How We See Embedded Processing

Apr 30 2007 10:30AM | Permalink | Email this | Comments (14) |
Blog This! using:  Blogger.com | LiveJournal |
Digg This | Slashdot This | add to Del.icio.us

If the relationship between "market penetration" (forgive me, my last job was in marketing) and the cost of the development boards is so simple (inverse proportionality?), every engineer worth their salt would quickly jump to the conclusion that such tools should be given away for free. Mathematically, there is no doubt about it—but the reality is quite different. It's human nature: when something is given for free, its "perceived value" is also reduced to zero. In my practical experience, free development boards are not trusted and most often left on the shelf to accumulate dust.

So, short of giving the development boards away, how can we define and justify any other arbitrary minimum price target?

Needless to say, this is a frequent topic of discussion at Microchip. It comes up almost as frequently as our design teams introduce a new product and, with it, the need for a new development board. Most embedded-control manufacturers seem to have struggled with the same question at one point or another, sometimes with comical results. In fact, in the desperate attempt to make the tool inexpensive they have sometimes compromised beyond what is reasonable on the component count, or the board size and functionality. The resulting demonstration boards, for example, are so small that they can become difficult to handle (I've recently seen a lot of advertising for demonstration boards the size of a quarter, barely protruding from the tip of a USB connector). Or, the "demonstration" capabilities are reduced to blinking a single LED (I've seen this way too often). When this happens, we can very quickly fall back to the perceived "zero value" impasse.

With the understanding that a development board has to withstand some minimum usefulness/usability criteria, how low can/should we go?

So far, the best way I've found to express the answer is by using a criterion based on identifying the minimum level of "economical sustainability" for the board. In other words, the board must provide the required functionality while being sold at a price that incorporates sufficient margins for the company to be able to pay for two things, in addition to the bill of materials: the initial development costs and the future maintenance. The initial development costs can be safely ignored, in most cases. The more successful the board, the higher the volume it is produced in and, therefore, the lower the incidence on the final price. The second part, the cost of future maintenance, is ignored all too often, but with severe consequences ultimately felt by the user. As a new board becomes successful and the volume increases, the cost of supporting the board, answering technical questions, providing software updates and so forth, tends to increase over time. If this cost is not balanced by a reasonable stream of revenue, it is very likely that the company producing it will soon be tempted to drop support and/or kill the board altogether.

The user pays the ultimate price, as they lose the ability to re-use the software and hardware developed and/or, as a minimum, they've wasted time learning/wrestling with one more development interface. As a corollary, this gives all embedded-control engineers a method for better judging a tool's value. If the price is too good to be true—barely equal to (or below) the sum of the components used—think twice before spending any of your time on it!

—Lucio Di Jasio, Microchip