Meroe Labs

We are a hardware startup in Ibadan, Nigeria. For now, we are building a drone, read up.

[meroelabs-blog]

Hardware is Hard

And making mistakes is frustrating.

With a frame ready, we proceeded to assemble the electronics. It seemed our drone would be ready soonest. The next steps were connections and calibration.

Now, we decided to calibrate using Mission Planner. It’s a fully-featured, easy to use program made by ArduPilot, who also made the controller we are using. One of the good things about the program is that the whole UAV does not have to be assembled prior to calibration.

One of the steps in calibration is the compass (GPS). Now, there are two commonly used ArduPilots: 2.5 and 2.6. Now, they look very alike but are significantly different: 2.5 has an in-built compass, 2.6 does not and requires an external compass.

Due to this, their prices are significantly different, the 2.6 being significantly cheaper. We had purchased a 2.6. *yup, being the cheapskates we are*

On that day, with eager hopes, we watched Mission Planner inform us that it could not detect compass. One of us had travelled into town for this. f%*gis#!

So, we had to buy an external compass, and the Turnigy-6M came to our rescue. Another setback of about a month. It seems ArduPilot phased out the controller with in-built GPS not to make more money as we initially thought, but to be able to move the compass away from the power source and rotors, thereby avoiding magnetic interference. Lesson learnt.

[meroelabs-blog]

How to solve a problem like a Quad?

The design decided and finished, the next thing was to go into a workshop and fabricate. The problem, which workshop can do all what we want?

For the drone arms, we decided to use wood. Why? Aluminium is preferable, but hard to get and probably expensive. Also, wood can be carved into any shape. However, most carpenters do not use decimal points in carving (we tried), the others use inches. Where do you find one that understands decimal points and SI units? University of Ibadan.

Luckily, we all came here. And Chuma had made some trouble in university with a Physics lecturer who became his friend later, so he allowed us use the Physics Wood Workshop.

Sketch of the arms we wanted.

Drilling the arms to M3.

The finished arms, the blank holes are for the rotor shafts.

The 3D printed plates also came along, we added a third (modeled by Kunle), picture of the original two below.

To check fit:

Another big problem with hardware, is that there are simply no complete hardware stores in Nigeria. The few we have sell mainly Arduino kits and breadboards for the same, but this is because that is what is selling. No place to buy dean connectors or bullet connectors or tiny screws.

We initially designed the plates to use M2.5 screws, but there is no place to buy that - online or offline. We actually walked through Oke-Padre.

The traders only sell from M3 upwards.

We need a McMaster-Carr.

[meroelabs-blog]

x or +?

No need for BODMAS, it’s not that deep.

Quadcopter UAVs (Unmanned Aerial Vehicles) either have builds in the shape of the x or of the + sign. This is informed by the way the arms of the build intersect.

A + build is symmetric about the centre of the plate and all the arms intersect at 90 degrees.

An x build is also symmetric about the centre of the plate, but the arms intersect according to the shape of the plate.

The choice depends on the centre plate. The centre plate carries all the typical components of the quad - that’s quadcopter for short - except the ESCs, motors and propellers. Lolade and Olamide did a retrofit of all parts to guide our Design team, as shown below.

That’s a + build, right? But it’s not that simple.

Our battery happens to be quite long, much longer than the other parts, this eradicates the use of a + build, as it requires two plates of the same size.

We had two builds for reference: first from Make, the quad there used fiber glass plates and wooden booms. In fact, Kunle modelled a frame based on this build on his Instagram. The problem with that build was holding the battery to the base of the centre plates.

The second reference build was from the Open Pilot Blog. This has hollow aluminium arms and three centre plates. Now, from engineering mechanics, hollow bodies have a higher moment of inertia than their solid counterparts. In plain English, they are harder to bend/crack/break, which is better for a drone so that it doesn’t break during testing.

Chuma had the idea that we should adapt the DJI F450 frame (see below), and he would 3D-print some of the parts to reduce weight.

The frame gives space for a long battery, and we would be adding an extra centre plate. We found a model for this on GrabCAD, thank you Brain Verkaart!

[meroelabs-blog]

Purchasing

How do you get parts for a drone when your base is Ibadan?

Not from FoodCo or Oke Padre, at least.

Seyi had drawn up a BOM with online stores we could buy parts. We had to buy components that people had tried and tested. The components from the earlier post came together to around $189.33. We thought ‘Ah, 40k, nice. Not a bad figure’.

Turns out we forgot to multiply the price of the rotors by 4. Each was $42.99 on Aliexpress. Another $129 almost doubles the price we thought we could get our drone components. So Temi thought it had to be cheaper, after all it was a 3D Robotics part, Aliexpress was only reselling. And really, 3DR sells the rotor for $18 a pop. Score.

Meanwhile, Aliexpress gives free shipping to Nigeria. Your item only takes a few weeks to arrive. Who has a few weeks? We do, so far it saves money. To get the 3DR rotors, we had to look for a friend of a friend (FoaF) who was coming back. And still, our order was cancelled twice. The first time, FoaF was packing his bags to Nigeria before we found out it was cancelled. 3DR has a history of wahala, apparently. We had no idea.

