Motor Commutation Explained: Featuring FETTECs new SFOC approach
Вставка
- Опубліковано 5 лип 2024
- If you loved this video please consider supporting me:
www.aos-rc.com/about
patreon.com/chris_rosser
USA: www.amazon.com/?&tag=chrisros...
UK: www.amazon.co.uk/?&tag=chrisr...
The latest product testing results are always on AOS Labs: www.aos-rc.com/aos-labs
To contact me about ESC/battery/prop/motor reviews and testing please use the form on www.aos-rc.com/contact
Timestamps:
0:00 Intro
0:30 Motor connections
1:19 Back EMF
2:09 Block Commutation
4:48 Sinusoidal Commutation
8:26 Space Vector Modulation
13:46 FETTEC Sinusoidal vs SFOC
15:48 Hardware
16:37 Outro - Наука та технологія
I really like how you show this much detail to us in an easiest way possible.
Thank you.
Best video yet brother! I finally was able to figure out why FOC ends up with less RPM than Trapezoidal control. Never made sense until I saw your example of why it's 86% of Vcc. Thanks
Nice deep dive Chris. All the methods presented here only utilize one FET per motor coil phase. I'm surprised that no effort has gone into using 2, or 3 FET's in parallel for each coil phase. Even with using the block-method, as very pure sinusoidal waveform, or other shape could be assembled (via making use of a Fournier series of the waveform) to produce similar waveforms, but do so with greatest FET efficiency. (such technique is common with solar inverters that produce 'pure-sinusoidal' power.
NOTE: The PWM frequency being used (12:21) is ~40x higher frequency that motor coil phase frequency. (eye-balling from diagram) Thus means the FETs need to be able to operate at these frequencies, while being able to deliver maximum current and voltage. Such high frequencies can lead to more heating (loss in efficiency)
I love new technology, this is awesome, thanks Chris.
This was superb! A potentially difficult topic presented extremely well.
Ive always loved go The technical side to the technology that we love. Thanks for all the work you do!
Thank you Chris for this fascinating Video.
Awesome explanation, thank you!
Absolutely amazing video! Thankyo
Awesome content! Thanks Chris.
Fantastic video. Thank you.
Very interesting. Thanks for explaining this.
Super interesting, thanks!
Great stuff! Thanks mate! Now I’ve got to go lay down in a dark room until my brain stops aching 😂
You got of easy. My brain started to ooze out my ears.
Thanks!
Nice, video can you put up the refrence document or weblink for it. Thanks
A good proof why science behind it is so important for ex .... my self i by a motor based on advertisement and fly like a crazy with out knowing i can get a better one based on tests like Chris dose
Any flight videos? Or does it even fly?
Good pres Chris, I recently moved into ebikes too, and see huge crossover in the knowledge. Now I can actually visualise a square wave controller. I wonder if SFOC would apply to ebike controllers and if the fact the motors are sensored would help with the back emf reading issue?
A sensored drive system (motor + ESC) doesn't need to measure backEMF to guess where the motor is.
nice!!!
Awesome! How long to wait )))) For official sells
What are the "additional motor parameters" the ESC needs to know? Hopefully just motor pole count?
That is super interesting... I look always for solutions for crawlers... you think, that could increase the slow speed torque and control....
Sweet, if it supports 8S when it releases I'll totally upgrade my fleet.
Thanks for all the info, Chris! 😊
Stay safe there with your family! 🖖😊
i wonder how them skew magnet motors would behave, since they are skewed, where is the 50% mark, where is the 0 point cross over, so how would they react in the traditional pwm style drive vs a sine wave driving style, interesting nerdy stuff, keep it up chris
How does trapezoidal commutation compare to sinusoidal? You can enable it on fettec escs in fettec configurator
Trapezoidal is the block commutation afaik
I could really use your help sir. I'm looking for a 20by20 4 in 1 esc for a Flywoo Explorer LR 4 V2 with Flywoo Dave-C-FPV 1404 v2 2750kv motors and a matek H743 mini V3 flight controller. This build will be for long distance and Arducopter. Thank you
What's the theoretical difference in performance (power, efficiency) between SVM and block commutation?
Judging by fettecs graphs it could be about 15% in efficiency which is the rate of their ESCs surviving the first arming.
Would it be possible to make VESCs support SFOC? AFAIK VESCs have shunt resistors for every phase and can also measure voltage on each phase, so my guess is that it would be possible to implement this from software on a VESC... ?
What makes you think Vesc doesn't have that already?
How is this different than Vertiq motor control?
🎉
please make comparion video betwen bldc motor and permanent magnet synchrounous motor
Bonjour, est il possible d'avoir le traducteur, Merci.
Vous pouvez activer les sous titres avec traduction automatique........
