Recently tested 96kHZ with thrust linear at 25. Drone flys amazing, ultra smooth, and very responsive. I could even reduce filtering as the motors are much cooler now. Prop wash is completely gone. I was so impressed that I am switching all my quads ESCs to enable 96kHZ.
Great insights. This perspective is just from the ESC, but to properly understand what happening we should be looking at ESC and motor together. A motor coil is an inductor and has a resistance, so unlike the MOSFET being driven by a digital signal, the coil will alter the power waveform. (a ramp at constant rate of change vs instantaneous vertical change in power) Depending on kV of motor, it's inductance (L) and resistance (R) will vary and have a power ramping time. This is governed by τ = 1/(R*L). It takes ~1τ to ramp 63% of power and ~5τ to ramp to 99% of power level driving the inductor coil (ie: motor windings). The point I'm working towards is that at lower PWM frequencies; less than 5τ (the natural frequency a motor coil), the coil will not become fully energized prior to the MOSFET turning off. Thus actual power would be less than expected if just considering theoretical square wave power source. At higher PWM frequencies there will be less variation a motor coil energing (de-energies) so coils will provided a more consistent power level with what is expected. The effect of motor coil on power soured by the MOSFET could be viewed by examining one of 3 phases powering a motor on an oscilloscope. (my insight into ESC firmware is limited, so not sure if ESC make compensations for this effect; but to do so they would need parameters on motor kV, (ie: R, L values) unless making generic assumptions) This deeper detail than the video; looking at the square MOSFET driver waveform and duty cycle. Thought it worth mentioning, for possible future topic (or those wanting more detail).
I've been flip-flopping between 24khz and 48khz. I've tried TL to help give better low throttle torque, but I still find 24khz to be better in this regard. For super stable cruising even in windy environments, 24khz just seems best. Now for propwash, I don't know why but 48khz almost always seems better. This is all anecdotal of course :) Awesome data here though! It will be fascinating to begin to understand how these ESC settings contribute to various flight related issues
Really interesting,many thanks.I,m off to Spain soon..so I can experiment more easily with the own frequencies...(e.g.its not raining,or blowing a gale,DAILY,..like here in England!)
Finally catching up - I'm amazing at how dead time dependent that is, even at the low end for braking torque, that's crazy. 7" Is plenty happy on 48kHz, but yes, adding thrust_linear into the equations drastically helps with control at idle.
Built my first drone in 10 years last week (yeah i'm old)... used stock betaflight settings and stock timing (24kHz) on a 7" 1300kv, mamba F722 build. Motors so hot i couldn't touch them after 5 min flight. 1350 mAh of the 6S battery used. Just tried setting the max and min pwm to (as an experiment) 128k in blheli_32. The motors barely get warm now, and i'm getting 30% greater flight time. Flew for 6 min and only 900 mAh used. I thought my telemetry was borked, but the charger confirmed it! Big yaw inputs cause a throttle blip for some odd reason... but I think i'm headed in the right direction. Will probably set a lower min PWM freq, but will try to compensate with PIDs before that.
This is amazing. This shows why 48khz is recommended for small whoops motors since they have less inertia so the initial and braking torque doesn't present a great issue. But the added thrust allows for more efficient energy usage, which is critical with the tiny lipos they use. Very interesting info , thanx!
Until now I had 24 kHz on all my 5" quads and 48 kHz on 3" and smaller but I will set my 5" to 48 kHz and give it a try. Of course, rapid breaking is desirable but I doubt I will notice and smoother motors are always nice.
Fun FYI, when dead time > 5% of the PWM period (50kHz = 20us period, so dead time should be < 1us) it increases low frequency harmonics which will reduce delivered power by a few percent and can also increase motor heating.
Been going back and forth with 24 and 48. It depended all on the build, but for bigger than 6" builds I've had most of the success with the tuning on 24khz, but then again with a lower timing than 23 degrees.
Take an esc and look at the output pulse length to the motor at 24khz and 48khz on the scope. Its not proportional at the same dshot value. You would expect the pulse at 48khz to be half of 24 but it won't be.. It will be slightly higher.
Regarding whoops, it is my understanding that the 1S whoops doesn't have a FET-driver and that the MOSFETS are run directly from the BB2 MCU. That gives the 1S boards an incredible deadtime of just 5. If you look at the almost similar 1-2S boards, they need a FET-driver and have more normal deadtimes of around 50. This makes the 1S boards (like Mobula6) prime candidates for 96 and 48khz.
