Comparing Development of Custom Android Hardware and Custom IoT Hardware

Hope everyone is figuring out how to make the most of living in pandemic times.  This month’s article returns to focus on Custom Android development, in particular how it compares with developing a Custom IoT product.  This article focuses on the devices, rather than the backend systems that interact with IoT and Android devices. The reason to focus on hardware is to keep the comparison parameters aligned since backend systems are another topic and can vary greatly from product to product.

Amount of development resources available

Generally the electronics in an Android device are much more complicated than in an IoT product.  For starters the low power IoT MCUs don’t support as much memory or peripheral electronics as an Android IC chip.  The same can be said about the firmware that drives Android and IoT devices.  Android firmware uses gigabytes of compiled code while the code of IoT devices could take less than 1 megabyte.  The firmware that drives an IoT device is what engineers learn at the early stage of their education.  Intelligence and experience affect the quality of an engineer’s work, but the basics are basic.  Creating a Custom Android device means making changes to a much more complicated and larger firmware.  Learning to navigate the Android source code requires more specialized education and doing it efficiently requires a lot of it.

For these reasons it’s a lot easier to find engineers who can make IoT products.  The community of engineers who get direct support from an Android IC manufacturer include high volume consumer brands and licensed engineering firms, often referred to as design houses. Getting the license costs tens to hundreds of thousands of dollars, may require special relationships, or commitment to purchasing a high volume of chips (which usually happens as a result of paying a lot of money or having a special relationship).

Cost of: Development. Engineering. Making samples. Testing.

Let’s say you want to make a Bluetooth remote control for turning on and off electronics in your room like a desk lamp, fan, etc… Here’s what you do.

Get an Arduino. Spend half a day watching videos or reading tutorials about setting up the hardware and programming. Get wires and other accessories that connect to Arduino. Depending where you are in the world obtaining the hardware could take between a day to a few weeks. Put it together.  Write code for the Arduino (or copy from the internet).  Start testing. Have trouble coding?  Find a nerdy 12 year old.  Buy them ice cream. You’ve got code; kid’s got compensation. With one Arduino and a pint of ambition an IoT product can be prototyped for less than $50. An Arduino prototype provides a possibly unique proof of concept, a starting point to learn from and build on.

After you’ve iterated the prototype a few times, fall in love with your product, and decide to move forward with a more robust and custom design there are many engineers (or engineering students) who can turn your Arduino based contraption into something that’s ready for its own PCB. Not all engineers are equal. If you find a good one then design files will be ready in a week or two. Use the internet to identify a prototype maker.  2-3 weeks later you have 20 pieces for the next level of testing. Total costs less than $1000. More complicated products will cost more.

Sounds good? That’s because it is, at least compared to what goes into making a Custom Android tablet or Custom Android smartphone. Companies that employ Android hardware and software engineers have much higher expenses and will charge more for their engineering services. If you find one that does ONLY development services, rather than one like Hatch that does only development AND manufacturing (or just manufacturing), their engineering costs will leave you unable to pay for that 12 year old’s ice cream (for well funded companies this isn’t an issue). The PCB used in an Android product is much higher cost and PCB manufacturers capable of making those boards require more money for doing a small production run. This is also true for doing SMT of an Android product and almost every other component that goes into an Android product when compared with a simple Custom IoT product. The companies making screens, touch panels, cameras, etc. that go in Android products are used to making their money on high volume. A request for making 20 pieces of a custom design without a deposit on a larger order doesn’t economically fit their business model, unless the amount you are willing to pay for those services can convince them otherwise.

Size of project

Because getting to the point of producing a small batch of an IoT product requires less capital, smaller production runs are more feasible as the up front costs don’t need to be amortized over so many units. This is one of the reasons that so many IoT startups have sprung up over the past 5-10 years. Everyone with an idea can find a rich old lady to pitch their connected lipstick device to.  The tech is great, but efficiently executed development and accurately identifying market demand presents challenges. The value is in finding the right project manager who knows how to manage it properly, making sure Ms. Daisy’s investment isn’t spent unwisely.  As low power and low cost IoT devices have opened doors to many new products this has resulted in a plethora of hobbyists-with-hustle confusing a cool idea with a viable business (few of which have worked, but good on them for trying). In most cases they should focus on building gimmicky party favors for friends instead of product companies.

Due to the amount of capital required to go from start to finish with a Custom Android project a company must be sure that the product can deliver a defendable and unique value to a high volume of users.  The reality is that Android hardware offers incredible value compared to what people used before Android was a ‘thing’.  Before Android companies would make devices based on WinCE.  Remember that?  It was used in POS systems, medical devices, industrial hardware, and plenty of digital paper weights.  Compared with Android, that historical Microsoft product required more processing power, had less hardware options available, wasn’t open source, and came with a per unit licensing cost.  To catch up Microsoft has reduced or removed the licensing fee on small screen devices, but Android continues to define itself as the go to platform in mobile devices.  Even though Android lowered the barrier to entry the cash resources required to complete a Custom Android product means that they’re usually much bigger projects than with an IoT product.

