10 Insider Tips from a Hardware Design Startup
After battling pandemics, component shortages and a myriad of technical challenges in domains that range from wearables to automotive, Asaad Kaadan, Founder and CEO of Hexabitz shares 10 insider tips on leading a hardware design team that survived the rollercoaster of the last two years with more than 4,000% of revenue growth.
This article appeared first on Hardware Pioneers website on 12 AUGUST 2021.
1) Hardware is hard, get over it
It’s the truth, so you better accept it and embrace it in order to survive and thrive in your hardware business. The internet is full of stories about software startups’ hefty bottom-line margins, magnificent growth rates and agile development magic.
The truth is, hardware startups need significant capital investment to launch, have usually lowered growth rates and building physical things just takes much more time than compiling lines of code. This does not make hardware startup success any less meaningful than its software peers.
In fact, hardware can be easier to monetise once you figure out your Product-Market-Fit (PMF) as we found out in Hexabitz. We had to rack in serious credit card debt to survive the first few years. Once we found our PMF, it took us only 17 months to reach 3,000% actual revenue growth, pay off all our debt and invest in our growth, all of that with zero external funding.
2) All your market assumptions are wrong
I usually hate generalisations but this is the only statement that I might consider to generalise! It’s almost always the case that your initial market assumptions are wrong. It happened with us in our own products and with quite a few of our customers as well. I’m not an expert in market analysis but I can attribute this to a few factors:
- Founding team (usually technical) is inexperienced with quantitative market analysis and value research.
- Inherit personal bias to one’s own ideas. This is particularly true in consumer electronics. What is favourable for you, might be okay or even negative for someone else.
- Hardware products take a considerable time to launch, on the scale of a few months to a year or more for complex products and this is only for the MVP. Usually, the market conditions might have changed by the time you are ready to go to market. One notable example here is the multitude of startups that launched products related to the COVID-19 pandemic. Many products did not hit the market at the right time and by the time they were ready, the world had already moved from prevention to vaccination and to treatment.
This does not mean you have to give up on your products or ideas! But be prepared to change course. Your product might work well in some other market you haven’t even considered or it might need few (or lots) of modifications to become a hit. Regardless of what you do, when the market speaks, you have to listen.
3) Have domain experts on your team
Electronics and PCB designers tend to think they can carry out any hardware design project in any domain they get exposed to. This is true for simple and textbook IoT projects.
However, once you dive more into actual products, you will probably need domain experts to help you avoid costly mistakes and understand the domain you are dealing with. It’s often the case that certain regulations, industry know-how or domain minutiae will affect the choice of electronic components or manufacturing process.
If you are building a PCB for a medical device, you better have physicians or biologists consult for your team. If you are in the entertainment industry, you better have some audiophiles listen to your music!
4) Pandemics are not always bad
There’s no questioning the dire health and economic consequences of the COVID-19 pandemic worldwide. However, challenges breed opportunities. We had an uptick of customers in 2020 with about 30% for pandemic-related projects. It’s not just the social-distancing trackers and disinfection machines though.
When people spend significant time in quarantine, they tend to innovate more, invent new projects or come up with new initiatives. The world was in constant-hustling before the pandemic hit. Hustling is good but sometimes you need to self-isolate and rethink how to solve the world’s most-pressing challenges.
5) Expect delays and then more and more and more delays
Delays are the enemy of stakeholders and product managers. They are unavoidable, though, in the course of hardware product development and you have to learn to live with them. Pandemics happen, stocks get depleted, factories close, shipping gets delayed and ports and canals get closed (remember the little digger rescuing the massive ship in the Suez canal?)
The challenging part here is not just cushioning your product planning with safe margins here and there. It is, based on my experience, being able to communicate the situation to customers, investors and stakeholders.
Outsiders don’t usually appreciate the complexity of modern-day supply chains and when you try to explain, they will give you that strange sceptical look like you’re trying to scam them!
Why can’t I get my overly-complex, custom-designed hardware in two weeks? PCBWAY ships me PCBs in three days! Because designing and building actual hardware products is a complex process that takes teams of engineers months of design work, supplier negotiations, shuffling components across the globe and multitude manufacturing technologies topped off with software development, testing and validation.
A hardware product is not simply a PCB!
Common questions that crop up include:
- Why do I need to ship components from the United States to China to manufacture them into PCB assemblies – aren’t these already made in China? (Yes but you can only buy them in the United States).
- Why do I need to ship my hardware from China to India or Eastern Europe or the Middle East to develop its software – why can’t you make everything in China? (Because PCB fabrication houses don’t do product development).
