In the EETimes report on this year's International Solid States Circuit Conference (ISSCC), called Pictures From a Silicon Exhibition, there are two references to MCUs that operate with a supply less than one volt. These millivolt MCUs may represent the ultimate in low-power design. It will probably be a good five years before the technology reaches commercial availability, however.
There are MCUs already available that can tolerate an ultra-low supply voltage. The Atmel 8-bit tinyAVR, for instance, can run with a 0.7V supply. But these devices simply use an on-chip charge pump to boost the internal supply voltage to the level needed. The millivolt MCUs, on the other hand, do not. They run with their internal core less than a volt.
Intel is a pioneer in this ultra-low voltage field and was demonstrating its Claremont chip at the ISSCC. This device was first announced at the Intel Developer Forum back in September. The code-named Claremont processor runs near the turn-on threshold for transistors, in the 300mV to 400mV range. This near threshold voltage (NTV) processor is not the basis of any planned products, according to Intel, but is an experimental unit that will aide in the development of such processors in the future.
Also at the ISSCC, some German researchers from the University of Paderborn's Heinz Nixdorf Institute of Electrical Engineering were showing their CORVEA ULP (ultra low power) processor.
The CoreVA experimental processor from the University of Paderborn.
This device, also experimental, uses adaptive techniques to keep its power demand as low as possible, and proved capable of performing useful work running below 350mV. The paper describing their device is only available in German, but Babblefish does a reasonable job of translating it for non-German speakers.
This technology, once it reaches the market, promises to further expand the range of possibilities for smart networks and embedded computing using methods like energy harvesting to power them. It doesn't take much solar, thermal, or mechanical input to produce a half volt at a few hundred nano-amps. When that is all the processor requires, barriers such as location and accessibility simply disappear.
Sure, this is all a few years away. But it probably will be less than a decade. So, now might be the time to start brainstorming about what you would do with a processor that needs almost no power. Got any ideas?
Microp 3/6/2012 2:21:28 AM User Rank Program Manager
Low power MCU- a promising technology
Rich, such low power consuming devices have a promising future in MCU/Embedded world. In most of the high end devices/appliances energy harvesting is a major concern and such low power consuming chips can incorporate for a better durability.
Rich Quinnell 3/6/2012 12:23:55 PM User Rank Blogger
Re: Low power MCU- a promising technology
Your mention of energy harvesting is right on the money. The people working on these near-threshold devices often refer to the ability to operate on a single solar cell as a feature of these devices. I presume that even under artificial lighting a cell can generate enough power to run one of these. Thus, anywhere that the users can see, these MCUs can operate. It makes for a tremendous application space.
Rich Quinnell 3/6/2012 11:30:54 PM User Rank Blogger
Re: Low power MCU- a promising technology
Microp, so what would you use such a low-power MCU for? The technology seems promising, but at the moment I am at a loss as to what would be useful to do if I could operate essentially without any power, even using energy harvesting.
Maybe we can turn the problem on its head, and consider applications where power is available but not voltage. A single AAA cell would have plenty of headroom to run one of these devices with peripherals. Further, it could run for a long time because it would not be dependent on the battery's terminal voltage being near its peak.
Any other suggestions out there? What could you do with an MCU that needed less than a volt in order to function?
Northstar, interesting concept, but I see that the post is from 2007. I haven't seen any wearable electronics cloth come out yet. And if you to put biometric sensors all over, say, a shirt - then what? Would this be used for a medical monitor of some kind?
Ryszard, yes those sound promising. The difference between body temperature and ambient has already been proven to be enough to power wristwatches and the like, so such monitors could easily work off of body heat. And with a microcontroller you could do more than simply monitor, you could trigger an alarm of some kind. Thanks for the example.
I notice that there is some kind of trend for this sensors into garments lately. I remember 10 years ago that I talked with a mountain rescuer and he told me that their jackets are equipped with a tracking device. If, for example, they are injured and lost, a detection device was able to locate them. This seems pretty common today, but at that time, this "device" was very tiny and embedded into jacket's textile. For that time, it was cutting edge technology.
Rich Quinnell 3/8/2012 11:23:17 AM User Rank Blogger
Re: Low power MCU- a promising technology
Northstar, interesting article about Apple's smart garment" patent. The stated benefits include an ability to monitor "wear patterns" for the garment, so they can be compared to those predicted for the garment. If I interpret that correctly, it means they are watching how the garment degrades over time as compared to expectations. The initial uses were with Nike for running shoes.
I wonder what they do with this data? The optimist in me says "design better, more durable shoes" but the cynic says "design shoes to support turnover in inventory." I sincerely hope it is the former.
