Blog

How not to fail at electronic products

Time and time again we read about a hot new Kickstarter tech product. How could we live our lives without this latest gadget? Hundreds of backers and millions of dollars are raised, progress begins and enthusiastic posts come along. “Functional demonstrator ready!” “First prototype revealed!” “Developing into prodcution model!”…. then quite often the mood takes a cooler tone. Stories of component delays, design overspend, manufacturing partners letting them down all start to emerge.

Then there’s the compliance testing.

Slowly the cool mood turns to pleading, “Please bear with us!”

“We underestimated the technical difficulty” is one of the most common reasons that teams report running into trouble.

Even big companies fail at products sometimes. This isn’t about a product being a flop though – sometimes things go wrong even if a market is waiting for something. I decided to write this article because I believe the barriers to entry have fallen a long way, and developing custom electronics is no longer the preserve of those with hundreds of thousands of dollars behind them.

Developing electronics is the easiest it’s ever been

Thing is, on the surface, designing electronic products these days looks easy. So much is available from low-cost dev boards, free toolchains, modules, shields etc – that almost anything can be dreampt up and prototyped using cheap and widely available modules. This is an amazing way of seeing what works. Within a couple of weeks you can go from nothing to a collection of dev boards, some code – even a custom PCB design can be laid out in completely free tools like Circuitmaker from Altium or KiCAD by watching a few tutorials and knowing a little bit about board design. Then you can send it off to your favourite Chinese board spinner and have some professionally-produced boards in your hands for the best part of $50. Code can be spun up using tools like STM32CubeMX or the amazing mBed from ARM and before long, with a bit of care, you’ve got your very own bit of tech!

From this point, the team then has two options:

  1. Embark on the detailed design process with your team
  2. Hire a design agency

Option 1 is great if you have some in-house talent, or are experienced in bringing products to market. The teams that can turn a winning formula often do again and again – every time is a learning experience and it only serves to make things more efficient for next time. Hiring the right kind of engineers for a fast-moving startup can be a nightmare if you don’t even know how to judge an average engineer for a good one – and even great engineers might not work well in that sort of environment.

Option 2 buys you time with a pre-built team that has the right kind of experience and knowledge, which saves a lot of headaches, but it can be dauntingly expensive – and sometimes leads to being trapped with dwindling funds and an incomplete product. Maybe requirements are misunderstood? Perhaps the design team at the agency have their own take on how it should work, which is not really what you want? Maybe there’s a language barrier – especially acute if you’re trying to combine design and manufacture in the far East. Sometimes, however, they know things you don’t – for example, sometimes you just can’t get access to the same kind of hardware that the Chinese Tat Wizards use – so that gadget based on the photo frame keyring that you can get off Aliexpress for $3 is a nonstarter because nobody will sell the parts to you for that sort of cost. Or they might even seem reluctant to use a particular part because it wouldn’t scale into production – the dev board that you based all your code on can’t be made into a production part because it uses an obsolete microcontroller, for example.

The last little bit takes as much time as the rest of the project

Most clients I’ve worked with who don’t have prior product engineering knowledge are always hugely surprised at the cost and time spent on the stage between having something that seems to work OK, to having something that is ready for market – even if that market is only extended field testing. There’s a lot that goes on under the hood in terms of subtle changes that can have a great effect on a product’s cost to manufacture, reliability or just it’s performance. Even solving things like supplier issues and component availibility can be really draining on a project’s resources just as you’re trying to get to market.

Then there’s complience testing – this can be an expensive journey for the unprepared, especially if you’ve got radios on board – and what modern gadget is complete without a bluetooth connection or cellular module?

So what’s the answer?

There’s no one-size-fits all answer, and you could be frustrated by all the options that seem to get you 90% there. But if I had to pick some pointers, they’d be:

Do what you are best at.

