Prototype Version (left), and Final Version (right).
This month Hatch introduces a new kind of article, one that profiles a unique architecture for Android hardware. Overtime, as we showcase more unique hardware, you may find something which serves as a better platform to test your software than a generic Android tablet.
When Hatch first started making custom Android products around 2015 most demand revolved around mainstream retail tablet and smartphone devices. As visionaries imagined new use cases and industries created new demands, modified hardware architectures gained in popularity. One of the earliest high volume use cases that evolved from a niche are point of sale (POS) devices. Now there are several engineering and manufacturing companies that exclusively focus on this category since the global demand justifies a high level of support. Same thing is true for Rugged Android tablets and phones, however the high prices of these may still justify going custom if order volume is high enough.
Modifying hardware for a specific use case is only part of what’s needed to deliver an end product. The other, and arguably more important, part is designing the software to go with the hardware. Software entails a front end app and, usually, a back end system that interacts with the app. Once stable hardware is developed for industrial or business applications it stays the same for years, however elements of the software are regularly updated to improve user experience or resolve bugs.
Hatch sees the growing popularity of niche applications for Android as a good thing for the industry and synergistic with our business model. A new generation of niche Android products make it easier for entrepreneurs to realize their vision by running tests on existing hardware. Often existing hardware can be used for prototyping and beta testing. If testing goes well using existing hardware the next step is making custom Android hardware. Below we look at a device that’s been designed to replace the traditional keyless entry system.
This Android device combines traditional systems with new technology. More powerful electronics supports new use cases, such as keyless entry using facial recognition. The facial database is stored in the device, either uploaded in a file or acquired using the camera, and recognition processing happens locally on the device in real time.
A Wiegand interface designed into the device’s motherboard connects to an external card reader. Wiegand is an older and still popular wiring protocol for connecting swipe card readers to an underlying access control system.
In addition to Wiegand there are several other interfaces including LAN (Ethernet), USB, RS-232, Power, and GPIO buttons. The RJ-45 Ethernet port can be used for PoE (power over Ethernet) or maintaining a hard-wired internet connection rather than relying on Wi-Fi. Other ports can be used for opening and closing access points mechanically or as defined by the user.
The IP65 case has been designed to withstand light rain and dust. This helps protect products exposed to rain or other outdoor conditions.
If the current architecture of the product detailed above serves your purpose better than a standard retail tablet we welcome you to use it for your initial prototype. Once you decide to make a custom version, future iterations of this product can add mobile data connectivity, body temperature sensing, or whatever customization serves your purpose best.
Design a tablet for photographers to quickly review their shots and copy photos from the camera’s memory card to a spare card.
Use Cases
Photographers want to quickly review photos on a clear and large screen then move images off their camera’s memory card so they can get back to taking photos again. Hatch’s client identified this unique market and turned to Hatch to manufacture a custom Android tablet tailored to serve it.
Hardware and Mechanical Requirements
9” Screen
No Wireless Connectivity (mobile data, Wi-Fi, Bluetooth, etc)
Reference design modification to add extra SD card slot
Assistance with Industrial Design
Mechanical Engineering
Creating prototypes
Tooling for device case
Manufacturing process optimization
Create finished product testing procedure
Trial and Mass Productions
Time Frame
This project moved fairly quickly from Hatch’s end. Two months passed from kickoff to initial prototype. The customer took a few months to thoroughly test the initial prototype, which is time well spent. Another month was used to refine aspects of the firmware in parallel with producing the tooling for the tablet’s casing. Trial production took a month and then the customer needed extra time to refine and debug their app. The whole process took about 8 months, but most of this time was used for app related development and optimizations.
How it Happened
The client initially approached Hatch about working together when they first started conceptualizing the project around 2017. At that time, as a new company and concept, they couldn’t commit to Hatch’s minimum order quantity. We wished them good luck and ended the discussion. They came across another tablet supplier who said all the right things, won the business and then forced the client to place a large order (which they weren’t ready for) or lose their deposit. If everything went well that’s where the story would have ended, but after deceptive business practices the other supplier also shipped bad products and were difficult to work with. At that point the customer contacted Hatch again and asked for help with getting out of their dilemma. Typically we don’t get involved with this kind of situation, but the client seemed like good people in a bad position. We agreed to help them fix the problems and manage production for the rest of their order from the first supplier before redeveloping the product ourselves.