This does give me an idea for these low-voltage MCUs. How about an orthotic insert for shoes that monitors pressure and walking patterns, using piezovoltaic power from standing and walking, that can then have their data downloaded to a physician's computer for the design of custom orthotics inserts for your shoes? An electronic Dr Scholls.
northstar 3/8/2012 11:29:54 AM User Rank Program Manager
Re: Low power MCU- a promising technology
And more, the data it will be automatically downloaded through an inductive circuit (or Wifi or some other wireless method) when shoes' owner step into a mall, for example. That way, data were collected without stressing the "subject". But some data could cross privacy boarders ... track your path literally!
duanebenson 3/6/2012 12:48:16 PM User Rank Blogger
Related hardware
Have you heard anything about any of the external components typically needed in an embeded system also operating on down near threshold voltage? I'm thinking about things like MOSFETs, LED, MEMs devices. I know there are a lot of level translation methods, but it would be nice to be able to keep the entire system down in the same low-power range.
I do agree with Duane, we need low voltage peripheals. For example wireless sensor node: sensor, ADC, RF transmitter. Then all could be supplied using e.g. thermoelectric generator.
By the way, in MCP560x core works at 1.2V and I/O at 3-5V, but it is not very energy efficient processor.
Some of the newer PIC processors, and of course some of the ARMs, operate down to 1.8V but I haven't built anything to run at that voltage yet. There is the peripheral issue and 1.8V is an odd voltage. The standard battery won't cut it and Lithium would require a regulator to drop the voltage. Barring a specific need, I don't see much point to 1.8V. 1.2 could get by with most single cell batteries, so that would make more sense.
Duane, does that 1.8V need to be regulated? I would think that the MCU had its own internal regulator and that it would tolerate a supply voltage down to 1.8V but up well above that because if its internal regulation. Does the spec sheet show a broad range?
One of the chips I've used, the PIC18LF26K22 can operate down to 1.8 volts. I don't see anything in the datasheet about a built-in regulator in that chip. The datasheet I have is still in preliminary state so there may be a bit on information covering supply that I haven't seen yet.
Some PICs do have internal regulators though. The PIC18F24J50 wants the core to stay at 2.5 Volts or less. It has an internal regulator for the core voltage allowing the I/O to be supplied with and operate at 3.3 volts. The data sheet notes that if the main input voltage drops below 2.5 volts, then the core will just follow operate at the supply voltage.
The ARM LPC1114 that I'm getting familiar with, doesn't show an internal regulator. The electrical characteristic charts show things like supply current on a continuous line from 1.8V to 3.6V. I'd interpret that to mean that any regulated voltage in that range would be fine. Some of the charts specifically call out different curves for 1.8V, 2.0V, 3.3V and 3.6V. Again, that would imply to me that it operates at any regulated voltage in that range.
PIC18F24J50 has internal regulator while LF version doesn't and must use external.
I think LPC1114 has internal LDO because Vcc is explained in datasheet as "supply voltage to the internal regulator" and other ARM MCUs have LDO built-in.
STM32F1 internal regulator has output voltage 1.8V, STM32F2 and STM32F4 1.2V.
PIC32MX5xx/6xx/7xx internal regulator output is 1.8V typical. So if you use 3.6V power supply, 50% of input power is lost.
Rich Quinnell 3/12/2012 2:33:07 PM User Rank Blogger
Re: 0.9V Silabs
Gutman, Thanks for the additional reference. Like most other currently-available MCUs, this one can operate with a supply below 1V because it has an on-chip DC-DC converter to bring that voltage up to the 1.8V that the core needs. Thus, this part and others like it are not what I would consider millivolt MCUs, which operate with core voltages near transistor switching thresholds. This near-trheshold operation means they use much less power than those with a core at 1.8V, even if the supply voltage is below that threshold.
Still, good to have low-voltage operation MCUs pointed out. Helps address current requirements.
Rich Quinnell 3/12/2012 7:50:59 PM User Rank Blogger
Re: 0.9V Silabs
Yes, input supply voltage can be as low as 0.9V. The internal voltage is higher, however. There is a DC-DC converter described in the datasheet that permits single-cell operation from a battery despite the core's need for 1.8V. This is the difference between what is available today and the research devices. Today's cores operate above 1V. The research devices have cores that can operate below 1V.
To save this item to your list of favorite Microcontroller Central content so you can find it later in your Profile page, click the "Save It" button next to the item.
If you found this interesting or useful, please use the links to the services below to share it with other readers. You will need a free account with each service to share an item via that service.
Tweet This
