I went back to my last embedded project and tallied the hours and code, and the result was 50 lines of assembly-level code per 8 hour day, and this includes the algorithm development. This was a small medical device (2100 lines of code not including LUTs) where I did both the H/W and S/W. The S/W was real-time in nature due to a feedback controller.
So I think there are several caveats to the 200 lines per month number. It's no doubt accurate on large projects with a random assortment of programmers, and I find that just the problem of having multiple entities involved bogs the process down. In my example I had intimate knowledge of the H/W since I designed it as well, an advantage never afforded the typical project. But a factor critical to assess is the motivation and technical ability of the individuals involved rather than the use of a fixed line/day estimate; in fact I'd argue that using the smallest (carefully chosen) S/W staff possible has a significant benefit toward minimizing the development time.
Great point about agile development practices, William. I'm hearing more and more about those as I start to dive into cloud-based design tools and mobile apps. It's definitely a mind set change. Hopefully, it's a practice being woven into new curriculum to help engineers get their arms around how to do it adeptly.
Beth: As a former hardware guy, I feel comfortable saying this: Those of us who were educated in mechanical/electrical in the 1980s or earlier are have more trouble bridging that cultural barrier that you refer to. We've been trained to think in terms of hardware, and it's a tough habit to break. Maybe it's a bit of an ego thing, but it's tough to say, "Yes, software is the most important part of our product."
Charles, really good article and lots of good points!!
I know and have been on those projects that seem to keep going and going and begging for someone to put them down. Many good projects have over run budgets or just failed due to poorly written or no requirements or requirement creeping or engineer's free give aways.
If I had a .01 for every time I've heard an engineer say I can do this better and it would be nice to have this extra feature and just write it in without really understanding it is out of scope and most likely budget, I would be a millionaire.
I remember having to fight management to purchase good lab equipment to test our designs. It cost the company more money and the designers unnecessary time due to fighting with management to get what we needed to get the job done.
Great article, Chuck. You have hit many of the major pinch points in embedded development.
I like your first point "It's all about the software" and your second - "Software is a relatively new field... and we're really still figuring it out". The difficulty arises from Moore's law -- with every 18-months the capability of each component doubles as we continue to innovate.
Five years late with their product? That's more than three generations out of date before the team launched their first product.
If the development team decided to use traditional "requirements-based" project management, they need to freeze the requirements and build to the original specification. But as most teams find out quickly, even if they don't change requirements for new technology, the requirements change because of unanticipated difficulties and incorrect assumptions made by the original designers.
May I suggest an initial "Zeroth Point" on your list: "0. Use an Agile Software Development Method". Once the team develops using Agile methods, the subsequent components on your list are still important, but can be altered with less difficulty. The smart-phone industry is one of the best examples of successful Agile development. Pushing software updates to the devices, new models with new features arriving often, and using standard ports and interfaces so upgraded devices can be swapped out with relative ease are slowly becoming the norm.
Great job, Chuck, highlighting the common misperceptions and challenges associated with embedded development. The whole notion that the development effort is predicated more around software design, not hardware design, is a huge cultural barrier for many engineering organizations which may lack a deep pool of expertise in that area and for years, have priortized and emphasized non-software related development.
The tools are also a huge issue. You talk about the need to invest in development tools specifically around writing embedded code. There also needs to be an investment in tools that integrate the embedded development effort with the rest of the product design effort, both mechanical and electrical components. If all the work is done in silos, you can run into intercompatibility problems and design snafus late in the cycle when it is expensive to make changes.
The question of whether engineers could have foreseen the shortcut maintenance procedures that led to the crash of American Airlines Flight 191 in 1979 will probably linger for as long as there is an engineering profession.
More than 35 years later, the post-mortem on one of the country’s worst engineering disasters appears to be simple. A contractor asked for a change in an original design. The change was approved by engineers, later resulting in a mammoth structural collapse that killed 114 people and injured 216 more.
If you’re an embedded systems engineer whose analog capabilities are getting a little bit rusty, then you’ll want to take note of an upcoming Design News Continuing Education Center class, “Analog Design for the Digital World,” running Monday, Nov. 17 through Friday, Nov. 21.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.