I get desyncs in 48khz but no issues on 24. (With whoops blhelis) or I think desyncs... Heavy or sudden throttle changes and it flips or motors cut out. Or if I cut throttle after a high throttle maneuver the motors spin up instead of slow and the quad climbs instead of decends.
The 'UAV Tech BF Preset for 5-inch Quads, With Go-Pro' (up to about 700 grams AUW) has two options for PWM: 16 & 24kHz or 48+kHz > If in BL Heli-32, I have it set for 24kHz (Low) and By RPM (High) on an ESC with 128kHz capability, ?> Which option should I select? 16&24kHz or 48+kHz --- Both ?
I run all my quads on auto timing recommended by Bardwell for extra smooth response. Picked up that tip 2 years ago and since then all on auto. Bad tip? 23 better?
I also found this to work best, but additionally I had to increase the PIDs overall by about 20% on my 5 inch quads. On one quad I got some issues with D oscillations which are controlled really nicely with tpa of 0.8 at a break point of 1250. Posted an extended post on that over on FB just a few minutes ago, link to the blackbox log is there. facebook.com/groups/291745494678694/permalink/945354222651148/ Still you can feel the loss of authority when hovering in front of an obstacle. So I am staying on 24khz with my freestyle builds. 48khz is the way to go for racing rigs (for me).
Seems like we need a firmware capable of adjusting PWM frequency on the fly based on erpm of the motor so we get the best of both worlds. Low end torque at slow speed with lower pwm timing and more top-end thrust and smoothness in the upper rpm bands with higher pwm timing.
I've stuck to 24kHz on my bigger quads, but 48kHz on my smaller ones. As the frequency goes up, resolution also goes down and this is very noticeable on my bigger quad that just feels lackluster at anything above 48kHz. With lackluster I mean it smoothes out the throttle band too much and feels much more floaty.
There is a big discussion going on between RC crawler enthusiasts in the moment....What kind of esc would be suitable to have a , lets say 1800Kv motor, run smooth and ....very important silent ? 48kHz....more....?? Awesome video BTW ...kudos !
@@uavtech Yes I know ...holmes hobbies introduced a esc recently but its limited to 4s and the BEC is not what they thought it was. What would you consider a good ESC for low rpm running ? I read something about neutronrc coming out with a Gmodel esc that has AM32 firmware.
I usually use 48 with thrust linear but I do not run at the recommended setting of 25. After testing and finding that it did work, I got curious whether that high a setting was necessary. I experimented a bit and fount 10 to 12 was enough to clear up the issues I was getting with 48khz PWM
First of all, great content! I have a question, since I have developed my own flight control hardware and software, I am now getting interesting in ESC and how they work. Perhaps you can answer this? : how can a ESC know when to apply active breaking? For me its strange since FC only sends "power commands" input to ESC and the rpm will be accordingly all factors considered. BR Martin
I wonder if it's possible to make PWM frequency dynamic in ESC firmware... 48kHz at high throttle values for more thrust, 24hHz at low throttle for better breaking and more low end torque (less probability of de-sync on throttle chops/on reversed flow through props)
I know APD ESCs do it. But that just means I know they have an Auto option. I have no idea what it does or if it is even better. Like Timing, Auto might not be better.
@@Krunked the last firmware for the fettec ESCs has some kind of dynamic pwm. But you also have to take in mind that the fettec/kiss ESCs are simulating a sinus curve and not a fixed rectangular one. blog.seidel-philipp.de/kiss-sinhybrid-sinwave-backemf-hybrid-kommutierung/
Great video, could this be the reason for the throbble issue I had that went away When I switched from 48KHz to 24KHz PWM? As that would make sense, as the issue only occurs When you chop the throttle to a very low idle and the motor breaking torque would Be lower at 48KHz?
This is very interesting - I’ve been having a lot of low throttle oscillations using a blheli32 40a hobby wing 20x20 esc. Do you think dropping down from 48khz to 24 would fix that?
Thrust is the same for the same output power.. please contact me for details.. look at 10:34 in the video . you can find the project here. github.com/conuthead/AM32-MultiRotor-ESC-firmware.
0:49 literally wrong, like not true at all. escs use "on/off" and not "in between" because it requires a lot more components to achieve that "in between". and it doesn't have that much difference on the motor. But it is possible to have an ESC that does output smooth voltage radher than pwm , it's just much more complicated