As the first order was coming to an end Hatch started to redevelop and improve the product. Hatch had the benefit of learning from the mistakes of the first development, which helped our development go smoothly and make the customer experience even better.
The device had several custom hardware requirements that deviated from the standard reference design, the electronics design which ‘off the shelf’ Android tablets are based on. For example, remove Wi-Fi, Bluetooth, cameras, headphone jack, unnecessary apps, and increase battery size. Also Hatch needed to add a second memory card slot and make both memory cards full size SD, while the reference design calls for one MicroSD size memory card. Most of these hardware customizations also required related firmware modifications.
Second SD memory card slot for copying photos.
Result
The results were great. A large part of the success came from the lessons learned from the first generation of the product. Our client was able to earn a reputation for higher customer satisfaction than the competition (in their niche) thanks to the great job they did with redesigning their app and the improvements Hatch made to the hardware and firmware. This allowed them to raise the retail price, helpful in offsetting the high costs of fixing problems with their first generation product.
It’s hard to predict how things go in business, but some situations create good stories when told in retrospect. Sometimes, for whatever reason, people don’t connect with potential friends/customers/whatever on the first interaction, but a relationship develops in the future. This client has turned out to be a great company to work with and Hatch provided key value to them. Often companies contact us who aren’t ready to work with Hatch when they first reach out. I try to guide them in the best direction for them rather than sell them services they’re not ready for. Hopefully my guidance or something else they did brought success. It’s always a pleasure to see companies flourish and become able to help them.
Android Devices Show Resilience Through Rapid Adaptation
Custom Android tablets are helping people push through the challenges of the Covid era. New use cases and growth in existing niches have driven innovation and demand for Android hardware.
When Covid arrived in Q1 2020 many Android manufacturing companies quickly reacted with a pessimistic outlook on business prospects for the year ahead. Key component manufacturers decreased their annual production forecasts, lowering the quantity of components produced for the whole year. Agile Chinese tablet manufacturers responded drastically. Many factories transformed their production lines to make masks, gowns, and other PPE. Engineering companies shifted attention from Android PCB design to designing digital thermometers. From one month to the next, customers went from getting price lists for tablets to price lists for medical devices and PPE from the same companies.
Covid didn’t kill economies. It halted industries. Travel, restaurants, cinemas, and many other industries have been decimated by Covid. Other industries ignited. Online education, remote working, and ecommerce/logistics, are examples of industries undergoing a renaissance with heightened user demand and investments. Many workers from the hard hit industries lost their jobs. With less disposable income people have less money to replace old phones or tablets. These were the people that industry executives thought about when reducing forecasts for component production.
What was overlooked was that if people couldn’t congregate in offices and schools then they’d need to work remotely. This created a boom in sales for Android tablets and other devices used to work or study from home.
Android’s system architecture makes it relatively easy to add different peripheral electronics to the core platform. Innovators can also quickly build apps for a wide range of use cases.
The combination of fast hardware and software development are what make Android an ideal platform for quickly adapting as the world and technology evolve.
Android products have become more than a mobile phone or tablet for the consumer market, where several global brands sell fairly similar products. The Android manufacturing ecosystem in Shenzhen quickly responds to opportunities with hardware for serious business, health, and education applications at speeds that weren’t possible until recent years.
Examples of Android Devices that Grew as a Result of the Pandemic
Access Control Android Tablet with Body Temperature Sensor and Facial Recognition
Before the pandemic started there were a few custom Android devices with access control. Once temperature testing started becoming a part of access control, temperature sensors were added to the product. Now, instead of a person taking other peoples’ temperatures at entry points for office buildings, malls, or supermarkets, a tablet does this, and much more, automatically. It takes peoples’ temperatures, assigns the temperature to a specific person, and records that information instantly for future reference as a pandemic control measure (while this raises privacy concerns that is another topic for another article).
