Embedded Development: Computer on Modules vs Chip-based development
Chip-based or custom embedded development compromises the time-to-market and cost of the product. Computer on Module offers an ideal platform for building end-products for a variety of applications. The fictional case presented below illustrates the issues with chip-based development and how Computer on Modules can alleviate these issues.
"Harry is the CTO of BestECG Company that manufactures ECG machines. BestECG competes with two other companies in their region for supplying ECG machines to diagnostic labs and hospitals. The labs and hospitals were demanding more advanced machines that could provide 3D displays, real-time data capture, and higher processing; however, without compromising much on the battery operating time.
To achieve this, Harry decided to use the advanced GenX processor from a reputed silicon vendor. GenX processors, which is manufactured on 14nm node, offers a peak frequency of 1.2 GHz, supports advanced multimedia, and packs various power optimization techniques. This will provide a huge boost to the ECG machines’ performance and low power consumption. Harry contacted the vendor and ordered 200 pieces of GenX SoCs. As the sales volume was low, Harry had to pay high price for the SoCs. Then, he engaged his hardware designers for board development and mechanical design. Simultaneously, the software team started to port the existing software stack on a reference board based on GenX SoC. Multiple software components were written from scratch and the porting lasted for around 8 months. The hardware prototype got ready 1 month later. Then, the software stack, including application, GUI, device drivers, and middleware, was tested on the actual hardware with all the I/Os. After some rework, the software became production grade. The hardware platform along with the enclosure was also finalized. It took around 16 months for the project to be completed.
Then, Harry applied for medical certification of the full product. The entire team of 50 engineers with diverse experience in hardware, software and mechanical design worked on the project. The total cost including the salary, raw materials, etc. was reported to the management. Two weeks later, one of the competitors of BestECG launched an ECG machine based on GenX processor. The product got very good response from the market and sales of the competitor’s product picked up rapidly. Harry was really upset, as BestECG product based on GenX could be launched only after 2 months. BestECG lost huge market share. The delay of 18 months from concept to production had cost their company dearly.
Harry realized that he had to accelerate time-to-market for future products without compromising on development cost and risk. So, after some study, he found that Computer on Modules (CoM) offered him the best solution to achieve this goal. He chose Toradex as the preferred partner for CoMs.
In comparison to chip-based development, in which Harry had to buy the critical components such as SoC, and NAND, at higher price owing to low sales volume of the end-product, using CoMs, the cost was optimized as he paid for the entire platform. The CoM vendors leverage economies of scale by engaging in large volume business with silicon vendors, and thus buy the components in lower prices. The CoM offers an off-the-shelf platform that has production grade operating system, device drivers and BSPs, so Harry only needed a small team size of 20 people for the next ECG machine project. Multiple projects could be executed simultaneously within BestECG with only 50 engineers. The software team worked on developing applications and GUI. The hardware team designed the carrier board that has the application-specific I/Os. This carrier board houses the CoM by means of some standard connectors such as SODIMM.
Overall, Harry was able to achieve better utilization of the engineering resources and develop multiple variants of ECG machines simultaneously as the same carrier board can fit in multiple pin-compatible CoMs with different performance and price. The project, which earlier took around 18-20 months, now takes only around 10-12 months. Further, the input cost of product was also optimized. As the software stack from the module vendors was already tested in various other end-products, Harry also face less regressions issues. The issues are mostly limited to software integration and application layer.
Apart from controlling cost, Harry was also able to commit a long product life to his customers. The CoM vendors ensures drop-in replacements of obsolete hardware components such as RAM, SoC, and NAND, in case any of these reach End-of-Life (EOL). So, Harry don’t have to worry about the component obsolescence. Further, pin-compatibility among CoMs ensures that an application can be adapted as per future requirements without redesigning the carrier board. With the same carrier board, he can plug-in a CoM based on advanced processor to boost up the product’s performance. Earlier, with the conventional chip-based development, such an upgrade would have involved a redesign.
Harry was able to address embedded OEMs’ concerns about long product life and low sales volume by using CoMs in product development. Chip-based development cannot address these concerns as mentioned in the blog here.
Few months later, the CEO of BestECG asked Harry, why he chose CoMs instead of Single Board Computers (SBC) as an alternative to chip-based development? Would you like to know Harry’s response? Yes, then check out my next blog."
**This is a work of fiction. Names, characters, businesses, places, events and incidents are either the products of the author’s imagination or used in a fictitious manner. Any resemblance to actual persons, living or dead, or actual events is purely coincidental.