An ongoing challenge facing embedded system hardware and software developerswill continue to be how to manage MCU power consumption. Operating at lowestpossible power has always been a given in most traditional 4-/ 8-16-bitMCUs, as well as 32-bitters, where operation within constrained compute,memory and space requirements is necessary. But it is much more importantnow and in the foreseeable future for at least two reasons.
First, more and more embedded designs are being deployed in completelyuntethered environments, communicating data wirelessly and operating on aminimum of battery power or from what can be obtained by energy harvesting.Second, the submicron/nanometer scale CMOS technology that brings developersmore processing capability is a double edged sword.
On the one hand, more functionality means more power consumption. On theother hand, at these geometries non-linear quantum effects raise up theirhead in the form of difficult to suppress leakage currents, further complicatingthe power equation. While SoC designers and process engineers are doing theirbest to deal with such issues at the silicon level, most of the burden fallson embedded developers and the board and IC level tools and resources theycan bring to the problem.
This week’s Embedded Tech Focus newsletter includes a number ofrecent design articles, blogs, webinars and white papers dealing with suchissues. In addition there is a recent blog by Jack Gannsle that I found enormously useful in sortingout the key specs that are important. Some other articles and blogs thatI recommend include: Leaksand Drains also by Jack Ganssle, as well as:
Conflicting MCU power specifications
In all such designs, a persistent set of problems that developerswill have to deal with are the often conflicting and misleading claims byMCU vendors about the current, voltage and power parameters of their devices.
In his most recent blog “Asneak preview,” Jack summaries the situation succinctly: “Every MCUhas vastly different mixes of features. Some wake up quickly. Others executeat great speed so wake times are minimized, ” he writes. “Some preserveregisters and memory while asleep; others don't. ” So making comparisonsis difficult. Even the sleep current numbers, he writes, are frustratinglydifficult to parse as not all vendors give worst-case values.
Of several recent technical journal and conference papers I’vefound that have given me some additional perspective on the problems he raises,two that I found particularly informative are:
“Comparingenergy efficiency of 8 bit low power MCUs,” in which the authors analyze and compare the energy consumption and computationalefficiency of six different 8-bit microcontrollers from three manufacturers,and, .
“Lowpower vs high performance in 8/16- and 32-bit MCUs,” where an internationalteam of researchers examine whether modern 32-bit processors can rival their8/16- bit counterparts in terms of both power and performance.
But as far as hands-on tools and guidelines that developers of MCU-baseddesigns can use, there seems to be little useful information available. Butthat is changing. Help is on the way.
In recent blogs on Embedded.com, JackGanssle has taken on the challenge. He has found a number of tools he thinkswill be useful in this task and has reported on them in blogs on the I-jet,the Real-TimeCurrent Monitor, and the uCurrentdevice.
Now, in “Asneak preview” he describes the test setup he has created with thosetools. He gave the blog that title because the results he reports there arejust the tip of the iceberg. “I've been running experiments for 6 monthsto gain a deeper understanding about building ultra-long-lived battery-poweredsystems, ” he writes, “and will be reporting more results soon. ”
So stay tuned. I certainly will. I am sure he would like to hear fromyou about your experiences and your reactions to what he has found out sofar. So read his report and leaveyour comments on his blog.. Also, if you have some ideas and experiencesthat deserve more space, send me an email and submit your idea for a blogor article of your own in the site.
Embedded.com Site Editor Bernard Cole is also editor of thetwice-a-week Embedded.comnewsletters as well as a partner in the TechRite Associates editorialservices consultancy. He welcomes your feedback. Send an email to , or call928-525-9087.