backtop

Science MIT, TI Team Develops Low-power Chips
Levi Beckerson (Blog) - February 5, 2008 2:17 AM
Print E-mail del.icio.us 18 comment(s) - last by winterspan.. on Feb 7 at 7:25 PM
Prepare for the 100 hour cell phone, but don't look to battery technology for it.

A new breed of microchip is just over the horizon, thanks to development by Massachusetts Institute of Technology and Texas Instruments scientists. The new low-power microchip is to be presented at the International Solid-State Circuits Conference in San Francisco today by Joyce Kwon of MIT's Department of Electrical Engineering and Computer Science.

The standard operating voltage for microchips has kept small devices like cellular phones, portable music devices and digital cameras from having a long and healthy battery life. Most commercial microchips operate on about one volt of power. While this is not a great deal of energy, it adds up more quickly than the average iPod user would like. MIT & TI's new low-voltage chip operates at just .3 volts.

The lower power consumption could enable mobile devices like cellular phones and pacemakers to operate much longer on the same batteries as in use presently. With such low operating draw, the group hopes that some medical devices may even be able to be powered naturally, by body heat or kinetic energy collection systems.

A key to the new chip's design is the integration of a highly efficient DC-to-DC converter which resides on the chip itself. Reducing the number of separate components necessary by combining the converter with redesigned logic and memory helps to greatly reduce the power consumption of the new, proof of concept microchips.

The group hopes that with continued refinement, the new design could see use in commercial and medical products within the next five years.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled
May want to clarify this...
By Brunnis on 2/5/2008 5:56:13 AM , Rating: 5
"Most commercial microchips operate on about one volt of power . While this is not a great deal of energy , it adds up more quickly than the average iPod user would like."

This part makes little sense. The voltage used says nothing about the power consumed. Just like power is not synonymous with energy.




RE: May want to clarify this...
By bobsmith1492 on 2/5/2008 7:17:22 AM , Rating: 2
Very true:

Power = current x voltage (Watts)

Energy = power x time (Joules)


RE: May want to clarify this...
By TomZ on 2/5/2008 9:11:46 AM , Rating: 4
The equations you state are for steady state (DC) power consumption. For high-speed ICs, dynamic power consumption dominates, which is the power dissipated in charging and discharging the tiny effective capacitors. That power is roughly calculated as follows:

P = v^2 * c * f

Where 'v' is the voltage, 'c' is the capacitance of the pin/circuit/node and 'f' is the operating frequency.

A secondary effect is leakage which is calculated using the static formula, e.g., P = v * i(leakage).

From the above, you can see the importance of reducing the voltage, since power is proportional to voltage squared. For example, cutting the voltage by 50% reduces the power by 75%.


RE: May want to clarify this...
By Adonlude on 2/5/2008 8:53:17 PM , Rating: 2
What IC's specify Vcc/Vee power consumption in this dynamic manner? In all the data sheets I have read there are positive and/or negative DC voltages specified, usually with a DC quiescent current draw. This holds true for all the logic devices I have dealt with as well as computer CPU's.

Are you going into a more esoteric description of how energy is actually dissipated in switching semi-conductor devices as opposed to a practical description of how power dissipation is dealt with in the real world?


RE: May want to clarify this...
By geddarkstorm on 2/5/2008 12:37:21 PM , Rating: 2
Watts also equals Joules per second (J/s), so in the end we come full circle. Power being the amount of energy at any one time, and energy being the amount of power over a certain length of time. They are certainly interrelated and the article seemed to use them in the right way (proper temporal sense), at least as far as I see.


RE: May want to clarify this...
By spluurfg on 2/6/2008 12:22:50 AM , Rating: 2
Watts don't simply happen to also equal 1 J/s, that is how a watt is defined. Also:

quote:
Power being the amount of energy at any one time, and energy being the amount of power over a certain length of time.


...is backwards. Power (watts) is energy per unit of time.


