2
\$\begingroup\$

When recently flying a model airplane with friends, I was recording on a smartphone camera (I believe CMOS, 720p, 120fps), and saw, in one of the videos, a flash that was not visible to the naked eye, that covered the whole frame of the video. This occured at the same time that the aircraft collided, about half a meter from the camera. The frame appears to be almost pure white (#fefefc), with the left side (farthest from the brushless motor) having a slightly weaker white (#fefefb), although this is after the video was compressed. There are no visual artifacts or effects in the frames before and after.

The airplane has a single brushless motor, with an ESC (electronic speed control) attached. I don't currently know the specific specs of the motor and ESC, but can readily obtain them later today. The collision happened when the motor was still powered and spinning.

I am suspecting that the sudden collision while the brushless motor was still being powered caused some kind of electromagnetic effect (when it collides, the rotor is mechanically restrained by the ground), but I am not sure what actual effect is going on. I suspected simply a small electromagnetic pulse, but am not sure how it would affect only the camera module and not cause data corruption or crashes by affecting the CPU/memory. It's likely that the camera was set to a very high ISO/sensitivity in order to record at 120fps.

Could someone shed some light on how it was caused by the motor/motor controller itself, and how it managed to cause the camera to register a large flash?

\$\endgroup\$
  • \$\begingroup\$ You're saying that this "flash" is in the first video, near the end, as the plane lands near your feet? But it's not visible unless you somehow examine the video frame-by-frame? \$\endgroup\$ – Hot Licks Mar 29 '15 at 18:20
  • \$\begingroup\$ Or are you speaking of the flash that is sometimes visible at around 0:55? That kind of looks like a compression artifact, though the reason for it is not clear. \$\endgroup\$ – Hot Licks Mar 29 '15 at 18:26
  • \$\begingroup\$ @HotLicks It is sometimes visible, due to the way Youtube buffers and plays videos. It is always visible in my original recording, and I have not been able to replicate it except with other crashes when throttle is still on (which I obviously try to avoid as it damages the aircraft) \$\endgroup\$ – Andrey Akhmetov Mar 29 '15 at 18:38
  • \$\begingroup\$ I suppose if the prop is stopped dead when on full throttle it could emit some sort of EMP. And the video sensor, being exposed with no shielding, would be most sensitive to such a pulse. \$\endgroup\$ – Hot Licks Mar 29 '15 at 19:01
  • \$\begingroup\$ @HotLicks, that is exactly what I am wondering. What physical principles cause such an emission, and what actually happens in the sensor that makes it register a pulse weak enough not to cause issues to the rest of the device? \$\endgroup\$ – Andrey Akhmetov Mar 29 '15 at 19:02
2
+50
\$\begingroup\$

Meet the power plant for the Nutball UAV. It contains, among other things, a servo, a killer brushless fan motor, and a 2.4 GHz spread spectrum receiver.

Without having the ability to perform a physical experiment, we have to perform a thought experiment. The battery recommended for the pod has no controls, save the speed controller for the prop, which is not particularly checking up on the battery pack. Similarly the IR receiver is not concerned with the battery, except to use it.

So, what exactly happened when the loaded brushless fan motor suddenly gets halted?

Well, the battery cannot instantaneously respond to the change - it's simply not possible. Not that there's a whole bunch of energy there, mind you, just a rapid transient in supply voltage. As the transient subsides, the chemistry in the battery hardly noticed.

So we have established the conditions, and we have to ask ourselves; is it more likely that a catastrophic voltage occurred during a disconnection that left the craft operational, or is there a more efficient pulse generator on board the aircraft?

It turns out there is. A radio receiver isn't physically all that different from a radio transmitter, and a spread spectrum system has more bandwidth than, say, an ordinary AM radio. What's more, the one on this device is already tuned for 2.4 GHz operation.

From the EMP wiki;

An EMP arises where the source emits a short-duration pulse of energy. The energy is usually broadband by nature, although it often excites a relatively narrow-band damped sine wave response in the victim. Some types are generated as repetitive and regular pulse trains.

At the moment of the flash, you just happened to be incredibly close to a sharp change in voltage that was being passed to a broadband antenna, filming at a rate high enough to notice it, and pointing several megapixels worth of small antennas at it through the only part of your phone that would be able to notice that it happened.

Now go buy a lottery ticket and share it with me.

\$\endgroup\$
  • \$\begingroup\$ Could the downvoter please explain the vote? \$\endgroup\$ – Andrey Akhmetov Apr 3 '15 at 11:44
  • \$\begingroup\$ @SeanBoddy I +1ed, nice answer, nice research. But why doesn't EMP harm other camera's equipment? Maybe hexafraction could shed some light - if someone else's phone nearby experienced strange acivity. \$\endgroup\$ – RinkyPinku Apr 3 '15 at 13:39
  • \$\begingroup\$ @aim, that's just it- the phones would have to be at a similar distance and attitude to the event, and i suspect filming at a high rate. It might even not appear on a different device - compression and shielding might make it imperceptible on a different device. \$\endgroup\$ – Sean Boddy Apr 3 '15 at 15:42
  • \$\begingroup\$ My phone was the only device within a meter or so, and I had no signs of EMI other than the camera. No application crashes, no data corruption, no lockups or reboots. My camera was also completely usable within the next frame after the event. Maybe the way CMOS sensors work at high frame rates might affect them far more than the onboard CPU or RAM. \$\endgroup\$ – Andrey Akhmetov Apr 3 '15 at 16:31
  • \$\begingroup\$ Yes, by reciprocity a receive-antenna is physically a transmit-antenna. However, the circuitry for RX is significantly different than for TX: a sensitive low-noise amplifier rather than a power amplifier. I'm curious: How would a power supply brownout (which is filtered by capacitors and an LDO before reaching the LNA) cause the LNA to transmit significant energy? \$\endgroup\$ – Zulu Apr 5 '15 at 20:01
1
\$\begingroup\$

My suspicion is that the plane's physical wiring was momentarily changed: likely a transient disconnect of a wire connected to the motor.