You’re developing a product because you have an idea, you’ve researched the market and you’re getting interest. You’ve attracted attention because you did an awesome campaign. You are fantastic at marketing and have good business sense. Stay doing that. Don’t get bogged down in technical details, or at least, not more than you need to get the picture. Don’t become an engineer. If you are an engineer already, great, but if you’re running the campaign too try and keep that business head above the weeds!

Trust the experts.

Not everyone is trustworthy, and everyone is tired of experts these days, but if you can find a guru (industry term), then hold their advice as golden. You’ll know who they are. They’ll have seen the mistakes and pitfalls, and may have made many themselves. Maybe they’re a bit cynical, perhaps they’ve seen loads of similar things fail – so what? Harness that. What do they think a success would look like?

Iterate fast, fail fast and often.

If something isn’t going to work out it’s better to have spent 3 weeks on it rather than 18 months. Running iterative development cycles used to be the preserve of the best funded tech companies – those that could afford multiple engineering teams working in parralel to thrash out prototypes that explore all the possible avenues of design. Modern, collaborative tools allow this kind of work across teams that are multi-disciplined and multi-location. You can work on designs live, and keep sending making tweaks to software, mechanical, and more importantly, hardware within a couple of weeks. A problem is identified, a solution is tested, a design is updated, CAD is regenerated and send off. $50 and 4 days later new boards come in. components were ordered for 2 or 3 to be built. Boards populated and reflowed that afternoon. Day 5 and we are running code on a completely new design.

So what do we do next?

For a design owner reading this, what can you do next? My solution is to keep moving – stand on the shoulders of giants. Learn the tools. Watch the youtube videos from the gurus. Read the forum posts – but, more importantly – get the stuff built and start failing!

Torbett Design can also help you. With years and years of experience of everything from rough prototypes, proof-of-concept, technology feasibility studies through to the handle-turning of that fast iterative design we are more cost effective than hiring a full time engineer, or can work alongside your existing team to resolve capacity issues, reduce bottlenecks and apply experience to get things moving.

Drop me a line at james@torbett.co.uk or call +44 7968 030306

Torbett Design gains Cyber Essentials approval and DSTL R-Cloud supplier approval

Torbett Design Ltd is proud to announce its approval to the Cyber Essentials IT security standard by the National Cyber Security Centre and APMG International.

This certification means that the IT infrastructure and practices have been evaluated and found to meet a minimum standard for the security of information and the protection against cyber attacks.

DSTL R-Cloud

Torbett Design has also been approved as a supplier to DSTL for defence research in multiple areas. This adds us to an important SME framework and allows direct engagement with DSTL for projects in the scope of R-Cloud. For more information, see https://www.gov.uk/guidance/r-cloud

The Internet of Good Things

In a couple of days time I am heading back to South India for the first time since I worked there over 10 years ago. Back to one of my favourite places in the world, Kerala.

The backwaters of Kerala

In August, parts of the state experienced the worst flooding for nearly a hundred years, brought about by extremely heavy monsoon rains. Dams were at risk from being overtopped, and the necessary release of water made things worse downstream. 

The immediate humanitarian disaster and resulting relief effort was well-covered in the media, and climate change was of course highlighted as a probable factor. But what is less visible is the ongoing economic damage due to massive destruction of crops.

Related image
Flooded crops in August 2018. Image: AFP

India, and Kerala in particular, is a state of small-holder farmers who operate in co-operatives to grow and sell their crops.

When you buy tea, coffee, or spices like cardamom, turmeric or saffron from a huge company like Tesco, who in turn buy it from a giant company like Tata, they actually buy it from a guy who lives in a tiny house amongst his farmland and sells along with his neighbours to the distributors. They of course, get the thin slice of the economic wedge, and are very vulnerable to the effects of Climate Change. But things are changing. With the internet, trade deals can be made from every corner of the world, prices of crops negotiated and data trends used to demonstrate future viability.

I am working with startup Climate Edge, a London-based climate data consultancy that specialises in smallholder farming, to leverage the power of connected devices and the ubiquity of the humble 2G GSM network to open up the power of real-time climate data trends to the smallest of co-operative farmers.