RE: May want to clarify this...
By spluurfg on 2/6/2008 12:23:22 AM , Rating: 2
Watts don't simply happen to also equal 1 J/s, that is how a watt is defined. Also:

quote:
Power being the amount of energy at any one time, and energy being the amount of power over a certain length of time.


...is backwards. Power (watts) is energy per unit of time.


RE: May want to clarify this...
By Calin on 2/5/2008 7:18:36 AM , Rating: 2
The same circuit (non switching) will consume power in relation to the square of the voltage. Switching circuits with capacitive loads will consume less than that (I think), but there are still energy savings to be had: moving from 1 volt to 0.3 volts will reduce the power consumption by a factor of (10/3)^2 (11 times).
If this works out, and can be implemented in microprocessors, the days of heat pipes, copper heat sinks and 120mm fans are over - you might have again processors running with a small heat sink


RE: May want to clarify this...
By tjr508 on 2/5/2008 7:23:18 AM , Rating: 2
You forgot you just reduced your switching speed by (10/3)^(1/2).


RE: May want to clarify this...
By murphyslabrat on 2/5/2008 10:39:19 AM , Rating: 2
quote:
If this works out, and can be implemented in microprocessors, the days of heat pipes, copper heat sinks and 120mm fans are over - you might have again processors running with a small heat sink

Or, Core 3 Duo gets to 10Ghz


By winterspan on 2/7/2008 7:25:23 PM , Rating: 2
Core 3 Duo? More like "Core 3 Hexadeca" lol! @ 6Ghz


RE: May want to clarify this...
By tjr508 on 2/5/2008 7:19:42 AM , Rating: 2
Assuming they are using roughly the same transistors, higher voltage does mean higher power.


By learningtree on 2/5/2008 7:36:02 AM , Rating: 2
Of course, its correct that voltage, by itself says nothing about the power consumed.
However, in the context of integrated circuits, power dissipation is, in general proportional to the core voltage. For a given geometry(transistor size), the higher the voltage, the more is the leakage current. This leads to higher power consumption.
Thats one of the reasons for the reduction of core voltage of ICs/CPUs from 5V to 3.3V,1.1V to the present level of about 0.8 Volts.


Stable PSU more important than ever?
By InternetGeek on 2/5/2008 4:07:37 AM , Rating: 2
A DC-to-DC converter in the chip sounds like future PSUs will have to be more stable than ever if I understand the idea correctly (PSU provides DC current which then a CPU or Memory converts to the DC current it actually needs).




By learningtree on 2/5/2008 4:18:50 AM , Rating: 3
No, actually you are completely wrong. DC-to-DC converters have nothing to do with CPU or memory. In most cases, these are either linear shifters or switched-capacitor type. Of course, going by the requirement of high efficiency, they designed a switched-capacitor type DC-to-DC converter.


does this mean...
By inperfectdarkness on 2/5/2008 9:00:06 AM , Rating: 2
so i'm assuming .3 volts will equate to reduced necessary wattage to run a cpu.

if this technology is combined with room-temperature super-conductors (well, as soon as they're developed...) does that mean that your standard cell-phone battery will now be able to last months without charging?




RE: does this mean...
By tjr508 on 2/5/2008 10:58:01 AM , Rating: 2
I doubt it. This tech fails to take care of the number one and two causes of battery run down on phones:
radio transmitter
lcd backlight


By s12033722 on 2/5/2008 11:54:53 AM , Rating: 2
I think they have their work cut out for them. Dropping the signal level that far means that they need to really drop their noise floor or their bit error rate is going to go way up. I am not sold on this being a good idea at all.




"Game reviewers fought each other to write the most glowing coverage possible for the powerhouse Sony, MS systems. Reviewers flipped coins to see who would review the Nintendo Wii. The losers got stuck with the job." -- Andy Marken









botimage
Copyright 2010 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki