make that popped into my head while watching. The first is that I don't think you need a buck converter to power the MCU; an LDO like a LDL212PU50R can supply 1.2A at up to 18V input, and only requires 3x3mm board space and a couple of capacitors. The load on the 5V rail shouldn't be too high, so while the difference in efficiency might be a large percentage difference, the actual power draw will be negligible compared to the draw from the motors and you reduce the BOM cost and count considerably and free up a little space. Second is with regards to your capacitor section. Don't forget that in these circumstances fully charged lipo battery is almost an ideal low inductance power supply! The internal resistance is tiny and the cells are rated to ludicrous burst discharge ratings - in fact inside my spot welder it's just a really big lipo cell and some mosfets. With a good lipo cell, short power leads, a low inductance PCB design you'll be able to get away with fewer capacitors than you would with big long leads on a bench top power supply. I actually don't have the Z height for any electrolytic caps on my board, so I've simply piled on as many 10uF ceramics as I can physically fit in the space available and crossed my fingers. Finally, a point on driver selection. While choosing a driver based on the motor stall current might be fine for sumo bots, where you can jump in and switch off a stuck bot before it burns a motor out, I would suggest against it for combat robotics. A stalled motor can easily burn out, and you'll rarely see full stall current during operating conditions, so it can be sensible to choose a lower power driver (or apply a current limit if the driver/controller allows it) to avoid motor damage if a wheel jams during a match. While it doesn't really affect the outcome of a fight, it does save on parts and repair time! In my bot I'm using custom Nerf motors (Out of Darts Valkyrie), which on 3S have a 14A stall current, which is absolutely insane for a 130 sized motor. If I spin up a corner of the robot drive with a bench top supply and grab a wheel it doesn't go much above 5-6A until almost totally stationary, hence my choice of 6A drivers. That's a lot of words and I hope it doesn't come off like I'm trying to lecture you! You know your project better than we do, I just enjoy the technical discussion and hope to share some things I've learned both during this project (which has taken almost a year, eek) and a decade in engineering. I think if I share a link to my bot's build thread I'll probably get caught in a spam filter, but if you want to check it out it's called Fatal Deviation and there's a public thread on the Bristol Bot Builders' forum. I will eventually be publishing the whole project on my github, but I need to finish it first!
Thanks so much for this comment, I really enjoyed reading it :) Good point about the motors burning out, I haven't thought of that since I just wanted extremely efficient operation of the PCB itself regardless of whether the motor drivers could be too powerful for the motors or not. You can set an trip current for the motor drivers that I've chosen though, I just decided not to, and this trip current can be well under the 32 amp rating. That means that you get an appropriate trip current while not constantly hitting the motor driver's maximum ratings, which would improve longevity and probably also power consumption since the internal MOSFETs have a lower resistance. As for being able to use an LDO and not a buck converter, I made an ESP32-S3 board with LDOs with similar ratings (16V input, 1A output, 2x2mm WSON), and they caused the entire board to heat up to an unacceptable degree. It burns to the touch. My antweight combat robot control board included these same regulators and I was unsatisfied with the amount of heat that was generated by them, even though the board was probably only 30 or 35 degrees celsius. (I really don't like inefficient circuitry lol, even if it works and it's fine to use). The ESP32-S3 chip draws lots of current when transmitting and receiving RF signals (like, 100mA or more) so it was not good. As you probably already know, an LDO's efficiency is essentially just its output voltage divided by its input voltage, which would mean an 18V input, 5V output LDO would be less than 28% efficient. Buck converters hit into the 85% - 95% efficiency ranges, so I would much rather them... LDOs are still great devices for low current though. Anyways, thanks so much for watching and leaving your comment, have a great day!
@@akshaypadmanabuni2075 I just posted a new video on the PCB design and schematic associated with this board. It really varies from driver to driver though, and I gave some advice in that new video. Thanks 😁
everyone talks abt the esp32 being better than arduino but thats not really fair cause its much more expensive. the esp8266 tho, is more powerful but also the same price! though it has very low current output (
hey, I like to build 3Kg sumo robot too! Nice video. This video almost feel like a design review. I have some questions/comments. @2:06 (edit @2:16) You mention "consume 28A requiring anti power switch?" which is not true. You don't need a anti spark switch because you consume 28Amp. You need this of the bulk capacitors. At t=0 a cap is a short circuit. So when you plug the Xt60 of you lipo, the capacitors act as a short circuit thus creating this huge spark inside the XT60. The softstart prevent this spark. Not related to motor current. @2:44 you say you want to use the LMQ66430 because it output 3Amp. That's not a valid reason, you need to show me data that proves you really need 3 Amp. A table of all the peripheral connected to this board with actual measured current consumption would of been great addition to prove the reasoning. @8:22 regarding MLCC, you need to consider "DC bias". For example check bottom of the TDK product page of the C2012X5R1V226M125AC, you will see the DC bias graph showing that if you apply 16V to that 22uF capacitor, its capacitance is now 2.35uF!!!!!!!!!!! Hence why you need more in parallel to compensate for DC bias. I was not aware of desmos, that is neat. We use MathCAD. All the other components explanation is great. PCB layout is an other beast. Try to do some internship here in the Bay Area, you will love it!
