Making a PCB Motor spin 30,000 RPM FASTER

this is my tiny PCB motor and for the last five  years I have been trying to make it stronger and   faster but here's the twist the motor is just half  of the problem so in this video I'm going to build   four different controllers test switch one is the  most efficient and try to make it smaller my goal   is to use this research to select a new controller  for my PCB Rover and also explore the possibility   of other exciting projects along the way so  let's start by understanding how an ESC Works my   PCB motor is like a brushless motor it has three  phases the only difference is that these coils are   integrated inside this thin PCB to simplify things  imagine the rotor being like a merry-go-round and   the stator is like my hand trying to spin it  the switching controls the speed so by turning   on and off the coils they can turn a magnet on  top there is a total of six steps and for each   step there will always be one phase connected to  vcc another to ground and one that is floating   this was called a six step trapezoidal waveform  and a microcontroller can help us generate these   signals to rotate the rotor these signals will  have to pass through a gate driver to switch   six transistors and a driver like this one has  everything inside which is ideal for my low power   motor now the problem here is that the controller  is running blind one timing mistake will make it   loose synchronization so next we need to find  a way to read and control the speed and there's   two ways how we can do that the simplest plus  one is to put a magnetic sensor near the rotor   this way it can detect the polarity changes of  the magnets inside the rotor the rotor has eight   magnets so from just one sensor we can determine  the RPM from this value the microcontroller can   make small adjustments in time to sync the speed  I confirmed the speed with a laser tachometer that   is why I had to paint the rotor black and put a  piece of reflective tape on it the maximum speed   I measured was around 3500 RPM so this ESC has  a lot of limitations it's the same one I used   in my last project and apart from the poor speed  performance it also gets quite hot both of these   issues can be fixed by using a more powerful  microcontroller but why bother when we can   just replace everything with a single chip this  code free solution can directly drive the motor   generate its own six step commutation waveform  and control the speed from the onboard sensors   so it will kind of do the job of these two chips  at first I tried to order some development boards   for this chip but they were quite expensive so it  made more sense to open altium and design my own   board putting the stator on the same PCB and add  the sensor as close as possible to the rotor if   you'd like to test this development board yourself  the files are available for free and you can also   download altium designer and 365 from the link in  my description one disadvantage of this chip is   that it requires three hall sensors instead of one  I added a bunch of pins because it can communicate   with multiple interfaces but it only requires  one to adjust the speed using the duty cycle   wow the maximum unloaded speed I read with the   tachometer was 33 900 RPM it's going  way faster than I thought it would it's very stable the speed the drop  when I increase the load but I was   still impressed especially because  the motor was operating at much lower   temperatures this is because the duty  cycle of the six step waveform is being   adjusted based on the motor's load in  fact it draws less power without a load there are still some drawbacks with the specific  design the first one is that the motor cannot be   powered with a one cell lipo battery because it's  more intended to be using cars besides this we   also have the extra cost of the three-hall sensors  these take up a lot of space considering the size   of the motor which might be unacceptable for some  applications like this one this tiny propeller is   integrated into the rotor and made from aluminum I  don't have a 5-axis CNC so I got it made at pcbway I was very disappointed because it just  reached 6000 RPM the rotor is lighter   than the wheel so most probably this is  happening because of drag the extra area   of the sensors is also blocking most of  the airway so this is a very poor fan   to solve this I can either add more layers  and place the hall sensors directly underneath   the coils or keep the original design and go  sensorless high speed brushless motors usually   use these type of sensorless controllers so I  want to start by designing my own from scratch   and then try to simplify it I'll be using  the same Hardware I used for the version 4   PCB motor project this ESC supports both sensored  and sensorless controlled but I never tackled the   sensorless part because the rotor was still just  too wobbling I only had one of these prototypes   which I broke for the intro so thank you  pcbway once again for sponsoring all these pcbs as I said before to drive a three-phase motor  you always need to have a pin that is connected   to VCC, another to ground and one that is floating  and as the rotor spins this undriven face starts   generating a voltage which is known as Beck  emf this signal is directly proportional to   the motor speed and its midpoint voltage is  known as the zero Crossing Point detecting   this point can help us determine the timing  needed to keep the motor in sync while also   controlling its speed this is not an easy task at  all because any delays in the code will result in   the loss of synchronization this is why this  ESC has a much more powerful microcontroller   than this one as it is specifically intended to  drive high speed brushless motors now the first   part of this algorithm involves driving the  motor in open loop to get the magnet spinning   and generate back EMF in the coils while we do  this it's very important to start lowering its   duty cycle this duty cycle is going to act like a  speed percentage so in open loop we want it to be   as low as possible as soon as the motor gets  in sync we can start increasing it and this   automatically increases the speed of the motor  this high frequency pwm will introduce noise   to our back EMF waveform so I added a capacitor  to smooth things out and a potential divider to   bring the signal down to the microcontroller's  voltage level tuning these three capacitors is   one of