Thursday, August 4, 2016

Next steps

To get Airball into the hands of real pilots, we need to build something operationally useable. Sort of a minimum viable product. We can continue messing around with smart phones and slurping data from whatever Dynon units we can, but that's not widely installable or applicable.

Our job at this point is to build a probe and a display unit, which connect to each other wirelessly. This should allow installation in any aircraft -- certified or Experimental -- just like clipping a GoPro camera to a strut.

Probe design

As described in our paper, the probe has a large radius nose so that we don't need to mess around with calibration. Taking an idea from the 2nd-place winner of the EAA competition, it can be powered by a little windmill, so it's self-contained and never needs battery charging or wires.

The sensors we plan to use are Honeywell TruStability, the better versions of which boast accuracies of 1% of full scale. These provide an I2C interface.

The probe will emit data as 7-bit ASCII "sentences" separated by newlines, consisting of human readable pressure measurements in engineering units. This makes the protocol debuggable yet reasonably compact. It would look something like:


where the XX is a checksum and the newline ends each "sentence." This is very similar to NMEA sentences emitted by GPS receivers.

We may consider adding a measurement of barometric pressure, and a temperature probe, allowing us to infer true airspeed. We have an idea for adding TAS to the Airball display for even better situational awareness.

For the radio link to the display, we expect we'll probably use a point-to-point XBee. This will ensure that we do not interfere with Wi-Fi or Bluetooth devices used by the pilot elsewhere. It should also remove the need to "pair" radios -- each probe and display will have a hexadecimal code baked in (e.g. via coded rotary switches), so the link is basically a simple "wireless wire."

The sensor readings will be scanned by an Arduino, converted to engineering units, and then pumped onto the serial link. That's all the processing in the probe.

Our hope is that the innards of the probe, and the radio link, will be simple enough that, as soon as the windmill starts turning, the probe will produce data, which will be immediately visible on the receiving end. So it would be almost like airspeed "comes alive" on the runway, within a matter of milliseconds, without waiting for devices to pair with each other or DHCP handshakes over Wi-Fi or any of that.

Display design

The display will be an embedded Android board (probably a BeagleBone Black) running a stock Android that's been tricked up to launch our app on startup. It will contain a paired XBee radio to get data from the probe.

It will have a high brightness (at least 1000 nit) LCD panel. The high brightness display is really the main reason why we're choosing not to go with a regular smart phone.

We plan to add a single pushbutton and rotary encoder knob for all functions including modifying the display parameters (limit AoA values, the position of the annotations for various "interesting" values of AoA, scaling of the ball to airspeed, etc.):
  • Long-press (e.g. 5 seconds) to get into Setup mode. This brings up the currently editable parameter on the screen.
  • Rotate to modify the value of the parameter.
  • Short-press to cycle to the next parameter.
  • Long-press to exit Setup mode. If no activity for a while (say 30 seconds), Setup mode ends automatically.
  • Outside Setup mode, the knob acts as a brightness control.
Power can be provided by a simple 12V "cigarette lighter" cable.

No comments:

Post a Comment