Hi, thanks for your comment. At 2:06 (wrong timestamp btw lol) I said this to reduce the more complicated explanation. I am fully aware that the bulk capacitors are the reason why, not the current draw of the motors. It actually is related to the motor current, because the more motor current you need, the larger the bulk capacitors you require, and therefore the more current the capacitors will draw on startup. So, my claim is true, but it isn't simply just cause and effect as I might have made it sound like (my bad). Yes, I don't need 3 amps output on 5V, but the LMQ66430 makes for a smaller implementation and the fact that it can output 3 amps tells you that the MOSFETs inside have a lower Rds (on), thus yielding higher efficiency. Higher efficiency is always a valid reason. Also, this allows for attaching high power servos or other motors which may run on 5V. I know exactly what "DC bias" is, and I am aware of its affect in MLCCs. I thought that it would only halve the capacitance, since I saw that most graphs showed this during my research. I knew it got worse with the package size, but even the slightly larger 1206 capacitors with the same ratings have a poor DC bias too. I didn't check these graphs. Thanks for letting me know :) I did have two in parallel, one for each motor driver, but, that amounts to about 5uF... I could've used more in parallel but this would've made routing with adequately thick traces much harder, since I already had predetermined space requirements. Also, I didn't want to do double sided assembly because I had to assemble this board myself with a hotplate. Besides, I already had 1880uF of electrolytics in parallel to deal with the ripple current requirements of the capacitors. Also, I'm in the southern hemisphere bro 😂Have a great week man
Hi, I use Premiere Pro 2024. The subtitles (since that's probably what you're interested about) are made from the AI generated transcript of Premiere Pro, which is then transferred into captions (text) which is then converted into graphics which is then animated with the bouncy effect. I used all of the PCBs to some degree, just in different robots. Would you like to know more? Thanks :)
@@Hukkinen the board has a few issues like the power switch not working, causing massive sparks in the battery connector. I would be happy to put new boards on PCBWay’s community ordering page when they work though 😁
Hi. This video got me 50% more average view duration than my other videos. That makes me more money, and brings in more viewers and subscribers. I will continue to refine this editing style until that number goes up even more. Thanks.
What kind of robot would you build with a custom all-in-one controller?
I like so much a project called nematoduino. It uses the mapped neural connections of a nematode (a worm) to simulate the brains in the nematode inside an Arduino board. Maybe you can port that code to the board that you build?
@@3238juan Cool idea 😁
i have some tiny bots and learning about pcb design to make the bots even smaller. they are currently 35mm x 40mm x 30mm but i can get them much smaller with different motors, an all in one pcb and a small 50mAh battery.
@@superradmaker impressive! Hope it goes well 😁
Electrical engineer pushing 40 here, being impressed with your videos. Good job!
Thank you!
This kind of detail, resources provided and reasoning about your choices, logic breakdown of sections is invaluable when learning. Thank you
@@JonathanDann thanks! I posted the second part which goes through the schematic and PCB design, if you’re interested. Have a great day!
im just learning pcb design to make as small a pcv as possible for my robots and this is incredibly resourceful! thank you so much
You’re welcome 😁
Awesome as always. Getting started with learning how PCBs work as a hobby and various forms and uses of capacitance is quite confusing.
@@BamTehBrent thank you so much for the nice comment dude 😁
Amazing, looking forward to part 2!
@@makeradd3494 Thanks 😁
Nice work! Do you think you might sell your controller boards to other folks? I'm not doing Sumo, but it appears you've designed an excellent board that I might want to have around for an as-of-yet conceived project.
Also as far as using the motor specs to select your cap values it seems that is mostly to save cost by selecting the smallest cap that will work. Do you think you could you build "generic" boards to have around for future applications by selecting the bulk capacitors based on the max specs for the motor driver instead? Matching the high-power current and voltage specs of the 8245 driver with the largest (tallest) electrolytic cap that will fit in the board footprint? These would be versatile boards suitable for many applications, though they would be over-kill for some. I looked at the TI video you recommended and it seems to say that using a larger bulk capacitance is (in general) is a desirable trait, though from what I saw, it might affect braking for some techniques. What do you think?