- Why can’t you build everything in the US or Germany? (Because I have to bill you 10x and you already think it’s expensive!)
6) You don’t need to understand everything at first
This might be difficult for perfectionists but startups are not the work of perfection! They are the work of efficiency. Many times your customers don’t know what they want or why. They just have a vision and it’s your responsibility to make it a reality and to help them understand their own vision.
This is particularly true when developing novel products that never existed before. It’s an exploratory process that involves research and learning. You cannot perform scientific research starting with predetermined outcomes.
This type of product development that blends research and engineering together is the most-challenging but usually the most-rewarding work you do. At Hexabitz, we pride ourselves that about 50% of all our projects fall within this category.
7) EE and ME should talk from day one and EE and SW should talk before building prototypes
This is particularly true when the form factor is important, such as when you’re building wearables or consumer devices (or pretty much anything these days that does not sit inside an industrial cabinet.)
Present day electronic devices are complex, multi-board, three dimensional structures that require careful mechanical integration with the enclosure. Delaying meetings between PCB and mechanical engineers might result in PCB redesign causing unnecessary costs and delays.
Sometimes enclosure design, material or user experience dictates the selection of components and their placement. Modern consumer devices and wearables are full of rigid, flex and rigid-flex board combinations with complex folding structure and manufacturing processes. State-of-the-art CAD design tools offer tight integration between mechanical and electrical design.
Another, often forgotten, coordination meeting must happen with the software team. Software (or firmware) is usually an after thought in hardware product development because software can easily change without extra cost. This is not an accurate statement however, because time is money.
Sometimes, minor hardware adjustments might result in big performance improvements or time savings for software development. It doesn’t hurt to trigger discussion as early as possible and definitely before placing your prototype order.
8) Start with building a dev board
We all have the tendency to start prototyping with final designs that are as close to the end-product as possible. We are eager to watch our beautiful creation getting made and feel it between our hands. This is true for simple products.
In fact, I always advise to prototype using the final form factor for simple products and PCBs because there’s a high chance it works as-is, but most importantly, because designing for the final form factor will teach you multiple important lessons on manufacturability and the practicality of the design itself.
Complex products are a whole different story on the other hand. Don’t prototype a complex PCB with its final form factor. You already have enough technical risks and unknowns in the design itself. Adding form factor complications will exponentially increase design and manufacturing challenges and cost you more time and money in the long run. If you have a complex flex PCB, prototype it first with a rigid one.
If you’re dealing with a new computing platform, build your software team a development board if one does not exist off-the-shelf. This will speed up the learning and development cycle while waiting for final boards to be manufactured.
This should not go down to bread boards though! In complex systems with high speed signals, it’s usually impractical and a complete waste of your time trying to breadboard your way in. You have to build your own PCB that can carry the high speed or high power signals and present you with capacitive coupling, interference and all other real-life annoyance that comes with the end product but it doesn’t have to be the complete system.
9) Think about automated programming and testing early on
If you plan to produce even just a dozen of your boards, it is worth it to think about automated testing and programming fixtures early on. This boils down to deciding the nature of these fixtures (POGO pins and clamps, connectors, etc.) and deciding on locations of these connectors and testing/programming pads even if you don’t want to assemble the associated circuitry on your board or build these fixtures.
Sometimes your one-off project becomes a great success and you find yourself making hundreds, if not thousands, of these PCBs. You will thank yourself later if you had already given a thought to automated testing and programming instead of redesigning the board and making another set of prototypes! Sometimes the design is so complex that adding any set of testing circuitry after that fact becomes a major redesign and a big headache.
10) Don’t blindly trust reference hardware designs
We all love to look up some neat reference designs and just copy them! (The equivalent of programmers copying Stack-Overflow answers into their code). This is a great timesaver and you are strongly advised to follow professional reference designs especially if you’re entering a new domain.
However, you should not follow these designs blindly! I don’t want to question the integrity of semiconductor OEMs. I’m pretty sure their talented engineers and product managers have put a lot of effort into their reference designs and evaluation modules.
However, we had multiple instances where reproducing designs from respectable companies failed to achieve the advertised results. Sometimes, your testing setup is not exactly matching and the OEMs will provide you with best case testing scenarios. The design could also be highly sensitive to small variations in PCB properties, routing or even environmental conditions making its integration into your overall project a challenging task.
In summary, put your trust in high-quality reference designs but have some cushions in your planning for a redo.
