Stratio

8 things I hate about hardware

No, it’s not that I hate hardware, I just needed a catchy title for this article, bear with me and give your opinion in the end, I’d love to hear from your experience in hardware development.

I have been working with hardware development for over 20 years and I think I’ve earned the right to grumble a bit about the struggle it is to get a new product out to the market.

As the Head of Hardware Development and Embedded systems at Stratio Automotive I will share with you the most common topics hardware engineers have to deal with in the path of bringing great products to life, so here is the list of the major 8 steps a hardware development team will have to accomplish.

1. Understand what you need to create to satisfy your clients, hearing the main stakeholders with a firm attitude on limiting the complexity. This is true for all companies and projects but gains a special emphasis in a company with limited resources. Wrong identification of target clients or misinterpretation of their real needs may create a fantastic product that no one will buy. Complexity can also be the cause for the failure of a project. We know a relatively limited team can elaborate the architecture of an extremely complex equipment, however its execution, the adjustment of the details (where we usually find the devil), will make all the difference, putting into practice and complying with all legal standards can become so difficult and time consuming that jeopardises its viability. Moreover, the costs associated with tasks to adapt to industrial processes, such as the creation of highly complex PCBs and very elaborate enclosures, above normal industry standards, will put significant obstacles in the success of the project. Rule of thumb — Identify what is really needed and distinguish it from the accessory, MVP approach, where additional features must be very well weighted and considered. The “fish diagram” below is a visualisation of my opinion, the value perceived by the client will increase greatly with his essential requirements usually simple and inexpensive, as the requirements get exquisite and less valuable for the client (logarithmic curve) the cost to implement them escalates almost exponentially.

Fish Diagram

The sweet spot is somewhere in the middle, an over simplified design will lack important functionalities, an utterly complex design will heavily raise the manufacturing costs and consume the profit margins.

2. Study the state of art of all the components and modules you are about to use and choose the reliable. It is frequent to be inclined to use newly launched technologies thus creating future proof equipment. Again, this selection shall be made after a serious validation to make sure the devices or modules are mature enough. If a component is not mature or extensively tested, you will be helping the fabricant to write their errata or finding issues to correct in the next FW releases. Once your product leaves the factory it will be complex and risky to update modem, Wi-Fi, or Bluetooth modules. The second problem is related to new systems or trends that may not be adapted by the general industry which will leave us with a limited product. A practical example of this: Three years ago, we decided to use the LTE cat M1 modem module that unfortunately still does not have network coverage in many countries like Portugal, Italy and Austria which adopted NB-IoT to replace GPRS in M2M communications.

3. After having identified all the systems that we will include in our hardware design, It is now time to integrate them. Here we have different equipment that may not have been designed to operate together and we need to find a way to get them to work smoothly. How do we do it? We read a lot of datasheets and we do our best to understand every detail on how that component or module will work. Electronic components datasheets are an incredible source of information, it is frequent to find a microcontroller or a modem module with thousands of pages in their datasheets that need to be studied to clearly understand the operation of each component. Components should be selected to work at 50% or less of its design capacity or you will be accused of planned obsolescence, by setting the quality standards high you will get a good life expectancy in an electronic circuit. Never overlook the errata section, if you do that you will probably waste precious hours trying to debug issues already found by others. If you are writing embedded code, pay special attention to the IDE or compiler limitations, according to the semiconductor selection, this is even more important if you use a free or “open source“ IDE.

4. Once the schematic is ready it is time for the PCB design, in this section special attention must be focused on power supply paths and ground, high frequency signals, and then Footprints! The footprints need to be discussed with your PCBA company, specially the smaller pitches, we can design any footprint and any via, but some will be very expensive to implement in real life. Special vias like blind or buried impact time and yield, so adds cost. An overlook here will cost you large amounts of money in the production of a PCB with laser drilled holes instead of mechanical drilled ones.

5. So, you have completed the PCB design and produced a small batch of PCBS that will be used to build the first prototypes. Here you’ll find another barrier, many assembly companies will be available to assemble a pair of units for you, for the cost of a kidney. Manual assembly is only feasible for low complexity PCBs, anything having reduced pitch, BGA will need a complete pick and place assembly line to be set up (same cost to produce one or a million devices). Components will also be twice the price when compared to the price per unit if you buy 1000 units, we must live with it! If you don’t buy the components and just send the BOM to the assembler count with negotiating some alternative or compatible components suggested by the assembly company, don’t accept the suggestions without reading carefully all specifications, some components seem equivalent but may have temperature ranges or tolerances out of your requirements and result in an imperfect prototype.

6. Congratulations, you have your prototype built, it is time to evaluate it, load your firmware. Developing FW, programming and debugging is usually an endless mission only limited by your level of satisfactory perfection. During this stage it is also very important to start running some EMI pre-testing according to the desired usage and regulation compliance. One advice here, don’t send to production new PCBs once you find and correct the first design fault, try to patch it and continue the evaluation, we always find some more bugs in the way and the prototypes are expensive and time consuming to produce.

7. Design and create an enclosure. Here we have the choice of using off the shelf enclosures (plastic or metal) or creating a new enclosure. The most common is the last and the preferred material is some sort of plastic. It all starts by finding a competent industrial designer that will create an enclosure ready for mass production by means of plastic injection. A 3D printed prototype is nice to have to validate physical dimensions and overall looks but you can’t depend on 3D printed enclosures if you plan to produce thousands of units. The required moulds for injection are an expensive initial investment but they reimburse for these numbers, not to say that the overall quality and aspect is much better on an injected plastic enclosure.

8. Time for Certification. It is an important, expensive, and time-consuming process for any company creating a new product. To start this step, one must understand clearly which are all the certifications required for the product itself, according to technical characteristics and it’s end use multiplied by the number of countries where you want to sell your product. Special attention should be given to intentional electromagnetic radiators and products including Lithium batteries, expect higher costs if your product ticks those boxes. Safety and environmental certification are also required for most products prior to being sold. In short all testing and certification requirements should be considered from the beginning of the design of the new equipment, but the certification is only conducted in the final stages of production, the reason is simple, a change in the product would require the restart of the certification process.

Some of the issues may seem obvious, others not that much and to be honest this year’s new normal brought some more challenges to our bowl. The sourcing of components is getting harder day by day, the prices are also reaching historical maximums and the lead times are in the impossible to cope area, most components ranging the 10 to 20 weeks microcontrollers and Tantalum capacitors easily reach 30 weeks.

I hope this article has given you a simple overview of the process of developing a new electronic product. Your development strategy may be different, but in the end we all want to reach the same goal, achieving a robust equipment that satisfies the clients and strengthens the company.

The acronyms:

BGA — Ball Grid Array, a type of surface-mount packaging

BOM — Bill Of Materials

EMI — Electromagnetic interference

FW — Firmware (executable code)

HW — Hardware (the physical equipment)

IDE — Integrated Development Environment

MVP — Minimum viable product

PCB — Printed Circuit Board

PCBA — Printed Circuit Board Assembly