Yeah, you’re completely right that I selected only that much bulk capacitance to save cost and space. Actually running 64A (32 per motor) would need a ridiculous amount of capacitance to be stable, so I didn’t add that much in. My all-in-one controllers are meant to be ‘all-in-one’, that is, working on multiple platforms and robots while being a single board solutions. Since I’m 13 as of now, I’m in no position to sell my boards, but I can sell the designs and manufacturing rights to other people if they have the means to manufacture and distribute the boards. Thanks 😁
Hey, really enjoying watching your journey, that's a really nice board, especially considering you're just starting out! I've also been designing a custom drive and control PCB for my beetleweight, it was the only way to fit all the components I need in the space I have! I've just ordered rev2, using a Seeed Xiao RP2040 and 4x TI DRV8256 motor drivers (via a PCA9685 for PWM/port expansion), plus an IMU. Four independent drivers for the four motors mean I can do torque vectoring, and the IMU should allow me to close the loop on the control system.
I have a couple points to make that popped into my head while watching. The first is that I don't think you need a buck converter to power the MCU; an LDO like a LDL212PU50R can supply 1.2A at up to 18V input, and only requires 3x3mm board space and a couple of capacitors. The load on the 5V rail shouldn't be too high, so while the difference in efficiency might be a large percentage difference, the actual power draw will be negligible compared to the draw from the motors and you reduce the BOM cost and count considerably and free up a little space.
Second is with regards to your capacitor section. Don't forget that in these circumstances fully charged lipo battery is almost an ideal low inductance power supply! The internal resistance is tiny and the cells are rated to ludicrous burst discharge ratings - in fact inside my spot welder it's just a really big lipo cell and some mosfets. With a good lipo cell, short power leads, a low inductance PCB design you'll be able to get away with fewer capacitors than you would with big long leads on a bench top power supply. I actually don't have the Z height for any electrolytic caps on my board, so I've simply piled on as many 10uF ceramics as I can physically fit in the space available and crossed my fingers.
Finally, a point on driver selection. While choosing a driver based on the motor stall current might be fine for sumo bots, where you can jump in and switch off a stuck bot before it burns a motor out, I would suggest against it for combat robotics. A stalled motor can easily burn out, and you'll rarely see full stall current during operating conditions, so it can be sensible to choose a lower power driver (or apply a current limit if the driver/controller allows it) to avoid motor damage if a wheel jams during a match. While it doesn't really affect the outcome of a fight, it does save on parts and repair time! In my bot I'm using custom Nerf motors (Out of Darts Valkyrie), which on 3S have a 14A stall current, which is absolutely insane for a 130 sized motor. If I spin up a corner of the robot drive with a bench top supply and grab a wheel it doesn't go much above 5-6A until almost totally stationary, hence my choice of 6A drivers.
That's a lot of words and I hope it doesn't come off like I'm trying to lecture you! You know your project better than we do, I just enjoy the technical discussion and hope to share some things I've learned both during this project (which has taken almost a year, eek) and a decade in engineering. I think if I share a link to my bot's build thread I'll probably get caught in a spam filter, but if you want to check it out it's called Fatal Deviation and there's a public thread on the Bristol Bot Builders' forum. I will eventually be publishing the whole project on my github, but I need to finish it first!
Thanks so much for this comment, I really enjoyed reading it :) Good point about the motors burning out, I haven't thought of that since I just wanted extremely efficient operation of the PCB itself regardless of whether the motor drivers could be too powerful for the motors or not. You can set an trip current for the motor drivers that I've chosen though, I just decided not to, and this trip current can be well under the 32 amp rating. That means that you get an appropriate trip current while not constantly hitting the motor driver's maximum ratings, which would improve longevity and probably also power consumption since the internal MOSFETs have a lower resistance. As for being able to use an LDO and not a buck converter, I made an ESP32-S3 board with LDOs with similar ratings (16V input, 1A output, 2x2mm WSON), and they caused the entire board to heat up to an unacceptable degree. It burns to the touch. My antweight combat robot control board included these same regulators and I was unsatisfied with the amount of heat that was generated by them, even though the board was probably only 30 or 35 degrees celsius. (I really don't like inefficient circuitry lol, even if it works and it's fine to use). The ESP32-S3 chip draws lots of current when transmitting and receiving RF signals (like, 100mA or more) so it was not good. As you probably already know, an LDO's efficiency is essentially just its output voltage divided by its input voltage, which would mean an 18V input, 5V output LDO would be less than 28% efficient. Buck converters hit into the 85% - 95% efficiency ranges, so I would much rather them... LDOs are still great devices for low current though.
Anyways, thanks so much for watching and leaving your comment, have a great day!
@@h1tec aha, I forgot how power hungry the ESP was a buck converter sounds like the right option then!
Thanks! :)
Always like your videos man, keep it going!
Thanks!
Are there any online resources I can learn to create a motor driver from?
