Mike Mitchell

image
consultant

Biography has not been added

Panometric

's contributions
Articles
Comments
    • I agree with Wcolley3 that flux contaminants can be a big issue. Many CMs advocate noclean, but I found found this residue will absorb humidity and become slightly conductive. So now we wash and dry all boards in manufacturing even if the flux is "noclean".

    • There's a better way: Use a current mirror with the device on one leg and charge a capacitor with the other. Then measure the resultant charge. The charge will be linearly proportional to the cumulative current used, regardless of the rate. It's more accurate than a log scale and does the math of integrating the waveform for you. It's hard to see how making the out log will simplfy the measurement.

    • Nice article on emerging standards that leverage previous work and merge two areas of development that do not often cross, namely embedded and web services. The CoAP mechanism seems like a good compromise over the TCP / UDP conundrum. I too was forced to build some protocol over UDP to get the determinism, but still be able to deal with errors on the channel. But it is better to have a standard way to do that so that network analysis tools like Wireshark can follow it. I doubt XML would be used over wireless networks, but I could see JSON as being a reasonable compromise from binary proprietary protocols. As I read the working group recommendation on EXI, I can only say that someone forgot the KISS principle. http://www.w3.org/TR/2011/REC-exi-20110310/#key-exiencoder A point you may have overlooked is that web services can usually be combined with https to get a secure channel. Most proprietary protocols only have security by obscurity. But as sensor networks become more important, this is increasingly unacceptable. http://www.panometric.net