the key elements of this project setting  the value to large can attenuate the signal or   introduce a phase shift the reason why I know all  this is because I have already worked on a similar   project a long time ago but this motor was a  different kind of Beast its phase resistance   is very high compared to other brushless motors  and it barely generates any back EMF so I wasn't   even sure if it would work this could have been  a complete waste of time in fact it took me weeks   trying to tune the open loop algorithm to get it  to spin fast enough to detect back EMF I ended up   opening the motor to 11 000 RPM in open  group but the sensorless algorithm still   wasn't stable so I kept testing the software  until I found that one of the ADC channels   was sometimes outputting garbage data I managed  to fix this problem with just a while loop and   this immediately showed a sign of improvement  but the motor still wasn't holding in sync   so then I added some software filtering and  averaging to help the motor out but the main   issue was that the two back kmf slopes wasn't  symmetrical I decided to ignore one of them   and base my timing on this one point if this  doesn't work I think I'm just going to give up omg i think its working the motor was spinning faster than 33 000  RPM at this maximum speed the Back emf was   getting a bit saturated I didn't understand  completely why this was happening but at   least it was stable and not getting hot it's  also managed to spin the wheel to around 22   000 RPM and the propeller to 18  000 RPM which was very impressive it isn't that visible in the footage but some  smoke was getting exhausted out the only problem   I was left facing with this controller is  the startup sequence because both of the   wheel and propeller wasn't always switching to  the sensorless algorithm this means that I have   to either customize the open loop algorithm  specific to the load or add current sensing   but to do that we need more real estate on the  board which brings us to the next point as fun   as this engineering challenge was this ESC doesn't  really make sense in my mind it's just too big for   the motor so the next step is trying to make it  smaller one chip sensorless controllers are far   more common than sensored ones as their most  popular applications are low power brushless   fans in laptops most of these controllers are  usually designed to operate with standard motor   parameters so I wasn't sure that they would  work with my PCB Motors High phase resistance   and low back EMF in fact from the three developer  boards I tested only one chip worked and it still   involved some tuning this specific chip required  an additional pin connected to the midpoint of   the phases soldering a wire directly to this  point didn't work that well but simulating it   with three resistors did according to the data  sheet this is still okay to do however its speed   still wasn't that high so I decided to try it out  on a PCB to see if it will improve but nothing   really changed this chip was also limited  to rotate the rotor in just one direction   so I made some more research and found this chip  which supports both clockwise and anti-clockwise   rotation it was a bit of a gamble designing  this one with my PCB motor because there were   no development boards available to test it out  on but the results were completely surprising it didn't beat the 33 000 RPM high score but   both the wheel and proper rotated  just as fast as my sensorless ESC it seems that the algorithm it's using is  a little bit different from the one I use   because it's switching the pwm off at the  midpoint to avoid filtering which I think   is quite neat this ESC also requires a  common point and it's datasheet doesn't   mention any simulation resistors but I am  using them here so I try to remove these   resistors and soldered a wire directly  to this point and it works just as fine so technically these three simulation  resistors can be eliminated which makes   the ESC even smaller the only thing that sucks  about this chip is that has a five seconds lock   time out which might not be acceptable  for some applications the exposed copper   heatsink did manage to keep the motor cooler  because air was passing through the vias but   it was also blocking the exhaust so I decided  to connect a non-heating PCB Motor to this tiny   controller which had the same circuit and  this turned it into a little PCB motor fan   this ESC is practically weightless which is  ideal when you start thinking about thrust four of these motors can only theoretically  lift around 5 grams so we still need a little   bit more trust to make a PCB drone feasible but  I have a feeling that we can juice more power   from it by redesigning the propeller and the  motor remember that in the past my PCB Motors   performance was limited by temperature so it was  designed not to get hot at low speeds and now   it's pretty clear that at high RPM temperature  is no longer a problem I think it's important   to start differentiating between high RPM and  slow RPM PCB Motors the sensorless escs are not   going to work at slow speeds because the motor  doesn't generate back EMF the sensor chip also   doesn't have a slow speed limit so I think  the best option to drive the motor slow is   by using a custom controller if we use the same  algorithm it still makes the motor spring quite   a lot but if we replace the square wave with the  sinusoidal one it makes things much more smoother   ways for now I will just focusing on redesigning  the motor for high-speed applications if you don't   want to miss that remember to subscribe and  check out these other two videos if you want   to learn more about PCB Motors special Thanks  goes to all my patrons who directly supported   This research feel free to join if you want  to get a glimpse of my upcoming projects

Channel: Carl Bugeja

Views: 1,806,495

Keywords: brushless, brushless dc motor, brushless esc, brushless motor, circuit, circuit design, dc motor, dc motor fan, design, diy, diy motor, drone, electric motor, electronics, esc, how an esc work, how to, how to make, how to make esc, making, making a pcb motor, micro motor, motor, new motor, pcb, pcb design, pcb drone, pcb motor, pcb motor drone, pcb stator, rc, robotics, rpm, sensorless pcb motor, simple motor, small motor, smallest motor, speed controller, spin, dc motor project

Id: NX7GHqq28uU

Length: 13min 40sec (820 seconds)

Published: Thu Jun 22 2023