Custom Android Hardware- How to Approach the Unknown

Developing new products takes creativity.  Creativity combined with experience makes that creativity more productive.  The more experienced the product creator, the more efficiently their product is made.  People who are trying something for the first time, whether making a custom Android device or learning to ride a bike, go through a process of learning what works and what doesn’t.  In the world of custom Android hardware development we are routinely faced with new requirements, since each project is different.  The experience we gain from past projects gets applied to new projects, but the requirements are rarely the same.  Just like riding a bicycle isn’t the same as riding a unicycle, but the bicycle experience helps with learning how to ride the unicycle.

When starting a new project we usually have some learning to do.  Experience from previous projects sometimes provides insight into new challenges, but rarely are custom requirements the same.  More importantly than having specific product development knowledge is knowing how to find the best solutions.  Take for example a project where the client requests a very bright screen that remains power on for extended periods of time.  We know from other projects that bright screens require more electricity.  This means electric component changes are necessary to the main PCBA in order to supply more electricity to the screen.  We also know that 1) a bright screen, and 2) running for an extended period of time increases heat generation.  Planning for this early is the value of experience.

Based on the things we know about screen brightness and extended run time, we make the initial design to the best of our ability.  At the same time we fully expect surprises, that we didn’t account for, to come to light through initial prototyping.  Expectating to learn new things is one of the key lessons that isn’t specific to a product, but to a process.  No one will ever foresee all the problems of a new product, but having a process in place to catch those problems is often the fastest way to find a solution.

With the intention of finding problems, we want to set up early prototyping in a flexible manner, allowing for quick changes.  For example remove the components that control electricity flow to the screen from the Android tablet PCBA.  Instead, use a small add-on board, that’s low cost and easy to customize, in order to test different specifications.  We’d also want to hold off on finalizing case design, knowing that we’ll need to test different screens and back lights before finding the right one, and each screen has slightly different dimensions.

The key idea is to keep initial prototyping as flexible as possible.  Test as much as possible.  As testing reveals what works (and what doesn’t work) lock in specifications.  This same process may happen in parallel for several parts of a product.  Once all the unknowns start to become clear it’s time to make a more formal prototype, one in which the specifications are not flexible, and could feasibly get mass produced.  Testing still goes on at this stage to check for problems that could emerge from the integration of all these unknowns.

Many custom Android products, and new electronics in general, go through a long evolution from concept to reliable mass market product.  Often the first mass production results in more of a functioning concept than a final product, and it could take a few rounds of mass production to address all the surprises that arise.  This doesn’t mean a first generation product is bad, just that the next generations are better.  Like in any artistic process there’s never a real end point, but a point at which the creators agree a product is ready for release.

Knowing that product development is constantly evolving, staying as flexible as possible for as long as possible is the ideal way to achieve the best result in the shortest amount of time.

Smart Development: Testing your Custom Android Device Early

There’s a saying about testing products that applies to all mass manufacturing, but maybe even more true for electronics and custom Android hardware in particular.  The saying goes like this: a problem with a product costs 10 times more to fix at each subsequent step of the development or manufacturing process.

This adage holds true during the earlier stages of development, and the number is probably much higher as the product gets closer to the customer.  In the beginning of a custom Android development we work with low quantity prototypes.  Prototype PCBs, prototype casing, and single digit volume of peripherals such as cameras, screens, etc.  Even packaging for individual units and shipping cartons go through a sample review process.

It costs money and time to get a round of prototypes done.  Low to mid thousands of US dollars and 3-5 weeks are standard.  Problems that are discovered at this point don’t necessarily require a new round of prototypes, although they might.  Most hardware problems identified at this point are solved by making manual adjustments to the electronics or casing.  For example resolving issues with the electronics may be done by replacing components or adding small PCBAs, depending on the nature of the problem.  A case that doesn’t fit well can be cut by hand or a specific part of the case can be 3D printed with an updated design.  These kinds of changes are relatively quick and low cost.

Electronics often go through multiple rounds of manual changes until reaching a satisfactory resolution.  Making changes manually allows us to check different solutions in a quick and low cost manner.  Once a solution is found, the engineers implement it into the design.  This way it’s not necessary to go through new rounds of prototypes to test engineering changes until there’s reliability in the result.

After the prototype stage comes trial production.  Trial production is a small pre-mass production run of 50-500 units.  Quantity depends on many factors, mainly how many units the customer wants to test in the wild and unit cost.  Hopefully only software related problems (that can be updated with an OTA solution) appear at this stage.  Keeping the trial production quantity low reduces the cost impact of a hardware related problem.  It generally takes 4-7 weeks to prepare and assemble a trial production; waiting for parts to arrive represents about 4 weeks of that.  The devices from trial production go to labs for certification testing, undergo age and stress testing.  Others go to real world users, as they may come across problems that weren’t discovered by people who designed the product.  Sometimes problems come to light at this stage; why there’s a trial production stage to begin with.  The time cost and financial cost of a trial production run outweighs those costs at the prototype stage by at least 10 times.  Finding as many problems as possible is the goal.  A manufacturer or brand only benefits by raising issues at this point.  Finding problems early is a question of digging for gold, not searching for fault, and should be applauded.  The cost savings of finding problems now amount to gold compared with finding problems at the next step.

After trial production the stakes get even higher.  At this point brands usually have plans for distributing the goods to their customers. The amount of money tied up in a mass production quantity exceeds all previous stages.  The amount of time invested to produce the products also exceeds previous stages.  There should be no design problems since products have already gone through a design stage, prototype stage, trial production, and multiple revision stages.  On the initial production run small problems are likely.  Products always get better over time, but the ‘problems’ at this point should be more like optimizations rather than issues that impede proper usability.  If something major goes wrong at this stage the brand, manufacturer, or both look bad.  Costs include the time and money to go through mass production, and reputation of the brand.  Still it’s better to catch a problem now than at the final stage.

The final stage is when the overseas brand or their customer receives the product.  Now there’s no turning back.  Shipping products back to China incurs a high (and for better or worse sometimes arbitrary) import tax.  Products considered ‘used’ aren’t even allowed back usually.  That means any problems which appear once the goods have landed need attention in the destination country, where assembly know-how and infrastructure are usually limited.  In many situations it costs more to fix the goods than buying them again.  The cost of reputation repair has no limit.

I remember on the first order I had for a retail customer, they made a late request which was implemented unsuccessfully.  Back in 2005 this customer ordered CD players.  They wanted packaging to hang on a peg.  Then they requested packaging to also stand on a shelf.  I had just visited the factory and checked the finished products (for the first time ever).  I checked a sample and the packaging stood up.  I approved it.  Goods arrived and the customer set up the products in their hundreds of locations.  A week or two later they said the packaging didn’t stand up sturdily enough and kept getting knocked down.  I offered to design a retail display to hold up the packaging.  They decided to return the full 2500 pc order.  I had little choice in the matter and took back the products.  Lesson, have your customers check the products as early as possible.  It’s worth delaying a shipment to avoid a recall.

Live and learn.  Hope this lesson helps you at some point. 🙂

Using Hardware Components to Remotely Monitor Product Quality

Whenever possible I try to write about actual topics that come up during the course of Hatch’s normal business activities.  Editorials are great (for whatever my opinion is worth), but real-life experiences, especially innovations, share useful and tangible value.  This month, together with our client, Hatch came up with an innovative solution for remotely detecting faulty hardware that, until now, went unnoticed without the client making an on-site inspection.  The specific case we’ll summarize involves the screen of a product.  The concept and solution we came up with applies to other components as well.

The product is a custom Android Wi-Fi tablet that’s used as an advertising display.  Multiple displays get installed in hundreds of locations.  Following installation, the company manages them from a backend portal.  Employees who work at the locations have no relationship with the display provider.  They also have no motivation or easy way to report problems.  After a display runs non-stop day after day for several years there have been cases where the screens stop functioning properly.  Hatch was tasked with figuring out potential causes of the problem without getting back any of the defective samples to analyze.

We identified three potential sources for the problem:

  1. Failure of screen’s backlight.
  2. Screen burnt out.
  3. Damage to regions of the screen.

If the screen’s backlight stops working or the screen burns out, the voltage going to those components changes.  With that in mind, we conceptualized a strategy for ICs in the device to read anomalies in the electrical current to detect problems.  Creative electrical engineering combined with related modifications to the device firmware now triggers a notification when the backlight or screen fails in entirety.  Regional screen problems may or may not affect the electrical current.  Most likely not, but detecting the other issues still improves the situation.

Due to the nature of an advertising display screen performance is critical.  With an automatic self-assessment solution in place, Hatch’s customers can get information about certain screen problems in real-time.

Hatch has been working with this customer for over 5 years.  During that time we’ve implemented several electronics and firmware iterations to achieve ongoing product improvement.  Custom Android products, unlike commodity electronics, often get used in commercial or industrial applications.  While the core hardware requirements don’t change as often as retail products, opportunities for optimization appear over time.  Working with a long term development and manufacturing partner, Hatch’s clients leverage our intimate knowledge of their product to implement continuous improvements.  This results in performance benefits that add intangible value to our clients’ businesses.

Entering into a custom Android hardware partnership extends deeper than just supplying a product.  A dedicated and capable supplier grows with their client, not just by providing hardware, but by empowering the client to become more successful as a result of Hatch’s dedication to them.

How to Decide Whether Price is a Good Value

How to value custom Android hardware

I like having an expectation of price when buying something.  You probably do, too.  An expectation makes judging value easier.  Human nature has something to do with this.  It’s clear if some products are priced fairly or not.  For example commodities like gasoline or bananas.  Quality of common grades of gas or breeds of bananas don’t vary greatly between different sellers.  Therefore pricing shouldn’t vary greatly either.  Many products we routinely buy, whether for personal or business use, fall into this category for the most part.

The closer a product gets to the customer the more potential for adding value.  This seems to apply universally, from custom Android projects to dinner.  On one end of the food spectrum is something simple, cheap, and low quality, but edible, like instant noodles.  On the other end is a culinary experience of globally sourced ingredients, prepared to the exact specification of a culinary artist, and served in a majestic environment.  Both fight hunger, albeit are priced very differently.

What justifies the price difference beyond the cost of ingredients?  In the example given above it’s the ambiance, presentation of the food, and taste.  The value customers feel from a high end restaurant experience far exceeds the cost of the actual food.  Buying products follows the same logic.  A mass produced product that is similar to many other products has little added value.  For example, a pair of generic headphones usually costs only slightly more than the seller paid for it themselves.  Customers pay more for branded products, even those that are very similar to an unbranded product.  The higher price comes from the reliability, quality, and sometimes prestige the brand represents.

Not all customers value things the same way.  For some the brand matters, for others they just want the commodity.  The value a custom product manufacturer brings differs from the value a commodity commodity device seller provides.  The custom device manufacturer needs to spend more resources developing the custom device even though the volume is lower for a custom device than a mass market device.  Custom Android device buyers often have niche use cases that deliver a higher value to the end user as well.  More work goes into making a custom product.  That means the product must deliver a higher value to the end user than a commodity product; the same way a dining experience delivers a higher value than instant noodles.

A custom Android tablet.

Most of Hatch’s customers use their custom Android products for business applications.  Their products offer a high value to these businesses and require a manufacturing partner that offers an equally high level of service and knowledge.  If a commodity product breaks the customer simply exchanges it at the retailer.  If a business product breaks that means a company is losing money by not having it work.  Also fixing the problem costs more than exchanging a product at a retailer.  It usually requires a technician to deal with the problem.   Business buyers should look at commodity costs as a clear point of reference.  The art of determining a fair price for the added value comes from quantifying a supplier’s capabilities on one side; and the cost of something going wrong with the project on the other side.