This got me thinking about whether anybody is doing an ML based speed controller. It seems like you could drive a specific motor as a generator and use that as a training set to learn ideal commutation. Given the voltage/current sensors I bet it could learn SVM and perhaps some other/similar tricks that folks haven't thought of (even specific to a particular motor)
I don't see how you would go about using SVM in motor control. Maybe some LPC or some micro (few neurons) neural net since you are basically doing regression here. These algorithms are so computationally intensive running them at kHz range would be impossible on those small chips + you introduce latency. And you might gain a few percent at most.
@@GK-dd5ci I agree with you if we assume the ML is running in realtime, but you could make a much simpler speed controller that is parameterized in a few ways, and then do an offline ML process to learn from telemetry on the specific motor (for example in generator mode).
For example, say one of the magnets in a particular motor is a little weaker than the others, this ML could identify that and correct for it by changing some parameters in the realtime controller (not ML)
Chris what's going on with your videos lately? Every other frame is duplicated, making the video very choppy and rough. Might want to have a look at your render settings.
its not every other frame. What you are seeing is a 25fps to 30fps conversion.
Nice video. Im just not sure about the sinusoidal current of a BLDC. The winding isnt harmonic so there shouldnt be a sinusoidal current. Rather a trapezoidal I think. Because when I set my ESC (BL32) to modulate PWM to sinusoidal I get worse performance. Big BLDC motors or EC motors with harmonic winding or synchronous motors should be driven by sinusoidal voltage. However, Ive never measured a BLDC motor so I dont know for sure. Just how they told us in university...
I thought my soldering looked bad I don't feel so bad now lol
What ESC could possibly support SFOC?
Vesc
But they still use pwm drive to simulate real SVM drive. So it's like modified block commutation.
It'll probably be a while before this gets to whoops, but I think it looks promising. Smoother is always better. Also, I wonder what motors driven by this tech sound like? Smoother and quieter? No difference?
Not trying to be a Debbie downer but with Fettec’s reputation of their esc’s catching 🔥 constantly or melting fets off the boards in flight I’m going to avoid them. I like the open source stuff we get, I just hope this new closed source tech gains some reliability for those who give it a try since there’s little support if esc’s start randomly blowing up.
A very big part of that is due to a simple design oversight on their part - in order to support their 'Racewire' LED's ALL the motor pads on a Fettec ESC have VBAT present on them upon power up/when disarmed. Now say you overheat a motor inflight (and burn the windings) or make the beginner mistake of using extra length motor screws into the windings, or just damaged motor wire insulation -now your entire carbon frame is at VBAT! Then combine that with many quad frames getting connected to GND via Antenna SMA / Vista hardmounted to carbon / Metal FC screws etc and you have a recipe for fires.... Oh and Fettec in their infinite wisdom decided to put the ESC current sensor resistor on the negative rail in the ESC so they have literally NO way to detect the fault condition!
wow
test shots vs
its crazy how complex esc really are
from the circuit point of view pretty simple, actually
even from a firmware point of view they're not that complex in the grand scheme of things. but yeah, I agree this much more complex than the naive solution.
I remember you going through this a year or two ago and finding it didn't make a practical difference in flight? Ignoring that it would be cool if it were from ANYONE but FETTEC.
I haven't been paying attention for a few months. What's the problem with fettec?
@@The_1ntern3tI’ve heard a lot of complaints on Facebook groups that their escs burn up pretty easily. My 45a esc is still kicking tho
@@The_1ntern3tas Venrad said there's a fairly consistent complaint of the quality of their ESCs. I on the other hand got baited with their takeover of KISS software only to get stuck with their own software and their replies in private were not consistent with a reasonable company. I will never have anything to do with them.
You're thinking of BlHeli_32s sinusoidal mode. That's different.
That design looks like scaled down VESC (originally designed for electric skateboard), which is not a problem because VESC is an open source project.
Is he hinting that voltage spikes aren't possible anymore and we wouldn't need to add on to the ESC?
when can we try one?! do they need some beta testers? :)
So its a digital esc
Seems so complicated and prone to desyncs.. we'll see how this approach goes in practice. But why not use a sensor on the motors? I don't think it adds more than 1-2$ to the total price of the motor. Also reduces the risk of burning escs and motors in case of accident or turtle.
That's called a flux observer
It's cool you found a way to "monetarily" dissipate power. I wish I knew how to make my FETs perform monetary tasks :-) 5:58
I think you confused voltage and current quite a bit here.
why fps so low. it's distracting
You might want to revisit your "more efficient" statement... efficient how? Switching losses are just that losses hence not efficient.
The losses in switching can easily be less than the losses of driving the phase at max voltage (and current) early and late in timing
Sounds expensive. 💵💵💵💵💸