Several different companies started making their own version of this product with little twists in design or components. Because of how many companies made their own version of this product it may not be necessary to custom make anymore since many options are available. Some of these tablets are built into a self driving robot device that approaches people for use in hospitals and restaurants.
Educational Tablet for Children
From elementary school to colleges, remote education was forced upon students around the world. With so many students studying from home the global demand for tablets tailored to serve them exploded. While the electronics don’t vary much (or at all) from standard tablets, specialized apps and casing designs differentiate educational tablets. The huge demand for educational tablets overcompensated for the people with reduced discretionary spending and drove up component prices across the whole industry. The price increases were also a direct result of Huawei hoarding parts due to trade restrictions and reduction in production forecasts from component manufacturers, as mentioned above.
So Many Use Cases. So Many Android Custom Solutions.
Covid has brought to light how the Android platform allows for diverse and fast development of new products. Hopefully demand for pandemic related products will go down soon. Whenever that happens there will be new opportunities calling on engineers to adapt Android in new ways. If you want to bring a unique Android product to life, Hatch can help. We’re always happy to talk with entrepreneurs and engineers about your project.
In this article ‘supply chain’ refers to all the contributors of components or services that go into manufacturing a finished product. More specifically, in the case of custom Android manufacturing, that means suppliers of hardware such as screens, PCBA, cameras, cases, etc. Also different specialized service providers such as electronics, mechanics, software, and assembly engineers. General services, such as logistics, which aren’t specific to Android hardware are not covered in this article.
What’s a mature supply chain?
Supply chains become mature when multiple suppliers gain expertise and compete on individual pieces of the supply chain. A supply chain consists of many different elements so when there are many companies competing in one niche of the supply chain, over time that creates optimal service and price (theoretically). For example there are hundreds of companies that design and produce plastic cases for electronics around Shenzhen. Several of them are particularly focused on designing cases specifically for Android tablets or smartphones. There are so many case suppliers and so much demand for Android device cases, some case suppliers carve a niche by specializing in tablet or smartphone cases.
This creates teams of engineers with specific retained knowledge that apply directly to making Android devices. Each case company has engineers with experience designing cases for optimal performance and manufacturing efficiency. For example an experienced mechanical engineer designs the camera slot in a way to ensure the camera easily snaps into place, blocks light from the flash, and doesn’t overheat. They know what works, and have seen many things which don’t, creating a reliable and efficient supply chain. More experienced component manufacturers working in a supply chain makes that supply chain more mature.
What’s an immature supply chain?
Supply chains are immature when there are limited suppliers with deep (or any) expertise in bringing a finished product to life. For example when developing a new product, especially taking that product from prototype to mass production, countless surprises and mistakes happen due to inexperience. It could be that the product was designed by non-Chinese engineers who choose components that are difficult to get in China, necessitating a design modification. Even for products fully designed in China, components don’t always work well together. Some materials don’t mix well together for chemical or electrical reasons. Some components come with different calibrations that require tweaking for consistency during mass production. Immature supply chains result in surprises that take time to resolve.
Production line staff at factories don’t usually understand the point of the custom modifications as it pertains to the product’s use case. Because of this sometimes small variations in workmanship have a bigger impact on product performance than the staff realize. Over time mistakes become lessons, but until that happens the supply chain is immature.
How does a supply chain become mature?
Immature supply chains become mature over time as orders of the finished product increase. For example when Android tablets first started shipping from local Shenzhen supply chains around the end of 2010 the product quality was low and unit cost was high. As demand exploded new suppliers entered the supply chain. Even with these new entrants it took a while to meet demand and have prices start to come down. Over time the profit margins decreased and maturity of the supply chain evolved. Once supply caught up with demand they needed to improve product quality and price in order to compete. Sales volume and, in turn, competition are the key forces that drive a supply chain to maturity.
The more orders suppliers produce the more opportunity they have to learn from mistakes. This drives improvement of their components and, ultimately, the finished Android tablet. As volumes of Android tablets and smartphones exploded a massive supply ecosystem came into existence. The same events take place in the creation of supply chains for lower volume products, albeit with relatively fewer participants, as demand is lower. A large company producing high volumes of a unique product will also create a mature supply chain. This happens somewhat through brute force rather than purely organic capitalism. In this case entering the supply chain is more restrictive than with a product anyone can buy since there’s only one customer who chooses the individual suppliers.
How to work with an immature supply chain
It’s necessary to work with an immature supply chain whenever creating a new product or modifying an existing one. Getting from idea to mass production means going through a learning curve. Over time production runs become more efficient by identifying and fixing problems. That’s why initial production runs should start with smaller quantities. Problems are easier and less costly to fix with lower volume.
In addition to extensive testing with low volumes, having multiple suppliers for high risk parts is also useful to hedge risk. High risk parts refers to parts that have a higher tendency to go out of production or become difficult to source. Often a component’s future availability is tough to predict. From experience, cameras and screens seem to have the highest risk in custom Android devices.
Supplier value in a mature supply chain
Once a supply chain becomes mature so do the buyers. Experienced retail buyers know what’s important to look for in a supplier. The supplier traits which matter most in a mature supply chain are quality, speed, price, and reliability. Because finished products coming from a mature supply chain are so similar to each other there’s rarely room for significant product differentiation. Strides in innovation are made in small, manufacturing related, baby steps. Any big technological innovation will not likely come from a manufacturer, at least not for Android products.
How an immature supply chain is helped by a mature supply chain
If you want to make a custom Android product your development and manufacturing partner, such as Hatch, plays a big role in leveraging mature supply chains to create custom products. Some suppliers in a mature supply chain are willing to support smaller production runs at a higher margin. By working with them Hatch leverages the experience and efficiency of the mature supply chain to reliably produce what is needed for the lower volume custom product.
Conclusion
Anytime a new commodity product or product category kicks off a supply chain must be created from scratch. The initial entrants do what they can to provide a ‘good enough’ early stage product, but as demand for a product grows and more companies enter into the supply chain the quality gets better. When making something custom leverage mature supply chains to gain a head start.
Whenever working with a new and immature supply chain, go slowly to find problems early. Reduce risk by thoroughly testing multiple rounds of a trial production and making clear requirements, sometimes even SOPs, for the suppliers to follow. All mature supply chains start as new and immature ones. They’re both important at different stages of a product’s lifecycle and through constant intersection help each other to grow.
This month’s post takes a pause from custom Android hardware to talk about a fun side project Hatch has been working on: Snipes, a brand new ice hockey training product. While we’re usually making products for our customers’ businesses we decided to have some fun doing our own project together with a couple of friends.
Due to Covid related travel restrictions, I ended up staying in China more this year than usual. In an effort to make the most of this, I got together with an ice hockey friend to develop a unique ice hockey training product that we wanted to use ourselves. This is a story of envisioning a product that didn’t exist, developing it through a great team effort, learning about the online marketing side, and ultimately receiving a great response from the target market.
The idea was clear. Create targets that go inside an ice hockey net to shoot at. Each target has lights and a sensor. When the lights of a target turn on, the player shoots a puck at it. If the target is hit then the lights turn off and the lights of another random target turn on. From this basic setup, various different game modes can be created. We named the product Snipes.
Taking an Idea to Something Real
Step one was finding the right target material that could sustain the force of being hit by 6 ounces of hardened rubber flying 100mph / 160kph.
We ordered different materials to test and found one that we felt comfortable with.
Next step was hooking up an Arduino to lights and sensors, but first we needed to decide which kind of sensor to use. After testing a few different options we went with a Piezo sensor. This senses the level of vibration.
Before attaching everything on the targets we created a mini version with just the lights and sensors to make sure that we could simulate the desired functionality.
First ‘working’ prototype was created and destroyed. That happened about 5 more times.
At this point we decided to go with wireless connections between the targets, meaning custom electronics and firmware were created. This development took a few weeks. When the first samples were ready the initial result was promising.
The first Arduino prototype didn’t use any case for the electronics at all. All the components were held onto the target using tape! We evolved the mechanics using a FDM 3D printed case, which was better than nothing, but broke. That’s when we went with CNC milled cases. These provided a nearly mass production level of strength.
Something Real to Selling Promises
We went through several rounds of fast and cheap prototypes before the final set of custom electronics, firmware, and casing was ready. Each time a problem was found, an improvement was made. Once the problems became harder to find and the testing went smoothly we took a video of people using the product and started our pre-sale launch.
Early response from the launch indicated that people liked the product. We went ahead with opening the tools for the casing mold and decided to commit to a mass production order as our pre-sales were going well. We forecasted how many would sell in the launch campaign and wanted to make extra inventory to have available after the campaign as well.
The target ship date was set for early December, much earlier than I felt comfortable with, but still possible to accomplish if we worked hard and had good luck. Products will ship individually from the factory to the pre-sale buyers in hopes of delivery before Christmas.
Instead of using a normal crowdfunding site we used paid social media ads to send customers to our own pre-sale page. My partners thought that since our product was fairly niche all sales would come from targeted advertising rather than being on a crowdfunding platform. Instead of paying a percentage of sales to the crowdfunding platform we’d use that money to pay for direct advertising. I wasn’t sure how this would go over, but the logic seems to have worked. Maybe next time, if the product is intended for a wider audience, we’ll try going with a platform or hybrid approach.
The Shooter Tutor, a well-known ice hockey training product Snipes is most similar to, has been around for decades. When hockey players see videos of players using Snipes they instantly understand how to use the product. For this reason online advertising using videos has been successful. There’s little need to further educate the customer after they see a short video. While the application is obvious there’s nothing like it on the market so we’re in a good position.
Selling Promises to Products Delivered
This is the stage we’re in now. I’m excited about getting Snipes out to the hundreds of customers who believed in us and joined our pre-sale. Most of the materials for manufacturing are ready. We are working to implement some last-minute changes on Snipes’ firmware and then must test that.
Shipping in time for Christmas delivery is still a possibility. Need to see how testing goes with the new firmware and whether any unexpected delays pop up. Good quality is more important than shipping earlier. We’re confident that customers will love the product when they get it.
Next Steps
Making this sports training product has been a fun project. The development time is much faster than making custom Android products as the functions are more simple and case design requirements are more about strength than aesthetic design. The Bolt team has ideas about new products that will expand the brand’s footprint deeper into ice hockey as well as other sports.
Since people couldn’t travel this year I spent more time than usual in Shenzhen. That gave me the opportunity to take on a fun side project. Now and in the future, Hatch continues to work on custom Android development and manufacturing. With the growing prominence of IoT devices, we’re open to evaluating IoT hardware partnerships as well.
Since the explosion of Android tablets started almost 10 years ago several CPU (central processor unit) manufacturers have come and gone. In the beginning high profit and high volume drew companies into the space, but as the volume of Wi-Fi Android tablets dropped and competition grew a few clear leaders have emerged. Those leaders include Rockchip, Mediatek (which also makes CPUs for telecom tablets and smartphones), and Allwinner Tech. These companies all make chips based on the ARM architecture, so many of their chips compete head to head and are fairly similar in spec and performance.
To keep it simple this article will focus on Rockchip since Hatch routinely uses their CPUs. The main reason we use more Rockchip is due to flexible customizations (important when making custom Android devices), engineering support, and because we have stronger relationships with Rockchip engineers. The other companies also make high quality Android CPUs that work for custom projects and we are supportive of using them if our customer requests.
The list below provides information about the leading Rockchip CPUs that are recommended for use in custom Android Wi-Fi devices in order of performance, from high to low end.
RK3399Pro
The RK3399Pro is called a ‘server level’ CPU. It’s mainly intended for use in applications that demand high performance such as an Edge AI server, industrial applications, client-side facial recognition, and others of this nature.
The CPU contains 6 (hexa) ARM cores, consisting of 2 Cortex-A72 (1.8GHz) and 4 Cortex-A53 (1.5GHz), a NEON coprocessor, Mali T864 GPU, and an NPU. It can run multiple learning platforms such as Caffe and Tensor Flow. The RK3399Pro supports multiple display outputs up to 4k resolution and multiple camera inputs up to 13MP. It also has native support for USB type-C protocol.
(Before the RK3399Pro came the RK3399. The RK3399Pro is the RK3399 with the integration of the RK1808 IC. The RK1808 provides the NPU (neural network processor unit) mentioned above.)
RK3288
The RK3288 is a high end Android CPU that features excellent display support and strong processing power. This chip is used in a range of consumer, business, and industrial devices. Many large screen advertising displays use this chip. Applications include live video streaming with add-on electronics (like temperature sensor), laptops, and access control terminals. The CPU contains 4 (quad) ARM Cortex-A17 cores that deliver speed up to 1.8GHz and a Mali T764 GPU. It supports up to 4k display resolution, dual display output, and dual camera up to 13MP (with internal ISP).
RK3368
The RK3368 is similar to the RK3288 with slightly lower specs. It still offers excellent multimedia features and does a great job in high level consumer or business applications. The RK3368 is commonly used for educational tablets, TV boxes, advertising displays, and premium tablets.
This CPU has an 8 (octacore) architecture with 8 Cortex-A53 cores that run up to 1.5GHz. It supports up to 4k video HDMI output powered by a PowerVR G6110 GPU. The internal ISP supports 1080p recording and up to 8MP images.
RK3326
The RK3326 is a mid-range CPU that has become popular with devices like retail tablets and smart speakers that have moderate performance requirements and need the newest technologies. The chip is a step up from the RK3128 in that it provides faster processing speed and an internal ISP for image processing.
This CPU has a 4 (quad) core Cortex-A35 processor that runs up to 1.5GHz. The Mali-G31MP2 GPU delivers video output up to 1080p resolution. The internal ISP supports video up to 1080p resolution and images up to 8MP.
RK3128
The RK3128 is the entry level CPU intended for mass market consumer level devices. Its high volume ensures stable functionality and reliable performance in tablets and TV boxes for lower resolution displays. It’s a solid option for custom Android devices that need long battery life with limited functionality.
This CPU has 4 (quad) Cortex-A7 cores running up to 1.2GHz. It integrates a Mali-400MP2 GPU that provides video output up to 1080p resolution. The CPU also supports cameras with recording up to 1080p and images up to 5MP.
Some CPUs have multiple versions that offer slightly different features. For example, the RK3128 and RK3126c are exactly the same CPU, but only the RK3128 has HDMI connectivity and supports DDR2. For products that don’t leverage the full features of a CPU using the ‘cost down’ version (like the RK3126c) saves cost on the BOM (bill of materials). Once the requirements of a product are clearly defined with a client Hatch will help to accurately identify the most suitable chip to use.
When architecting custom Android hardware the first thing to consider is which CPU to use. A CPU is the central IC chip that powers a digital product, sometimes referred to as the ‘brains’ of a product. Choosing the CPU is the most fundamental design decision for your product because the capabilities and limitations of this IC predicate the capabilities of the entire device.
Two CPU companies that continue to dominate homegrown China products are Mediatek (Taiwan) and Spreadtrum (China). While Qualcomm, globally, is the leading brand for top tier mobile phone CPUs, it’s more suitable for high volume (100k+ units) consumer devices than lower volume (10k+) custom devices. Designing phones with Qualcomm CPUs requires more engineering expertise so development costs are higher. The performance benefits that Qualcomm CPUs provide are often overkill for what custom device customers need anyway.
Mediatek and Spreadtrum advertise many different CPUs on their websites. The reality is that only a few of them are commonly used by players in Shenzhen’s manufacturing ecosystem, the world’s source for tech hardware. The websites show several older and brand new CPUs, neither of which are good for custom products. Older CPUs may soon become obsolete and lack performance. Brand new CPUs are often too new for stable mass production and their features are overkill for the needs of most custom projects. The CPUs that have become popular gained traction because they offer the best value for the segment they serve.