Order size

In line with the above logic, customers wanting a small production of IoT products have plenty of resources to get that done. Doing a small production run for market feedback requires minimal investment and minimal commitment. Customers can find many prototype companies online to help with getting PCBs made, sourcing, doing SMT, and testing the boards. In most cases new IoT products are custom designs which means the buyer needs to get their own electronics designed. Since there are many engineers capable of doing the design the cost of design services is also not a major investment.  Producing a batch of prototypes is relatively low cost.  Final products also require a custom case. Making a mold costs between thousands to tens of thousands of dollars, depending on the size and material of the case. For prototype purposes using an existing case, while possibly not a perfect fit, avoids having to pay the high tooling fee.

For Custom Android development designing the electronics and customizing the firmware has a higher upfront cost.  Finding all the right components takes time and expertise.  Integrating and testing the electronics also takes more time. The people with the right skills for bringing a Custom Android project to life are expensive and the projects take longer than IoT development. As mentioned above companies don’t take on small Android projects without charging a lot for development or receiving a larger volume order commitment.  Amortizing the initial costs in a small volume order renders the unit costs too high for most customers. The mold cost for a Custom Android tablet or Smartphone is also higher than for a small IoT device. To bring unit cost to a reasonable level Hatch usually looks for a commitment of 10k pcs in order to engage a new project (although for some projects lower volumes are acceptable).

Real Life Examples

Simple IoT Product

During the pandemic period some of Hatch’s clients have had to delay orders or slow down development giving us time to work on a few side projects. One of the side projects comes from a startup that’s making a simple IoT device that involves Bluetooth, lights, sensors, and complicated mechanical design. Due to the confidential nature of the project I will not go into more technical details, but recapping the prototype development process will highlight the points mentioned above.

This product requires a special kind of material to work properly. To start with we ordered samples of different materials to find the right one. After deciding on the right material we got an Arduino and the accessories needed to test this product. My friend, who’s not a programmer, figured how to program the basic functionality we wanted to achieve in order to create a rough prototype for early testing. From getting the first sample material, the ‘starting’ point, to having something we could test took about 7 days.  After destructive testing we made changes to the construction. Another round of testing. Another round of improvements, and so on.

Our goal is to have a mass production prototype of the product ready 2 months from the starting point. This includes designing the electronics and firmware using a Bluetooth MCU instead of the Arduino board, customizing the design of the product, and finalizing materials and construction. We won’t do a custom enclosure for the initial prototype, allowing for quicker iterations to continue design optimization while giving us more time to refine the eventual design.

The goal is to have a trial production of 50 units done 3 months after the starting point which will ship to selected users for feedback. Based on a generally positive reaction to the 50 trial production units we hope to ship the first mass production about 5.5 months after the starting point. This goal won’t be easy, but the chance exists.

Custom Android Manufacturing

Typically a Custom Android hardware project requires 8-12 months to go from start to mass production, sometimes longer. A Custom Android Smartphone project we’re working on now started with sending multiple sample devices (products that are already being produced) to the client for them to choose specifically which screen, CPU, camera, etc. to use on their own device. This process took 3-4 weeks.

Once components were decided it was time to do mechanical engineering and PCBA design. Unlike with an IoT product a Custom Android device needs to have a custom case made in order to produce prototypes. The Android product’s PCB design needs to match the case design and the external components must be properly secured in place. There’s no generic casing to serve this purpose.  Mechanical engineering and testing 3D printed casing samples takes 1-2 months. Designing the PCBA and making samples takes 2-3 months.

These early stage tasks are just the beginning, but hopefully it’s already clear why making Custom Android hardware takes more time and cost than an IoT product. For an in depth review of the process check out this article outlining the steps to getting a Custom Android Device made.

Continuing with Android and Starting on IoT Development

Hatch has been working with Android hardware in Shenzhen since 2011. We were one of the first companies to work with Android tablets designed in China and are one of the more tenured companies still focusing on this space. The use cases and presence of Android devices in daily life have grown as much as we imagined they would. That growth is going to continue and Hatch will be there to drive it forward.

As we’ve already worked on several IoT projects (for internal purposes and for clients) our understanding of the development and manufacturing for these kinds of products has gotten to the point where we’re ready to start offering IoT development and manufacturing services to our clients. Unlike Android projects we’re (at least initially) open to work on projects without a mass production commitment so feel free to reach out if you’d like help with your IoT project. More to come about this in future articles.