An early NEXO prototype. Image: Climate Edge

By using low-cost data gathering tools that can be easily installed and maintained by the farmers, devices like this can empower them with the data to bolster their market position, and advise them on mitigating the effects of Climate Change by making small adjustments – for example, improving drainage when trends show increasing rainfall, planting shade crops in areas that experience less cloudy conditions or better suiting planted varieties to the local temperature trends.

The key to this enabler is not to force the small-holder to foot the cost of the infrastructure and data analysis. Although they are the obvious customer of this climate data, they represent the product. The customer – the real customer of their products – are the large international buyers, who are the ones buying the data analysis, and subsidising the system.

NEXO PCB. Low-cost, simple and robust design philosophies are used

Torbett Design has provided services to Climate Edge by developing the NEXO weather station hardware and firmware. The NEXO utilises a low-power ARM Cortex processor interfaced to the various internal and external climate sensors. A local RF link allows additional sensors such as rain gauges to be added without wires. GPS is used to locate the unit accurately on maps, and automatically update its location should the unit need to be moved, without the farmer having to survey or deal with mapping tools. Finally, a simple 2G GSM link uses a global SIM to transmit small but hourly data updates on any local cellular network. Should there be a cellular outage, the unit can store several months’ data for automatic upload when it is able to connect. A small but robust solar panel and charge management system keeps two generic Lithium 18650 cells topped up, these batteries are available globally should they need replacing. The entire system is built on a simple two-layer PCB which can easily be fault-found and repaired if necessary to reduce electronic waste.

 

New Lab Equipment

To add to my slowly growing lab capabilities, last month I acquired a Fluke 8840A 5.5 digit multimeter and a FeelElec FY6800 60MHz digital signal generator / counter. The Fluke is a second-hand device I found on eBay, and although it looks like it’s been well used it’s in good working order and still in calibration. I’m primarily using it to make low-voltage measurements across shunt 0.1Ω or 1Ω resistors to determine power consumption of battery-powered systems. With modern processors, it’s not unreasonable to expect average consumption in the low micro-amps if sleep modes are used correctly. In fact, quite often processors that might seem more power-hungry, such as low-end 32-bit ARM devices are cheaper and lower power than “classic” low-power devices such as 8-bit micros.

The Signal Generator is used to create test waveforms of particular frequencies, duty cycles or shapes to provide external stimulus, for example to simulate an encoder coming from a motor. Buying it was a bit of an experiment as it falls under the category of cheapy chinese equipment at around £99, but having watched some teardowns and reviews of the device online I am satisfied that it will do just as good a job as my go-to alternative, which is to program a dev board to produce stimulus waveforms. They’ve made some interesting, but not entirely bad design decisions in the device, not least creating a DAC from a huge resistor ladder driven by a high pin-count FPGA! As for long-term reliability, well my lab is not the most demanding of environments so I am sure it will last well enough. Except the Fluke, which I expect will out-live every other piece of equipment!

Fail fast, not hard, to succeed.

Too many technology projects fail because they try to run before they can walk. Tech is hard. Hardware is harder. We exist in a world where we are now surrounded by things which make the complex look deceivingly simple, and in bringing your technology idea to life it is all too common to shoot for the stars – create something beautiful, solid, functional…. expensive. Fail fast. Use rapid prototypes to find out what the market really wants, what features work, what doesn’t. Before you dive into the deep waters of engineering for development, certification and mass-production, let us take you on a journey of discovery of the art of the possible. Using years of experience, and standing on the shoulders of giants, so to speak, by utilising existing solutions and rapid development tools together we can create prototypes to test your ideas, markets and business model without breaking the bank.

No more shiny renderings of beautiful designs with no guts – create real, functional things in short spaces of time that prove out concepts, test markets or de-risk investment pitches.

Get in touch for a chat about your technology ideas, and let’s make them a reality.