@@akshaypadmanabuni2075 I just posted a new video on the PCB design and schematic associated with this board. It really varies from driver to driver though, and I gave some advice in that new video. Thanks 😁
everyone talks abt the esp32 being better than arduino but thats not really fair cause its much more expensive. the esp8266 tho, is more powerful but also the same price! though it has very low current output (
Graphical prezentation of this video super, excelent.!!
hey, I like to build 3Kg sumo robot too! Nice video. This video almost feel like a design review. I have some questions/comments.
@2:06 (edit @2:16) You mention "consume 28A requiring anti power switch?" which is not true. You don't need a anti spark switch because you consume 28Amp. You need this of the bulk capacitors. At t=0 a cap is a short circuit. So when you plug the Xt60 of you lipo, the capacitors act as a short circuit thus creating this huge spark inside the XT60. The softstart prevent this spark. Not related to motor current.
@2:44 you say you want to use the LMQ66430 because it output 3Amp. That's not a valid reason, you need to show me data that proves you really need 3 Amp. A table of all the peripheral connected to this board with actual measured current consumption would of been great addition to prove the reasoning.
@8:22 regarding MLCC, you need to consider "DC bias". For example check bottom of the TDK product page of the C2012X5R1V226M125AC, you will see the DC bias graph showing that if you apply 16V to that 22uF capacitor, its capacitance is now 2.35uF!!!!!!!!!!! Hence why you need more in parallel to compensate for DC bias.
I was not aware of desmos, that is neat. We use MathCAD.
All the other components explanation is great. PCB layout is an other beast.
Try to do some internship here in the Bay Area, you will love it!
Hi, thanks for your comment.
At 2:06 (wrong timestamp btw lol) I said this to reduce the more complicated explanation. I am fully aware that the bulk capacitors are the reason why, not the current draw of the motors. It actually is related to the motor current, because the more motor current you need, the larger the bulk capacitors you require, and therefore the more current the capacitors will draw on startup. So, my claim is true, but it isn't simply just cause and effect as I might have made it sound like (my bad).
Yes, I don't need 3 amps output on 5V, but the LMQ66430 makes for a smaller implementation and the fact that it can output 3 amps tells you that the MOSFETs inside have a lower Rds (on), thus yielding higher efficiency. Higher efficiency is always a valid reason. Also, this allows for attaching high power servos or other motors which may run on 5V.
I know exactly what "DC bias" is, and I am aware of its affect in MLCCs. I thought that it would only halve the capacitance, since I saw that most graphs showed this during my research. I knew it got worse with the package size, but even the slightly larger 1206 capacitors with the same ratings have a poor DC bias too. I didn't check these graphs. Thanks for letting me know :)
I did have two in parallel, one for each motor driver, but, that amounts to about 5uF... I could've used more in parallel but this would've made routing with adequately thick traces much harder, since I already had predetermined space requirements. Also, I didn't want to do double sided assembly because I had to assemble this board myself with a hotplate.
Besides, I already had 1880uF of electrolytics in parallel to deal with the ripple current requirements of the capacitors.
Also, I'm in the southern hemisphere bro 😂Have a great week man
Really nice. What software do you use to edit your videos?
And for the PCB, which one did you used?
Congratulations!
Hi, I use Premiere Pro 2024. The subtitles (since that's probably what you're interested about) are made from the AI generated transcript of Premiere Pro, which is then transferred into captions (text) which is then converted into graphics which is then animated with the bouncy effect. I used all of the PCBs to some degree, just in different robots. Would you like to know more? Thanks :)
Loving your vids!! 🎉
@@CordSchneider thanks!
It would be nice to click to PCBWay to order myself this interesting integrated module, if the price were right =)
@@Hukkinen the board has a few issues like the power switch not working, causing massive sparks in the battery connector. I would be happy to put new boards on PCBWay’s community ordering page when they work though 😁
Good work man!
Thanks! 😁😁😁
Nice video, keep it up, thank you :)
@@RixtronixLAB thanks!
Nice !
@@Nayel-Khouatra thanks!
Genius! 😎🤖
m8 someone as intelligent as you at your age shouldnt need that much help with the SAT's :p
lol well thanks… Still, getting a 1600 is going to be tough. I’m aiming above 1520 at the very least though lol
100% agree
Thanks so much guys :)
I’m on TH-cam.
Not TikTok.
Please don’t make TikTok’s on TH-cam
Hi.
This video got me 50% more average view duration than my other videos.
That makes me more money, and brings in more viewers and subscribers.
I will continue to refine this editing style until that number goes up even more.
Thanks.
you could misuse an ne555 for a mosfet th-cam.com/video/FKqkRDM0Al0/w-d-xo.html
@@SvenWoldt cool, I’ll take a look later 👍