(Relatively) High End Segment 4G
Both Mediatek and Spreadtrum are showing 5G chips on their website and a few large (mostly Chinese) mobile phone brands have already released phones using CPUs from Mediatek’s 5G Dimensity line. When new chips come out larger brands which reliably sell high volumes are the first to market with products. It takes a lot of engineering resources to develop a product using a new chip which is why CPU manufacturers partner with high volume brands to release the initial products. Once these brands have come out with products and the system becomes more stable Shenzhen’s smaller development companies start releasing products for smaller brands.
Because 5G solutions aren’t stable enough for smaller market brands yet they’re still not ready for most custom projects. At this point 4G CPUs are still the most practical for custom projects as the engineers are most comfortable making products on these platforms. Therefore, even for the high end segment, this article focuses on 4G CPU options.
Mediatek G90
This is Mediatek’s top of the line 4G CPU. It offers high performance for playing resource intensive video games and high end camera support. Mediatek describes it like this:
“An upgraded octa-core CPU boosts speeds up to 2.05GHz. Using a mix of the latest Arm Cortex-A76 and Cortex-A55 in combination with a large L3 cache, the G90 series increases multi-threaded performance up to 9%1 beyond direct competitors.
With speeds up to 800MHz, the latest Arm Mali-G76 3EEMC4 GPU can play Fortnite and PUBG incredibly smoothly, blazing through Manhattan benchmark up to 26%1 faster versus direct competitors.
Up to 10GB of LPDDR4x at 2133MHz1 (4266MHz1 effective) provides generous bandwidth and memory capacity, while the latest UFS 2.1 storage yields ultra-fast data-streaming.”
Mediatek P70
The P70 replaces the successful P60 offering up to 13% faster performance and 35% lower power consumption. The chip includes an upgraded dedicated AI Processor Unit (APU) that provides up to 30% more processing efficiency than its predecessor. This chip is marketed towards gamers, camera enthusiasts, and advanced AI applications. Could be a good choice for custom products that need high performance onboard processing. Here’s what Mediatek has to say:
“The MediaTek Helio P70 incorporates a powerful Arm Cortex-A73/A53 octa-core CPU complex with an impressive Arm Mali-G72 class GPU. Versus the prior P60, the faster clock speeds generate up to 13% more performance. In-hand, there’s support for 20:9 displays at Full HD+ resolution that allow users to enjoy the most beautiful and modern smartphone designs with full fascia coverage.
MediaTek’s exclusive CorePilot 4.0 technology manages how tasks are assigned between processing resources, with a power management focus that’s on maintaining a sustainable, fast user experience through thermal management, UX monitoring and Energy Aware Scheduling (EAS+).”
Mediatek G70
This chip offers similar, but slightly reduced, features as the P70. It became mainstream in early 2020, about a year after the P70. Mediatek has this to say:
“The G70 incorporates a pair of powerful Arm Cortex-A75 CPUs operating up to 2GHz, plus six Cortex-A55 CPUs in a single, octa-core cluster. These are interlinked and share a large L3 cache for improved performance plus up to 8GB of fast LPDDR4X memory. For gaming, an efficient, high-performance Arm Mali-G52 class graphics processor operates up to a speedy 820MHz.
A multitude of hardware accelerators, such as a dedicated depth engine, Camera Control Unit (CCU), Electronic Image Stabilization (EIS) and Rolling Shutter Compensation (RSC) technology enhances video panning and ultra-fast recording (up to 240fps).
Ideal for popular AI-camera tasks such as object recognition (Google Lens), Smart Photo Album, scene detection and segmentation with background removal, and Bokeh-shot enhancements.
With MediaTek NeuroPilot support and full compliance with Android Neural Networks API (Android NNAPI), developers and device makers have the best possible ecosystem for Android enhancements and app development, with support for many common AI frameworks.”
Upper Mid Range 4G
These lower cost CPUs provide terrific performance for many dedicated single app custom Android products. The CPUs in this category provide an excellent balance of performance and features.
