Ebee Drones are expensive, and there is no doubt they are well made. If I want to build my own drone it is worthwhile to see how the old Ebee was built. What components did they use? Is there anything that can be salvaged for my new drone?
A tenant of manufacturing, engineering, and life in general, is never do more work then you need to. For this reason I have found that most of the time, engineers are not focused on reinventing the wheel. They are taking off the shelf components and assembling them in new ways or modifying them to meet the needs of their projects. This drastically limits time consuming design work, testing, and the manufacturing resources needed to put together something. It doesn’t matter if it’s a $12,000 drone, or a $100 camera. Engineers build on the back of giants.
When looking at the individual pieces three components drew my immediate attention. The servos, the motor, and the Electronic Speed Control (ESC).
The Servo’s
I had never come across a servo with a machined metal body before. I know that the servos are replaced every 100 hours of flight time, and these servos had just been replaced. So if we could use these in our new build, that would be pretty awesome. The connectors looked to be the same as any servo I was used to seeing.
The ESC & Motor
The ESC was painted in a soft blue masking paint. Nothing is done for free, there is a reason someone painted over both the front and back of the ESC. It may be an insulative covering, or it may be a mask to help hide what the original components were. Since other components on the drone were masked over, I am inclined to think SenseFly is deliberately trying to force you to buy the components from them rather the risking the consumer put off the shelf components in their products.
ESC’s are a dime a dozen, but I was curious about this ESC, kind of wondering if I’d see some well known brand silk screened on the board.. I scrapped off the paint on a few of the chips to see if the numbers were legible. They were, an I was surprised to see the Atmel Mega 168PA. These little 8-bit microcontrollers are used in some Arduino knock off boards.
There wasn’t anything obvious that pointed to the original manufacturer. Since the ESC and the motor will be sized appropriately to our new build. I am not as concerned about trying to reuse this one. It might turn up in a future project, as it appears to be and off the shelf motor and ESC combo, and probably will work with any other radio controller.
The Motor was barren of any markings, even when the SenseFly label was removed. Like ESCs, outrunner motors are common. It is likely a high quality motor, given the expense of the servo’s and the drone but it is hard to say which one they used exactly.
The Distance Sensor
I believe the Ebee uses an altimeter and an optical distance sensor for measuring distance from the ground. I don’t know if it uses GPS data, but it wouldn’t surprise me. The distance sensor is used when it is close to the ground, I believe that is under 100 feet. In terrains where is isn’t perfectly flat, the Ebee will dive for the ground and then level out, looking for that distance to ground reading. In my experience this has been less than ideal as sometimes it is a bit too aggressive in it’s dive and it can bleed too much altitude and it doesn’t have time to level out before landing. I.E. it crashes.
Still, something like a distance sensor might be useful in our replacement drone. So what does the Ebee use? As with the ESC and the Motor there were no external “Makers Mark”. However I suspected the housing was likely sourced and the chip inside might have some clues. The package was glued together so I had to cut a flap in the rubber backing.
Figuring this sensor out has been a real trick. The A9500 made by Avago industries (The stylized A matches) is a an optical mouse sensor. I have the datasheet for it right here. However it is not used in that manner at all, according to Ebee, it senses the distance from the drone to the UAV, which makes it possible to land “Softly” in terrain. I put out a call to a Broadcam, which bought out Avago, only to find that model had been sold to Pix Art Industries. The conversation with the tech didn’t initially provide a lot of help.
To him it was a mouse sensor, but then I looked at it again, the chip might have been initially used for a mouse, but the lense is completely different. I suspect, and I might be wrong, but I suspect this sensor is indeed a mouse sensor. However by removing the original lens and applying some clever engineering on someone’s part, someone took a mouse sensor and repurposed it for a distance sensor.
Based on the data sheet I think it would be difficult for me to repurpose this part. I would be better off using a small LiDar for the build, if I want to have the same capability.
Telemetry (Radio Transmitter)
Given that the information is printed right on the modem this was pretty easy to track down. Including the operating manual for the receiver.
Looking at it, Sensefly more than likely had a custom board made that mounted the radio transmitter, and a package made for the radio receiver. There may be an opportunity to replace the radio transmitter and still make use of the existing receiver. We will have to dig into that further.
The AutoPilot
I believe this is the autopilot. It too was smashed up pretty good with one whole connecter sheared off the side.
The altimeter is readily available from DigiKey and can be found here. Airspeed sensors are pressure differential sensors, and the air speed sensor can also be found on DigiKey. At the very least form factor for this board is very likely custom, the biggest clue of that is that it has a SenseFly silk screen. There likely is a Developers Board that this small board is based off of.
Developer boards will often have lots of different components and capabilities which are either not needed or are under utilized in the final product. With anything airborne weight is a concern, and thus it is likely the Sensefly started with a developer board and then commissioned a board to be made that deleted the features that were not required.
In order compare any autopilot with the Ebee autopilot it is useful to look at the Microprocessor.
The PixHawk autopilot we plan on using for the project features a ARM STM32F427. These two MC’s are fairly comparable. Why SenseFly went with a the 407 instead of the 427, is likely to due with the 407 being adequate, the didn’t need the extra features or the extra power that the 427 offers. Given there are plenty of development boards for both MC’s it would be impossible to know which one was used. Other then we can say with a degree of certainty that an Autopilot with the ARM STM32F427 will do everything that the Sensefly autopilot will do.
The GPS
This is the last major component for the drone. There are small components such as the sensor for the wings being attached correctly, the Pitot tube, and the LED’s for status indicator. However those are pretty easy to replicate or replace. The GPS sensor is the last major component that will make or break the project.
As with the radio modem this has the branding all over it. So it was pretty easy to look up the specs.
According to the manufacturer this is the latest and greatest GPS Module.
So the GPS sensor really didn’t present much a challenge for us to look up. It’s a part of the Ebee so we’d be a miss if we didn’t include it here.
My Impressions
The Ebee is a quality product there is no doubt there. Though I find it frustrating that they are charging $7,800 for all the components and $12,000 for a replacement drone. There is nothing that I can point to and say “That is revolutionary”. With exception of the distance sensor, which I’m still scratching my head on. (I am certain, I just haven’t figured it out, and it’s not something that’s unique to senseFly)
Given that the hardware I can purchase for less then $1000 for equivalent or better parts, I have to look at the software for costs that are not easily compared. It could very well be that I am paying for the software license that integrates all of these components together into a UAV. Is this software better then what is on the market? Honestly I don’t know, the other option I am aware of it the Ardue Pilot. Which is free and completely open source. So if Ardupilot does everything that the Sensefly eMotion software does, then I find it very difficult to justify the differences in price. (Yes you download eMotion free, but that is with the Purchase of the drone. So the costs are built into the drone price.)
This is all to say that I am confident there is no “Magical Part” or “Voodoo” that is going on that cannot be replicated by a home built drone. There will be no fancy hard sided case (Yet) or it may not be as easy to transport, but it will very likely produce the same results with less cost to operate and maintain. Going into this project knowing that is the case helps build confidence that I’m not getting in over my head.