The future of development tools

Likewise, requesting a sample of a device is an effective way to get your hands on new or existing technology. The life of a design is finite, the faster a supplier can provide samples to their customers, the greater the potential for revenue over a longer period of time. Today, many semiconductor companies provide sample items to their eCommerce distributors and rely on them to ship samples to their customers. This means it is faster to simply buy one’s samples from one of those distributors like Digi-Key, Mouser, or

Farnell to name a few.

In doing this, the chip companies leverage the part management and logistics capabilities of their distributors and, in most cases, the samples are shipped within minutes of receiving the orders from their customers as opposed to days or even weeks from the supplier directly. Everyone wins. By outsourcing sample deliveries, the chip companies save cost and make their technology easier and faster to obtain by the widest swath of customers. The distributors benefit from more business. Their shipments increase to more customers which enables them to negotiate better shipping rates with their shippers and customers benefit by getting their parts sooner.

Changes

Another area that I predict will change revolves around development tools. Semiconductor companies generally believe that maintaining proprietary development tools is in their best interest. Their argument is that a development tool represents a high switching cost to their customers. Having a customer indoctrinated to one’s tools means the customer has taken the time to learn, often times, complex and intricate application programming interfaces (APIs), code libraries and bespoke integrated development environments (IDEs) that a semiconductor company has developed. In fact, I’ve met zealots of specific proprietary development tools and I might be called a zealot myself for my love of the good old MAX+Plus II IDE from Altera (now Intel).

Few developers like wasting time to learn something they do not think they need or will use. So, if a known development environment satisfies nearly all of one’s needs, semiconductor companies reason that customers will not arbitrarily switch from them and will thereby remain loyal to that semiconductor company. For this reason, semiconductor companies are unlikely to ever align on a standard IDE solution that all of them use. This is probably mostly true with programmable logic device companies - the microcontroller camp is less proprietary.

Above: Feather board standard from Adafruit

Nearly all new microcontroller devices introduced today are based on ARM or RISC-V cores and not much else, and most of their IDEs are based on Eclipse that they customise to make into their own IDE. Even though these manufacturers love the idea of locking a customer into their platforms, their platforms are pretty much restricted to the same features and interfaces. In reality, it is not that hard for a customer to switch from one of these semi-custom IDEs to another. It’s the libraries and code standards that make switching more difficult. Also, since these companies are semiconductor companies and not development tool companies, it is hard to find one with a truly visionary outlook or philosophy on development tools. I once worked for a semiconductor company that had such an outlook at one time. That company was Texas Instruments.

I worked for Texas Instruments when it transitioned from its modular port scan device (MPSD) architecture to what was first known as the Joint European Test Action Group (JETAG) which later became simply, Joint Test Action Group (JTAG). I was told the ‘E’ was dropped because Texas Instruments was American. MPSD and JTAG introduced the concept of boundary scan and it was a revolutionary approach to development tools. It changed the way microprocessors and microcontrollers are designed and how the systems based on them were developed.

Pushing the boundaries

Today, another type of company is pushing the envelope of development tool capability. They are doing this by focusing on making their customers more efficient (i.e. less cost to produce an output) and productive (i.e. more output per unit of cost) at producing their end products. Working separately, these companies are each prolific in their development of standards but they make their standards open to encourage adoption by others. Collectively, I see these new development tool companies as an industry in and of themselves.

This new type of development company is largely born of the maker and hobbyist movement. To make less skilled or less educated developers productive, they have rethought and continually reengineered the development process flow. This industry feeds makers and hobbyists with very straightforward and easy to use tools. Let me name some standards they’ve established: Uno interface (Arduino), Feather boards and ‘Wings’ (Adafruit), Circuit Python (Adafruit), Grove boards (Seeed), and impressively Click boards (Mikroelektronika or Mikroe) whose variety of board functions number around 1,000 and there are still other standards. In fact, some of these standards are supported by semiconductor manufacturers who accommodate these standards on their evaluation boards with its connectors and interface signals.

