I recently read a technical article about sensor linearity that I feel had a few errors. This reminded me of a few technical articles that I have read on other matters, especially data converters, that had errors as well. I do not mean differences of opinion, but these articles had errors of fact.

The sensor linearity article, for example pointed to a figure that showed a typical response curve of a thermistor with a knee at about 0ºC and stated that this is where the curve is “most linear”. I had to read and reread this to see if I had misread the statement because it is so obvious that this is where the curve is most nonlinear. But the words are ”The range of highest linearity for this device is roughly between -20 and 20ºC.” Hm-m-m-m!

 

An ADC question I once saw was “How can the TUE be 0.5% from a device that claims 21.6 ENOB?” The answer was an excellent one and right on the mark, except the part of the answer that said it “sounds like you are working with a 12-bit converter.” That is twelve bit converter. Well, you can never realize an ENOB (Effective Number of Bits) that much greater than the converter resolution. Getting an ENOB of 21.6 from a 12-bit converter just will not happen. With input dither and averaging you can realize a little greater ENOB than the converter resolution, but you will never go from 12 bits to over 20 bits.

 

My point to all of this is that we must apply a bit of reasonable thinking to whatever we see or hear. Technical writing is generally done by people with other primary tasks and the writing is secondary, so that it tends to be a bit hurried. Even if it is not hurried, we are human and can make errors. Our own errors are difficult to catch and people proof reading for us may be in a rush and miss errors. Also, proof readers may be so familiar with the subject that they see what they know to be true and miss some errors even if they are not rushed. So, apply a little reason to what you read and hear. If in doubt, contact the author to explain your concern or confusion and ask for an explanation.


- Nick Gray