Mediatek P35 (MT6765) and P22 (MT6762)
These chips appear to use the same core, with key differences in display and camera resolution. The P35 supports display resolution up to 2400×1080, dual 13+13MP cameras, and a single 25MP camera. The P22 provides respectable, yet reduced, display resolution up to 1600×720, dual 13+8MP cameras, and a single 21MP camera. Both chips come with a suite of Mediatek technologies like CorePilot, Imagiq, MiraVision, NeuroPilot, Pump Express, Tiny Sensor Hub.
Mid Range 4G
Mediatek A22 (MT6761)
The A22 has the same feature set as the P22, except it uses a quad core architecture while the P22 uses an octa core architecture. The A22 is a good choice for custom Android devices that need a solid feature set, but don’t demand excessive computing power. Noteworthy for possible IoT applications “its tiny, ultra-low power sensor hub is ideal for apps that require always-on sensor readings (pedometers or vocal-triggers), without having to engage the main CPU complex, providing significant power savings”, according to Mediatek.
Spreadtrum SC9863A
Spreadtrum trails behind Mediatek in technology, market prestige, and number of CPU models, but the limited options it offers do a good job covering the low to mid-range price driven market. The 9863A delivers a strong octa core architecture, with speeds up to 1.6GHz. The camera resolution tops off at 16MP and supports dual cameras. Screen resolution supports up to FHD+ (2160*1080).
Mediatek MT6739
The 6739 has dominated the market for high quality lower-mid spec consumer devices since 2017 (and probably for the foreseeable future). This means a very high volume of these chips have shipped out, making the electronics and firmware stable, exactly what many custom products need most. Facing competition from Spreadtrum’s 9863 on the high end and Spreadtrum’s 9832 on the low end, a quality reputation continues to drive sales, even at a higher price point. Here’s what Mediatek wants you to know:
“Combined within the MT6739 is a highly efficient 64-bit quad core, ARM Cortex-A53 CPU that operates up to a fast 1.5GHz. Its IMG GE8100 GPU operates up to 570MHz and supports the latest 18:9 display type with HD+ (1440 x 720) resolution.”
Spreadtrum SC9832E
For the most price driven full feature smartphone market the SC9832E delivers. Supporting cameras with a resolution up to 13MP, including 1080P video recording, and display resolution of HD+ (1440×720) or under, the SC9832E has had dominating success with entry level smartphones in price sensitive markets. As the chip has been used for years by both regional brands as well as tier 1 international brands its capabilities have been proven in the consumer market.
Low End Range 3G
Mediatek 6580 and Spreadtrum 7731
Both of these chips have had very long and successful lives, making them the headlining 3G chipsets for their respective brands over the past 6+ years. The specs of these two chips are fairly similar. Both have a quad core Arm-A7 architecture, support up to 1440×720 screens, and a 13mp (6580) or 8mp (7731) camera. There are few reasons to use either of these chipsets anymore, but a small market still exists. Custom Android hardware that doesn’t need high specs or high bandwidth may be a good match. These chips are probably still produced to serve legacy products more so than used to develop new ones.
An Alternative Approach
In situations where a brand wants to use a Qualcomm CPU there are now Qualcomm SOM (system on module) options available. Using an SOM allows for faster and more economical development of custom Android hardware using a Qualcomm platform. End products using SOMs are usually higher cost and, depending on the product, require a thicker case, but come with some advantages also.
A SOM is a module that contains a complete mobile phone PCBA in a miniature form factor. Some SOMs come pre-certified for different international safety and mobile network standards. This makes getting those certifications for the final product less expensive and faster. Products can be designed using an SOM by soldering the SOM on a ‘main’ PCB and then adding peripheral electronics to the main PCB. Peripheral electronics include ports, such as USB-C, ethernet, HDMI, etc, special electronics such as NFC, LoRa, etc, and interfaces with a screen, battery, touch panel, etc. If using Mediatek or Spreadtrum CPUs the IC chips are soldered directly on the main board.
There you have it, party people; a bunch of chips worth getting to know for your next custom Android project. Think about what works for you or give us a shout at Hatch, and we’ll help you decide.
Looking to enter into a business relationship with a new supplier or customer? Our tips to evaluating business partner relationships provide important pointers that will lead to success.
