Home > Free Sessions > Theatre Talks >

How to Design a Hardware Product from Idea to Market

Matiss Drusts - Watch Now - Duration: 42:37

Designing a hardware product from start to finish can be a daunting challenge. It requires various skills like circuit design and embedded systems design. The numerous potential pitfalls and points of failure are a major reason why many people put off developing their own products or why those that do try, often fail.

Therefore, in the case that you are looking to design your own hardware product or are working in collaboration with others, it is essential for you to understand how a hardware product is developed from beginning to end. This is what you will learn in this session.

  • How to design a hardware product from idea to market
  • General pitfalls to avoid
  • How to manage the economic of building a hardware product
  • How to design your hardware and software in parallel
M↓ MARKDOWN HELP
italicssurround text with
*asterisks*
boldsurround text with
**two asterisks**
hyperlink
[hyperlink](https://example.com)
or just a bare URL
code
surround text with
`backticks`
strikethroughsurround text with
~~two tilde characters~~
quote
prefix with
>

Bruce_Lueckenhoff
Score: 0 | 7 months ago | 1 reply

Mr. Drusts,
Thank you for a thought-provoking talk! Could you please share your thoughts on using System-on-Module approach?

(where the CPU, flash, DRAM and/or SRAM, and such standard things are provided by the System-on-Module, and a "carrier card" contains any custom logic, special interfaces, and so on)

MatissSpeaker
Score: 0 | 7 months ago | no reply

Well it depends on what your building and its application. If you need various functionalities and interchangeable modules then its the route you'll have to take but if you can avoid using system on module architecture then I would say that you should. Why, because it requires more design effort for a degree of added flexibility. Also a lot more failure points and time spent troubleshooting. Added cost of PCB and components. Making sure that the module and base board connections do not introduce noise artifacts or matching issues. The breaking of ground planes reducing signal noise immunity. That being said, if your building a complex product with desired flexibility then you may not have much choice, but for simple consumer products I'd avoid the complexity.

Pradeepa
Score: 1 | 7 months ago | no reply

Hello Matiss,
Great talk!
One point that I'd like to add for hardware/firmware parallel development is building a win32 simulator. It's going to slow you down first, but the ROI is huge. If your firmware engineers use strict abstractions for hardware developing a simulator means just building the firmware for your host PC.
Cheers!

piotr_zdunek
Score: 0 | 7 months ago | 1 reply

Do you think there is any benefit in using semi-agile management methodology in case of hardware development? I often see companies try to push for agile even for hardware development and it's often done forcefully, I wonder what is your opinion on that.

MatissSpeaker
Score: 0 | 7 months ago | no reply

So it depends. I don't want to say that its a remedy for everything. It works best if your developing a product. However, if your working in a consulting company and your client is Daimler who comes to you and say exactly what they want, exactly all the features they want, how long you have to make it and then hands you the money then I don't think its a useful strategy. Why, because they already know what they want, there is no iterative process, you probably will do a few spins of the board and that's it. The customer is already paying you thus there is no need to apply this methodology. Its more so for when your developing a full product from start to finish and are interacting with customers and looking for feedback along the way.

leandropg
Score: 0 | 7 months ago | 1 reply

What do you think of use 3D printing technologies for use in the Mechanical Hardware? These reduce the cost but is really useful in the process of manufacturing in mass?

MatissSpeaker
Score: 1 | 7 months ago | no reply

I think 3D printing is great for rapid prototyping in the early stages but not for mass production.

mariner
Score: 0 | 7 months ago | 2 replies

What is planned for today as the typical length of the life of a product? For business to business or industrial applications? For consumer market?

Steve_Wheeler
Score: 0 | 7 months ago | no reply

What you expect for product lifetime, and what is typical, may not be what you end up with. My company designed an SBC in 1993 that a customer designed into a product. Around 2010, we notified the customer that they needed to make a last-time purchase, because we were discontinuing the SBC. The processor hadn't been manufactured since at least 2004, and it and other parts used on the SBC were becoming unobtanium. They informed us that we had to keep making the SBC, because they had guaranteed to their customers that their product would have a 30-year lifetime. We managed to come to an agreement with them about how to move forward, but that blindsided us.