We got our battery, propellers and RC controller from Hobbyking. Seamless process, but hey, they do not make their own drones like 3DR. We chose to order from Aliexpress first, as it would take a while, but in this time the ESCs we wanted at Hobbyking went out of stock. *insert appropriate four letter word*

We had to re-calculate our drone equations. Well, not we, Olamide and Lolade. Our replacement was a Chinese brand sold by our friends, Aliexpress. Good news was it weighs 21g and sold for $20.99 for a set of 4. Our initial ESCs sell for $31.96 for a set and weigh 25g each. Not bad. 

Temi remembered we will need a Power Distribution Board. Smart guy, his ancestors are probably proud of him. Blogs can not tell you everything. 

If you’re planning a drone build, where you get them from is important. Pricing is also important, but sometimes the cost is higher at Store A for a reason.

[meroelabs-blog]

In The Beginning...

We started reading stuff off Oscar Liang and scoutUAV, so many people build copters all over the world. But they usually have hardware stores they can buy parts. We don’t, we had to make sure the parts we got were the best.

That’s where Olamide and Lolade came in.

Temi had written up a project brief (it was quite long so brief should probably not be the name any more) outlining the components: Frame - holds the components - , Motors - makes propellers move -, Propellers - rotate like helicopters -, Inertial Measurement Unit (IMU) - continuously measures the altitude -, Battery, Micro Controller - brain of the quad -, Electronic Speed Controllers (ESC) - checks/controls the speed of the propellers -, RC Transmitter - receives information from the ground pilot. Chuma found out about ecalc.ch, the best calculator for building your own drone.

Going further we found that an Ardupilot is a controller that has it’s in-built IMU, and a large network of users. So no need for an IMU, saves money. Maybe.

First, we chose a rotor. The AC 380 from 3D Robotics seemed like a good choice as we wanted 10A sourcing current for the ESCs to work. It also comes with propeller adapters, so we can buy any kind of propellers. Coup.

Hardest to fix was the battery. As our motor runs on 11.1V, we needed a battery with a power rating for the time we need. But we needed to know what our frame will weigh.

So Kunle whipped out his Solidworks and modelled a frame. Using Workbench we were able to agree on a suitable build. At that time, Kunle was in Benue, serving his fatherland. Bukunola did material selection, working from Calabar. We got the weight of the frame, added some kg, approximated to 1kg. 

Using this to calculate, at 75% motor efficiency, we needed a battery with more than 3440mAh. We picked a Zippy FlightMAX for this, rated 5000mAh.

For the ESCs, we chose the Multistar from Turnigy rated at 20A. Our transmitter is going to be the Flysky/Turnigy/iMax operating at 2.4GHz, because it can select rates and choose your hover pitch. And it’s cheap. It’s also Chinese and has well documented issues.

Last, the propellers. We chose the SF-E, measuring 10x4.7. That just means it is 10 inches long and has a 4.7 inch pitch. 

We ran our selection on ecalc.ch:

Not bad, at least on paper and by equations. 

Now to buy the components, that’s another ball game.

[meroelabs-blog]

Team

First, I’ll introduce the three guys.

Temi is our Electronics Lead and philosopher. You will find him talking about ETH, drones or both. He likes to buy books on everything from Arduinos to Motivation. We don’t like to engage him on this, well one of us does.

Chuma is the Design Lead and hype man. He has an account of every social media platform you can think of, but is not very social. Apple fanboy, he likes to show off his 3D printing business.

Seyi is our Business Development Lead and money bag. He hates white collar jobs and loves digital marketing. Also drives like LASTMA is chasing him and he doesn’t even live in Lagos.

And they assembled this team.

For Design, we have Bukunola and Kunle. Now, you will never see any social media presence of Bukunola, but she never sleeps. Her trademark is 12am emails; she handles mechanical engineering research. Kunle is a SolidWorks and Keyshot guru, he has won challenges on GrabCAD. He likes to argue scientific philosophy with Temi. We don’t disturb them.

For Electronics, we have Olamide and Lolade, for Research, and Ope with Hamza, for Software. Olamide and Lolade (female, because we are gender friendly like that) are full-time geeks, and actually look alike. They draft the best reports also and hate replying mails. Ope spends his time looking for who next to beat on Hacker Rank and is a boss with HackiDemia. Hamza is quiet, and lays mad code at Lemz Delivery. 

For Business Development, we have Didun. Fine boy, Overkill Lord. He gives PowerPoint slides instead of reports; we are his pro-bono gift to mankind.

You’ll get to meet all these people really soon.

[meroelabs-blog]

Preamble

So three guys decide to build a drone. Why? Different reasons.

But still they recruit a bad ass team for this. 7 other people. Currently, they are in Lagos, Ibadan, Ilesha, Calabar. They set up a Trello board and divide themselves into teams. The motive: design the best drone for the cheapest to fly for 15 minutes.

This blog will document our process. But currently we have gone a long way, so I will bring you up to speed in the first set of posts.

Who are we? Meroe Labs, founded by three ex-UI students.

Happy viewing.