You might be thinking that these tools are for amateurs and most of them are but not all. A distinction I made in my embedded world 2021 presentation is that makers and hobbyists specify what to buy for themselves, whereas professionals specify what their employers buy. This has nothing to do with the tools used. There are many professionals using the new company standards, tools, and products.

Still, if you are a professional, the expectation of you is to produce a nearly cost-optimised design - one that finds a proper balance between cost and functionality and this is where some of the tools I named do not totally fulfil the needs of the professional - but not all. To adopt a standard, one does not want to break it by circumventing some aspect to get one last measure of needed product performance. For example, there is no true real-time component to the Arduino platform, although I recently saw an ad for a company that makes an Arduino with programmable logic that enables the developer to run in hardware as opposed to software.

Above: Homemade pressure recorder for insertion into a Bulldog’s trachea

Speaking of software, the point of my embedded world 2021 presentation was to call to everyone’s attention that not all problems are software problems. All but the programmable logic companies’ tools focus only on software and not the integration of off-chip peripheral hardware.

Focusing on the blend of on- and off-chip hardware and software is the development tool paradigm of the companies in this new development tool company industry. These companies are application focused without being experts at anything more than integrating components into systems and this is key to developer productivity. They see software as the glue that integrates the hardware.

An Adafruit feather board, a pressure sensor, and some 3D printing was used by a veterinarian to develop a $50.00 air pressure recorder that is used to measure the pressure inside a bulldog’s trachea. Because all the hardware was readily available, all the veterinarian had to do is modify some simple software that he found online. I mention this only to point out how simple application development can be with the right tools. In many ways, the amateur tools make one more productive than the professional tools.

However, I would be remiss to convey that tools are either amateur or professional. There is a spectrum of users. Many professionals use amateurs’ tools and, I am sure to a lesser extent, many amateurs are using professionals’ tools. However, I want to call your attention to tools that are in the middle of these markets.

I mentioned Mikroelektronika earlier. While they invented the ‘Click Board’ concept that many large semiconductor companies have adopted like Microchip, Renesas, and others, this concept is based on their ‘open’ MikroBUS standard. This standard specifies the electrical and mechanical aspects of these boards. In addition to the MikroBUS, they also established a MikroSDK which specifies a hardware abstraction layer (HAL) to any board that accepts Click Boards.

In addition to this, they developed their own IDE that allows a developer to switch from one CPU to another and they foresaw and developed the ability to debug remotely using a WiFi connection to any other of their target boards wherever they are in the world.

They’ve developed this capability for some semiconductor companies. This is powerful stuff that we are not getting from the semiconductor companies themselves.

These development tool companies have a vision for the future of development tools. They are profitable as development tool companies; this isn’t a side-line operation. They fund their ongoing business by marketing to and supporting their own customer bases and, at least Mikroelektronika supports the semiconductor companies who have hired them to develop their development tools.

I ask, why should non-professionals have tools that make them more productive than professionals have and why must professional tools be subsidised by the sales of the chips they were developed for? I am seeing a trend in semiconductor companies beginning to outsource their development tool development. I predict this will become something like fabs and samples and be done by those whose only focus is development tools.

One client told me they have all the development tools they need but not all the tools that they want. Outsourcing development tools enables a semiconductor company to focus on where they add value and that is to augment those ever-evolving development tool platforms with hardware and libraries targeted to the applications they developed their chips to serve, while also reducing the time to market for their customers.

Randall Restle is a lifelong semiconductor professional who early on worked for Texas Instruments and most recently for Digi-Key Electronics as their VP of Applications Engineering.  He has now established a consulting firm where he helps technology companies improve their product development and market development plans including technical marketing.  His clients have mainly been international firms wishing to better penetrate the North American market.  He can be reached at randall@restle.us.