VOVANE Electric Dust Blower, 150000RPM Air Duster vs. high speed camera (WHL #121F1)
It’s been a while, and aside from being busy editing and uploading metric shittons of Beach Buggy Racing videos (as of today spanning until the last week of December, on a two-day-release cycle)…
…I have also pondered my options regarding the air duster. I wanna know if this thing really hits 150krpm with all the cheap battery, power electronics, bearings and whatever else is involved. 150krpm for such an inexpensive device sounds so alien to me.
Luckily, we do have a high-speed camera at work. It’s more of a microscope, but zooming in and not having the full air blast onto the lenses might be a good idea regardless. So one Friday after clocking out, I did run a couple experiments.
Safe to say I’m not done processing these, and the absolute velocity of the blades is actually a worthy opponent for the camera. It’s a VW-9000 unit from Keyence, and it is able to run at full 640×480 resolution (how tiny in today’s world!) at 4000 frames per second, or go down in several steps to a strip of a couple of pixels at 230400 fps and 1/900000s of exposure time. A couple months back, we had a sales visit from Photron, and boy is even their (six-figure!) 1MP model an entirely different league of camera.
Anyway, Keyence does ship a Windows tool for simple measurements and exports, but the motion tracker software seems protected by some stupid hardware dongle. Shotcut, my preferred video editor for now, does work fine with the exported .avi video files, but these run in 15fps mode for all frame rates of the actual capture. Adding a real-time clock to the video requires some tricks as outlined here in the Shotcut forums -> Upper limit of more than 99 hours for Timer filter? Suggestion – Shotcut Forum
In essence, the Timer filter does the job, but gets remapped from counting seconds to milliseconds. For the video below, captured at 4000fps, this results in a Timer speed factor of 3.75x (4000fps video encoded in 15fps is a 266.667x speedup; getting to a 1000x speedup to remap seconds to microseconds requires another factor of 1000/266.667 = 3.75). Then, the filter output is set to “SS.SS” or “SS.SSS”, so seconds.milliseconds, but now remapped to milliseconds.microseconds, and then some text overlay to the left and right is added, since the Timer itself cannot display addition text (prefix/suffix or the like). It is somewhat important to stay above a speed factor of 1x, as values below 1x will fail to update timestamps until the next step triggers. Not a problem at 4000fps, very well a problem at 150000fps, the highest setting used for capturing videos with the air duster. For that, the max time that the Timer filter can hold is simply too short (or the camera can capture too much footage at once, depending on your point of view…)
This footage is from running at 4000fps and on low fan speed. This setting is too slow to actually track the blades; on full blast there’s heavy aliasing, and because the blades all look essentially the same, it’s not easy to see how far they have rotated between frames.
There’s however two blades with a slight nick, one at the center, one almost halfway out to the tip, marked with the arrow below. At the end of the video at max speed, it takes slightly above 9 frames for this groove to reappear at the same spot, and say 9.2 frames at 4000fps is 2.3ms per revolution, and that is 434Hz or 26100RPM. And that is super close to the values determined in the original post #121 and the original marketing claims.
-> moved to 1 o’clock position
-> back to original position
(take the time values with a grain of salt, this is not accurate due to the described timing thing in Shotcut. I ran the numbers with the exact figures from the camera FPS readout)
Again, more details once I have processed this further, it might need some workarounds and/or a new version of Shotcut to do the final editing. But this footage once again confirms my photo diode capture with the oscilloscope, and it looks like the 150krpm aren’t just marketing… 😱