“No one got fired for hiring IBM” is a saying which means that IBM has such a good reputation that an employee would never get fired for hiring them, even if a problem arises.  Many highly skilled companies are small and don’t have the reputation of large competitors.  When evaluating a partner for a custom project, it’s necessary to understand exactly what additional value the partner will bring.  To do this, find out what kind of projects the partner has worked on before.  Ask for referrals from past clients.  Take into consideration how quickly and precisely the partner communicates.  If they offer sensible and clear answers from the beginning that’s a good sign.  If not then don’t expect them to start after paying the deposit.

Up front due diligence requires digging into the details of how businesses operate.  Spending time to get comfortable with a potential partner increases the chances of success for your project.

Tips for a Speedy Custom Android Development

Making a custom product takes time.  Usually, more time than our customers realize as there are never less details, and often more, during a development process.  But through experience we learn optimization techniques that have a big impact.  Below are some concepts you can apply for an efficient, yet reliable development process.

The App

When building a custom Android device, the key to quick development is identifying which parts of the product exist already.  That way you start testing your app on existing devices at an early stage, rather than waiting for your custom hardware.  Start by finding a sample of an existing product which uses the same CPU and Android version as your custom product, these are the foundation of an Android product.  While the existing product and your custom product won’t be exactly the same, the existing product can be used to start testing the basic functionality of your app.  Starting this process early saves time and generally results in a better app.

Special Hardware

To continue with efficient development, it’s necessary to know which components matter and aren’t standard.  Consider the purpose and focus of the product.  This determines what other components are important.  For example, if taking photos or video using a wide-angle camera lens is part of the device requirements, try to get samples of different cameras to test with the chosen CPU to make sure the camera quality is good enough.  We’ve seen the same camera provide different results with different CPUs (better results happened on CPUs with integrated ISP).  Ultimately this motivated a change in the device CPU because the camera quality was such an important feature of the product.

During this stage the hardware doesn’t need any case as the point is quick compatibility testing of the hardware.

Casing

Basic concept is to start prototyping as early as possible.  It’s possible to prototype even before the electronics are ready.  Hatch’s customers usually design the exterior of their product case with an industrial designer (Hatch does the inside, mechanical engineering, part).  Previously we had a customer approve a prototype and then a few months later change their mind regarding part of the external design.  The mass production tooling was already done at that point, so the change ended up costing additional time and money.

Even if the custom Android electronics aren’t done, if case design is very important to a customer, it’s possible to quickly get a simple PCB made with just case fittings.  For example, a blank PCB with just a USB, earphone, and charging port, rather than all the electronics components, so the customer can test from purely a mechanical and design standpoint.  If that prototype works the same one can be used once the full electronics are done.

Tips for Efficient Custom Development of Android Hardware

When conceptualizing a customized Android device, the 3 main areas to focus on are casing (mechanics), electronics, and firmware. Other aspects may include accessories and packaging (depending on business model), to name a few, but this article focuses specifically on the Android device itself.  Key factors within each area affect the development cost and time and unit cost. Understanding what impact different factors have enables a more efficient and beautiful development process. Read on for a summary of key details that matter in custom Android development.

Casing

Casing is unique as it incorporates two different disciplines of design; external and internal design.  Usually called Industrial Design and Mechanical Engineering, respectively.  If good looking external design isn’t a key aspect for your custom Android hardware then using just a mechanical engineer for both should be fine.  Many mechanical engineers mistake themselves for industrial designers anyway.  Usually external design means less for products that users don’t see or are purely functional in nature.  For the sake of this article we’ll look at the best way to approach a product where design matters.

