Who has the lowest power MCU?
I'd be very careful posing this question in any public forum of any kind, real or virtual. It is like asking ten doctors for an opinion: you'll get ten opinions and be none the wiser.
Surely Google, knows the answer - “Google, Google, on the wall, who has the lowest power MCUs of them all?” The top 10 results from the search term "low power microcontroller" were as follows (bare in mind, Google matches my searches to my profile):
1. TI Ultra Low Power MCU (pay-per-click advert)
2. 32bit low power mcu RX100 (Renesas) (pay-per-click advert courtesy of Silica)
3. Ultra Low Power MCU - ARM Cortex-M0 from Freescale (pay-per-click advert)
4. MSP430 - TI (first true search result)
5. picoPower - Atmel
6. eXtreme Low Power - Microchip Technology
7. Ultra Low Power MCU - Silicon Labs
8. C8051F96x - Silicon Labs
9. EE Times article on this topic
10. Low-Power MCU - Maxim (yes, Maxim have MCUs)
Now I have a list of some potential suppliers to research. I also know that TI pays top dollar for advertising and Renesas is getting good promotion courtesy of Silica! And curiously, Silicon Lab's 8051 MCU is the first actual MCU device highlighted;, what about their newly acquired EFM32 devices from Energy Micro?
So having trawled the net with this topic in mind, Energy Micro, now part of Silicon Labs, seems the most balanced in describing the aspects of an MCU that are critical to achieving low power consumption . It’s a combination of factors, including a low-leakage technology for when the MCU is in a low-power state, a well performing CPU to quickly get tasks finished when active and autonomous peripherals that can undertake a certain amount of work without needing to wake the CPU up.
So, as Douglas Adams said in The Hitchhiker's Guide To The Galaxy, perhaps the reason we are not getting an answer that helps is because we are asking the wrong question. Perhaps the question should be "Who has the lowest power MCU for my application requirements?" And this is exactly where the fun starts!
As Peter Clarke (EE Times) in his article mentions, everyone wants to claim the moniker of “World’s Lowest Power MCU”. But a data sheet comparison (as listed in this article) only provides some guidelines that we could plug into a rough Excel sheet calculation based upon how we intend to exercise the part. Reading through the comments left by readers, we sift through the usual “10 doctor’s opinions”. Except I see, from my perspective, a ray of hope in the form of the person going by the name of ‘schneiderham’, and I quote:
“I am seriously considering doing my own comparison test--an all out brawl--of MCUs that manufactures' are claiming to have "The industries lowest Power MCU". I plan to create three typical Low power applications and then optimizing to each processor and comparing their power performance. So far in my list of possible MCUs are: MSP430 XLP AVR picoPower RL78 STM32L Energylite EFM32 Does anyone know if anyone else has done this? Does anyone else have a MCU they would suggest?”
Exactly! This is what we really need - a side by side comparison. Only one issue with this though, who has the time to install several toolchains and IDEs, learn enough about each MCU to be able to write a comparative application code and then have the money to invest in the necessary hardware (i.e. both development kits for the MCUs and test equipment that can measure accurately from nA to mA)?
Evidence online suggests this is a challenge that many of us face. So how do you really tackle this? When faced with a new low-power MCU design challenge, do you just stick with what you know, or do you review the market, if so how deep do you go? Mere data sheet comparisons or do you dive in and get your hands dirty? Do you have the time to undertake detailed research? Is your laboratory kitted out with the equipment to undertake such tests and is it available when you need it?
Please leave your commens below to discuss your approach to choosing low-power MCUs and what “low-power” means for your applications.
CommentsAdd a comment
January 29, 2015 17:22
For me the lowest power MCU seems to be the PIC16F1503 from microchip XLP. See http://www.microchip.com/pagehandler/en ... ducts.html. It uses only 25uA/MHz. Of course the actual power consumption should depend on the application, but this should be for a low demanding application. The power consumption can even be lower since in sleep it only uses 25nA! Can anyone beet this?
March 3, 2014 21:01
For sure there are many facets to be considered and, yes, I agree that claiming an MCU is the "lower power" is becoming more and more a vacuous statement. Efforts are being made with, for example, the EEMBC low power benchmarking (see http://www.eembc.org/benchmark/ulp_sl.php) so that, as far as possible, an apples to apples comparison can be made.
In conjunction with RS Components we did demonstrate an on-line, web-browser based system for running hardware experiments to make a power consumption assessment. The demonstration was shown at Embedded World 2014 last week and received a warm response from both engineers and semiconductor manufacturers. We hope to be able to share more on this approach to making real hardware comparisons via web-browser controlled applications in the future.
Best regards, Codinghead
March 1, 2014 17:18
Today on the market vendors are figthing to show the lowest value on datasheet and other promotion materials but some keys paramters are still missing. For instance talking about ua/MHz is not really relevant if users has no idea of the platform efficiency (and not the core only). As an example a MSP430 has a score of 0.6 coremark/MHz while a STM32L1 (STMicroelectronics ARM coretx-M3) has a 2.9 coremark/MHz. It means the Cortex M3 core device will stay 4 time less in active more than the 16-bit core MSP.
Also vendros are tlaking about lowest power mode and very fast wakeup time but the energy and peak of current that occurs during fast is quite never mentionned except by some vendros.
Depending the aplication ultra-low-mcu will have different meaning. For example if I want to build a fire alarm detector, a gas meter or a mobile phone the ultra-low-power MCU will not be the same.
I was at embbedded world 2014 last week I have seen Atmel with the D20/21 family which is low-power only (not ultra) because of the high quiescent current. Also STMicro has announced the STM32L0 (Cortex M0+) which seems far ultra-low-power vs. Kinetis L family or EnergymicroMCUs both on Cortex-M0+ too.
I got a Nucleo eval board with Arduino connectors, I will make some test and let you know.
Thanks for reading.
November 4, 2013 11:09
Thanks for your comments. What is driving your need for low-power in the applications you work on? Are they battery powered or is there something else limiting your power budget? Which three MCU families would you turn to first in your search for a low-power processing unit for you next application?
November 3, 2013 11:43
I am new to this site and i want to know what is this all about. I really appreciate if you orient me everything in here. - Texas Lending
November 1, 2013 09:59
In finding the lowest power CPU possible, there seems to be no method to avoid a completed design before knowing the actual power consumption of the MCU. One could even choose several different MCU's and achieve nearly equal power savings with careful design strategies. In many cases, the choice of CPU may have much less influence on the power consumption than the careful design based upon ones understanding of a particular MCU. I could even say that the compiler chosen may influence the power consumption depending upon how closely the actual compiler models the strategy of the designer/programmer. All in all, calling an MCU the "lowest power MCU" seems almost facetious in lowest power design.