The tendency of those specifying embedded hardware to dangerously flirt with employing commercial PC alternatives flares up routinely and usually correlates with the latest fads in retail computing technology. However, even outside of these “spikes” I continue to observe a worrying quantity of projects taking the very riskiest of gambles with their brand’s reputation.

I had at one point imprudently presumed that the days of needing to “sell” the benefits of embedded, industrial-grade computing implementations were well behind us; ostensibly they are not, hence my “back to basics” analysis for those considering taking that risk – those seduced by the lure of a quick profit fix.

Suitable? Disputable

Requirements matching can often be a box ticking exercise, usually focusing on functional and easily qualifiable elements – two USB ports, three RS-232 interfaces, et cetera. Satisfying the environmental parameters can easily be glossed over during this process, and our industry’s end products are rarely in an air-conditioned office!

These are harder to qualify as some aren’t quantifiable, for example how ruggedized an enclosure must be. Whilst most won’t, it’s possible that some commercial manufacturers do consider these quintessentially industrial parameters – for example drop, shock, and vibration protection levels – but how reliable is this information? Is it certified? Do these figures apply to the last unit as much as the first?

IP ratings are often incredibly important in embedded computing enclosures. Industrial manufacturers go to great lengths to qualify and certify their product’s assigned rating. Commercial alternatives rarely have a rating at all, or indeed one worth shouting about – if they do, to what level is this verified?

Given generally stringent approval requirements for embedded products, exhaustive supporting documentation has always been a prerequisite in our industry. Does the commercial alternative offer sufficient certification of conformity to negate potential legality?

Finally, electronic components typically have varying price-scalable temperature variants. Is it likely that a commercial designer, naturally looking to minimize his bill of materials (BoM) cost, will opt for those suited to your operational temperature needs?

Reliable? Deniable

Those selecting electronic components purely on cost rather than industrial suitability and reliability are those to be feared in the embedded sphere. Susceptibility to approve usage of grey market parts, combined with corner cutting in PCB material quality and inferior soldering techniques can combine to huge detriment to a product’s lifetime.

You could argue that a cheap commercial product’s prerequisite, beyond satisfying its functional need, is to survive just past the length of its warranty – why longer? If it lasts forever they’ll never buy a new one and achieving much more than the warranty is over engineering at the expense of profit – the bean counters would be outraged!

Embedded and industrial computing platforms are quite the opposite. Substantial care is taken to create a product with the longest working life possible and huge effort is made to prove this product longevity, as the consequences of these failing is not just a trip to the shops to purchase another!

What we’re discussing here is the difference of a primary focus on price versus one on quality – the quality of every single production unit. The “throw it away and use another” mantra is the inversion of our industry’s key performance indicators, or KPIs.

Reliability doesn’t just stretch to the physical product, what’s often even more critical is the reliability and continuity of supply. What availability guarantees can genuinely be made on commercial products? What commitment is provided to not change the product? Perhaps an irrelevant change for them, a mere upgrade of BIOS version can wreak havoc further down the line.

Any reliability issue may stop your product functioning, grinding production to a halt! Potentially the product remains fully functional, but often stringent industry approvals dictate that any change requires a full re-approval! Imagine maintaining a fixed BoM in such circumstance.

Risk

One can synopsize every aspect above in a single word: RISK. Making quick profits today risks customer satisfaction and brand reputation, decisions that can devastate a business financially.

When this eventually happens, the CEO won’t be asking for the employee who saved them a few dollars – he’ll be asking for the employee who shattered a revenue stream, costing the company millions!