External design, unlike all the other areas covered here, is exclusively based on perception and emotion.  Intricate details affect branding, perception, and usability.  Hatch highly encourages our customers to work directly with an industrial designer (rather than through Hatch or any other final product supplier) on the case’s external design for the highest quality results.  Hatch will provide the dimensions and other relevant information to support the designer.  To allow a smooth transition to the mechanical designer, the industrial designer should define gap tolerances between the parts of casing, exact colors, and materials.  Usually industrial designers do a good job on the latter two, but making the case fit perfectly can take a lot of time.  Having this requirement from the beginning gives a clear goal, allowing a supplier to test internally before sending to the client.

It’s best to use a mechanical engineer with product specific experience to create the mechanical design.  That’s why Hatch takes control of mechanical design once external design is finished.  Hatch’s mechanical designers have made and modified hundreds of Android devices like tablets, phones, and custom hardware.   Without start to finish product specific experience even a very qualified mechanical engineer won’t foresee product specific design nuances.  Those nuances are related to finished product quality and, through support of supply chain partners, component selection.

Only through seeing multiple products go from start to finish do hidden design optimizations become clear; this is the definition of experience.  For example structures within the mechanical design to strengthen ports or block dust from entering the screen area.  Sometimes ‘the perfect’ design requires an expensive or difficult to source component.  That must be avoided (and there’s no such thing as perfect design).  Collaboration between Hatch’s experienced engineering team and a client’s mechanical designer combines the best of both worlds.  This has a higher cost for the client and maybe overkill, but certainly covers all bases.

Simple takeaways:

  1. Make sure your industrial designer specifies gap tolerances between parts of casing.
  2. Use a mechanical engineer with Android device experience for expertise with quality manufacturing details.
  3. Have the mechanical engineer work with a supply chain partner (like Hatch) to verify that the components suggested by the industrial designer are easy enough to obtain now and into the foreseeable future.

Electronics

All projects have specific performance requirements that call for different kinds of electronics.  Potential for remorse comes when choosing commodity components like a screen or battery, designing around that, and then finding out that the screen isn’t crisp enough or the battery doesn’t last long enough.  When customizing an Android device choose components individually, by testing them in other devices first.  If there’s a screen you’re considering then get a sample of another device that uses the same screen to see if you like it.  Once you decide on all the key components then make an initial prototype to check that they work well together.

As much as possible stay within the framework of the CPU manufacturer’s reference design.  Clients have asked for a certain Bluetooth chip that’s not included in the reference design.  A change like this requires changing the tried and true reference design.  This change opens the door for new bugs.  It also takes a lot of engineering to make this modification.  Changing the reference design usually comes with a high development cost and higher mass production risk.

Simple takeaways:

  1. Sample key components individually in existing products before deciding on the final list.
  2. Test a prototype that integrates those chosen components before making mold.
  3. Stay within reference design; changing it opens the door to unpredictable problems.

Firmware

A combination of knowing the result you want from customized Android firmware and clarifying what customizations are possible at the onset of a project sets the stage for smooth development.  The key here is to ask specific questions to the right people so you get reliable answers.  Many Chinese suppliers hire inexperienced sales people based on their young age and English ability.  Young means they’re cheaper and English means they can serve as a conduit of information to the customer.  Oftentimes the answers these people give customers are incomplete or just wrong.

Hatch uses direct connections with IC companies and their licensed engineering companies to directly access management level engineers.  While the engineers don’t usually speak English that’s not a problem and we’re usually able to get real answers about implementing complicated customizations.  Like proper directions before a new journey, knowing whether the customizations you want are feasible avoids going in the wrong direction.

Simple takeaways:

  1. Clear definition of what’s needed before starting the project.
  2. Find the right person to confirm feasibility of customizations saves losing time going in the wrong direction.

Any project that tackles the unknown is prone to risk and surprise.  Applying the takeaways, learned through first hand experience, listed above when approaching the project will maximize reliability while intelligently managing risk.

Developing an Android Phone with Global Coverage

In the early days of mobile phones, a single phone could be used around the world relatively easily.  In the 2G era the whole world used a total of 4 bands so a ‘quad band phone’ could support voice and (where available) data services anywhere.  Newer generations of cellular technology got more complicated with different mobile networks choosing different wireless protocols and frequencies to operate on.  As a phone’s compatibility with more networks increases so does the complexity of electrical engineering.