MatissSpeaker
Score: 0 | 7 months ago | no reply

The length of lifecycle of a product would vary widely depending on the industry that your in. For automotive and automation industries it could be as much as 5 to 7 years. For customer markets its usually allot faster, like 1.5 to 2 years.

leandropg
Score: 0 | 7 months ago | 1 reply

Hi Matiss... You suggest do customer interviews in the early steps of the creation of the product... You suggest make this interviews for example in Linkedin... How I protect my idea of that the other people copy my idea and execute? In this part of the process does not exist product yet

MatissSpeaker
Score: 1 | 7 months ago | 1 reply

Good question. But the answer is very simple. Nobody really steals unproven ideas. Its a common fear that many people have starting out but its just not based in reality so much as it is based in Hollywood movies and TechCrunch blogs. We all have ideas, they all have potential but potential is nothing. People that you interview have lives and other things to think about, they generally don't drop everything just because they hear an idea they find interesting. And what is the worst that could happen? somebody thinks its a great idea and forks it. Well they still need to build it. They are essentially no further than you are. There is really no need to worry about this at the begging since it will just get in the way of you delivering great product.

leandropg
Score: 0 | 7 months ago | no reply

Thanks Matiss for you advice

daleonpz
Score: 0 | 7 months ago | 1 reply

Hi Matiss, great talk, I really enjoy it. But I would ask you something. How would you estimate your development costs for industrial applications? Do you have some rules of thumbs or points to take into account? As you said, economics are really important, if not the most important.

Too add also. This book "Testing Business Ideas: A Field Guide for Rapid Experimentation" by David Bland,
is really great for learning how to test ideas. You mentioned many that can be found in that book.

And For HW simulation, some people use QEMU (which stands for „Quick Emulator“) with Docker containers.

Great talk again, I learned a lot :)

MatissSpeaker
Score: 0 | 7 months ago | no reply

Thank you, I'm really glad you enjoyed it.
The development cost of your product will vary based on features of the product and not really the industry your in (unless its something crazy, like space). The best way to estimate your product cost early is to make a rough design of the product like a POC or preliminary production design.

That being said...you don't need to spend days or weeks doing these. You can estimate costs on paper. list out the features, make a simple block diagram, identify major component(controllers ,coms , any sophisticated expensive components), which in turn will be driving the hardware cost.
The certificates needed will again depend on the components used and features of the product.

Thanks for the book recommendation, I'll check it out. I'd like to return the favor however. Take a look at Sprint, by Jake Knapp. Its geared towards software dev but offers very good insight that can be applied for any product and business.

Erwin
Score: 0 | 7 months ago | 1 reply

Hi Matiss, really nice thoughts about not treating hardware the same as software. I fully agree that your so called "semi agile" approach could be a beneficial process if established. Have you any thoughts on how to push a running project which works more like the "chaos" strategie (I like this term) to semi agile? Or do you think turn overs are nearly impossible?

MatissSpeaker
Score: 0 | 7 months ago | no reply

Hi Erwin,
Very interesting question. I haven't actually considered the scenario of moving a (currently running project) in to this framework. I would think that its NOT IMpossible. It would come down to assessing at which stage of development your product is in and planning from there. That being said, its more of a shift in thinking. If your developing alone, its easy but when your working in a team, I can see how there might be issues. But...If you can get your team to NARROW down in focus, and decide which stage you can tackle next then it could work.

leandropg
Score: 1 | 7 months ago | no reply

Hi Matiss... I am watching your video... I laughed when you say: "Strategies Development... 1) Chaos"... jajajaja... Is a good way to describe it

NicoPeper
Score: 0 | 7 months ago | 1 reply

Hi Matiss,
talking about avoiding typical obstacles...
you mentioned the obvious: developing hardware is often a staged development. How would you on the long run mitigate the risk about hardware availability (beside using off the shelf components)?

MatissSpeaker
Score: 0 | 7 months ago | no reply

Hi Nico,
In order to minimize the risks of hardware component availability I would first of all look at the components lifecycle, just to make sure it will be available for some years to come.

Another tactic would be to find suitable replacement components on the market, with a feature set very closely resembling the one your current component has. Often times the package types are similar or close to similar, and you can add pads for the replacement parts in the PCB design, that is if space is not an issue. We did this same thing recently with an LTE module. We had concern it might be out of production in the near future. The same manufacturer was about to release a new slightly modified LTE module, with part dimensions and pad layout almost identical to the one we were currently using. Therefore we had a replacement in mind, and just by adding a few extra pads on the board we could future proof the product.

OUR SPONSORS