Nominally, energy is delivered to the motor through the ESC from the battery via MOSFETs configured in "half-bridge" configurations. The rate at which magnetic fields in the motor grow and decay is limited by the voltage applied across the windings (remember that by Faraday's law, \$\frac{\partial \Phi}{\partial t}=\frac{V}{N}\$) -- a measly 4.2V to 12.6V, depending on how many batteries you've stacked. The MOSFETs cannot allow more voltage than the battery provides, because even if they turn off their body diodes will conduct (and even if they didn't, the MOSFETs would go into avalanche breakdown, somewhere around 30V), so \$\frac{\partial \Phi}{\partial t}\$ isn't high enough to generate the energy at high-frequency needed to interfere with other electronics.

However, if the wires connecting the motor were momentarily detached, \$\frac{\partial \Phi}{\partial t}\$ could be huge, because the voltage across the windings would be limited only by the breakdown voltage of air. This is in the kV range, so you can see how this could cause an EMP containing energy into the MHz and possibly GHz.

As for how the EMP affects the camera... I'm not sure I have a good answer for that, and there are probably numerous valid explanations. Here's my attempt:

The CMOS sensor comprises a large array of pixels, where each pixel comprises a photodiode, a row select switch, and a small amplifier. Because it's so compact, I can't imagine an individual amplifier being affected much by the pulse--let alone the entire array of pixels. However, I can imagine the power supply for all the amplifiers being affected. The power will be provided probably by some sort of linear regulator, which will use closed-loop feedback to regulate its output (as presented by a voltage divider to a differential transistor pair). By Lenz's law, the EMP could cause a voltage glitch at the regulator's feedback point, causing it to under- or over-estimate the regulated voltage and momentarily shoot it to some undesirable value. This causes all the pixels' values to be wrong. The regulator returns to its normal setpoint milliseconds later, but some frames have already been captured by then.

As has been suggested, the camera may be susceptible because it's unshielded. Another possibility is that, since nearly all the signals in your camera are digital, they have noise immunity that prevented them from being impacted and it would take a stronger EMP to get bits to flip. Perhaps the camera has the only analog circuitry that was especially EMP-sensitive.

\$\endgroup\$
  • \$\begingroup\$ When you mention the wire disconnection, is that just inductive kickback, or is it some other effect specific to motors? \$\endgroup\$ – Andrey Akhmetov Apr 2 '15 at 10:06
  • \$\begingroup\$ @hexafraction, yes, I'm referring to inductive kickback. Normally inductive kickback is limited by the H-bridges to the battery voltage, but during a disconnect can fly to thousands of volts. \$\endgroup\$ – Zulu Apr 3 '15 at 1:09
  • \$\begingroup\$ The first thing to check would be if it worked after this. Assuming the wire came off. I am also curious what kind of phone you have. When I hear unshielded my first thought is that most smartphones are a really good faraday cage. Now if you mean that the photosensitive elements which are the lens may be causing this I might believe that an excess voltage through the rectification circuit could overload the circuit and cause an anomaly. Whether this is an overloaded signal to some ADC in the phone or just an analog filtering issue, I am curious. \$\endgroup\$ – mcmiln Apr 4 '15 at 7:43
  • \$\begingroup\$ @mcmiln: I think it's possible to have a transient disconnect, which occured during the crash and disappeared after it. The primary weak point here is the plug connector between the ESC and motor. These connectors nominally maintain really good contact, but the rapid acceleration of impact could change this. For the purposes of EMP generation, the disconnect only needs to last microseconds. \$\endgroup\$ – Zulu Apr 5 '15 at 20:08
-3
\$\begingroup\$

I think you had a power supply dropout or overload. When a spinning motor is suddenly stopped, as I think it would when it hit the ground, there is a sudden surge in current into the motor as the controller tried to get it back up to the desired speed (the controller does not know it hit the ground). That will make it increase the voltage sent into the motor, which increases the current. And since the motor is not spinning anymore, there is no reverse EMF being generated by the motor, so the current goes way up until the supply overloads. That probably causes the power supply to drop its output voltage, which resets most of the electronics. Curious that the video continued to record after the crash but maybe its controller is set to deal with momentary supply dropouts.

\$\endgroup\$
  • 2
    \$\begingroup\$ This answer doesn't make any sense. It's clear that the camera is entirely separate from the aircraft. \$\endgroup\$ – Dave Tweed Mar 29 '15 at 15:31
  • \$\begingroup\$ @Dave, that is correct. The camera is on a separate device, with its own power supply. \$\endgroup\$ – Andrey Akhmetov Mar 29 '15 at 16:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.