It became more difficult for phones to provide global coverage on networks starting with 3G.  As the presence of IoT connectivity and custom Android devices grows Hatch shares what to consider when designing a phone for optimal global coverage.

Generations of Wireless Telecom Protocols

The 2G network uses the GSM standard.  4 frequencies 850, 900, 1800, and 1900MHz covered all the world’s main markets.  Most phones only supported dual bands, but it was easy to find quad band options also.

3G introduced incompatible network standards.  Phones that were built for UMTS often didn’t support CDMA.  Plus within each of the different 3G networks there were at least 5-6 different frequencies.  Only a few high end phones provided global coverage on 3G.  Generally more expensive phones and telecom devices support a wider range of frequencies than lower cost ones.

4G also introduced incompatible networks, but the LTE network had much better coverage than the competing WiMax technology.  Across the world there are over 10 commercially used frequencies on the LTE network.  More than most phones can support.

5G looks to keep things more simple.  The 5G NR (New Radio) network has been introduced in a handful of countries and doesn’t appear to have any rival standards, although there are over 15 different frequencies on this network.

Decommissioning of Older Networks

In many countries legacy 2G and 3G networks are shutting down.  Mobile networks try to move subscribers to the newer networks so they can cut the costs of maintaining older networks.

Backwards Compatibility

Phones with the newest wireless technology are backwards compatible.  That means the newest phone which supports 5G will also support 2G, 3G, and 4G (wherever those networks exist).

New Network Benefit

Newer wireless standards, like 5G, are great for quickly consuming tons of data.  Older generations offer slower data rates.  Mobile users who want to stream high res movies or play graphics rich video games have the best user experience on 5G, but for many IoT or lower data use cases even 2G sufficies.

Designing a Global Android Telecom Device

If you’re making a custom Android telecom device to use globally the key details to consider in regards to choosing frequencies are outlined below, based on the information provided above.

First consider the device’s data consumption profile.  If it doesn’t consume a lot, for simple IoT purposes or low data applications, 2G will keep you in business.  Obviously if your application requires more data consumption then 2G may fall short.  Then research which networks your SIM card will roam on in different countries where the device will be used.  The roaming networks are determined by the roaming agreements of the original SIM card provider.

Next check whether the countries where the device will be used still have 2G (or 3G) networks. Information about countries sunsetting 2G and 3G networks is easy to find on Wikipedia (where information for this article comes from).  Then research which networks your SIM card will roam on in different countries where the device will be used.  The roaming networks are determined by the roaming agreements of the original SIM card provider.  Notice which networks and frequencies overlap the most between different countries based on the carriers your device will use.  What often happens is that not all the bands within one generation (ie, 4G) can be covered by one device.  This happens because so many different frequencies exist around the world.  To overcome this, start to look at which frequencies on other generations (ie, 3G) are most used by the roaming partner networks.  The idea is to aggregate the most overlapping networks in different countries so that the device can work anywhere it needs to on at least one generation (2G, 3G, 4G, or 5G).

Designing a phone with multiple custom ranges of frequencies.

Traditionally the frequency bundles in mobile phones are broken into US, Asia, and EU regions.  In each regional configuration the frequencies are all set to serve that local market for best coverage across all generations.  That’s a different logic than getting as much coverage as possible globally.  Most generic mobile phone suppliers allow the wholesale buyer to choose the regional configuration before manufacturing.  For example if buying phones to use in the US then the buyer would choose US configuration.

When making a custom Android device you can choose which bands the device supports based on your specific needs.  Up to 3 custom configurations can be created, so you don’t need to stick with the traditional US, EU, and Asia configurations.

Looking to the Future

In the world of technology having common standards sometimes works better than having options.  There are varied reasons for the existence of competing standards, however financial interests probably have the biggest impact.  Whatever new roads appear as you pursue making a custom Android device, Hatch is there to help navigate them. 🙂