Dangerous datasheets

Datasheets themselves are not that dangerous of course.  The paper variants perhaps would make good tinder for the fire pit (I don;t know this for sure, never tried), but in general they do not detonate or release toxic fumes.

It’s just that top page that is dangerous, the one that marketing wrote rather than the engineers.  Those promises of features and performance that will then be fine-printed over the next 50 (or 500 pages).  Components can be complicated these days, especially in the electronics world.

We have a project from a customer to redesign a product line based on such datasheet first pages, promising 20% better performance than the component currently being used.  This components is the key component in the system, and paying a few cents more for 20% better is a no-brainer from a product line point of view.  It’s being heavily marketed as a next generation technology, and its figure of merit indeed sounds cutting edge.

Except the new super component may not really be better in our application.  The devil is in those 50pages down below.  On the technical side, its key performance criteria are speced at different operating conditions than the incumbent component.  We’ve dome a slew of calculations and built up a complex spreadsheet to model the relative performance of the two at various real (for our application) operating points, , and of course talked with the apps engineers and tested real parts.  We’ve even gone so far as to use calipers on the curves and used third-order curve fits.

On top of this there is the usual muddy area of nominal versus worst case, and even more complicated, a system of binning that effectively creates a number of grades.  Besides the additional technical complexity added by these grades, there is a supply chain side as well, as there is a significant difference between the sales policies of the two respective vendors as to which grades they will sell under what conditions.

Net, net, what started out looking like a runaway victor now looks like a photo finish.  Work still to be done on both the design and supply chain side.

And of course this is a recurring issue across the industry, not just in doing side by side comparisons.  We do a lot of “rescue” redesign work for various customers, with root causes oftentimes rooted in page 43 or page 433 of data sheets.  Common examples include decoupling schemes and power supply sequencing (in one case literally on the four hundredth page).  And we’re not immune to such mistakes ourselves–I personally burned up a thermo-electric cooler sample recently  from not reading up on the details of heatsinking (fortunately it was just a prototype, and no it did not explode either).

Happy reading.  Better put another pot of coffee on.

Chuck

2 Comments

  • Tak

    Thanks for sharing srepub informations. Your internet site is really cool. I’m impressed by the details that you’ve on this web page. It reveals how nicely you perceive this topic. Bookmarked this site page, will come back for extra articles. You, my friend, ROCK! I found simply the info I currently searched everywhere and just couldn’t come across. What a perfect web-site.

Leave a Comment