Dan Zhao and Simone Fabiano at the Laboratory of Organic Electronics, Linköping University, have created a thermoelectric organic transistor. A temperature rise of a single degree is sufficient to cause a detectable current modulation in the transistor. The results have now been published in Nature Communications.
"We are the first in the world to present a logic circuit, in this case a transistor, that is controlled by a heat signal instead of an electrical signal," states Professor Xavier Crispin of the Laboratory of Organic Electronics, Linköping University.
The heat-driven transistor opens the possibility of many new applications such as detecting small temperature differences, and using functional medical dressings in which the healing process can be monitored.
It is also possible to produce circuits controlled by the heat present in infrared light for use in heat cameras and other applications. The high sensitivity to heat, 100 times greater than traditional thermoelectric materials, means that a single connector from the heat-sensitive electrolyte, which acts as sensor, to the transistor circuit is sufficient. One sensor can be combined with one transistor to create a "smart pixel."
(Score: 2, Interesting) by Anonymous Coward on Wednesday February 01 2017, @03:29PM
This is very interesting work, but one needs to keep in mind that this is inherently a "slow" process. A change in temperature needs to be generated, which is related to the thickness of the thermoelectric region; you can make it operate faster by making the thermoelectric region thinner, but that leads to more instability in the temperature difference. The paper itself [nature.com] mentions that this would work well for something like skin temperature monitoring, which occurs on times like 20 to 200 seconds. The idea of a "smart pixel" implies fast, camera-like response, but in this context they are really talking about imaging things with relatively steady-state temperatures.
(Score: 2) by The Mighty Buzzard on Wednesday February 01 2017, @03:42PM
It's a shame it's too big to be used in a CPU/GPU. Automatically switching off bits of the processor based on heat for instance so that you literally could not burn a chip up any more would be kinda sweet.
My rights don't end where your fear begins.
(Score: 2) by takyon on Wednesday February 01 2017, @03:55PM
I think a lot of people will come in here looking for the HAMR of transistors that allows further shrinking and vertical scaling, but it does not look like that whatsoever. Too bad.
My computer was using something like 1000 threads earlier. If we don't have a core for each of those, we're nothing but savages. Make it parallel to V-NAND development, put 4 cores on each layer, and we can have a 400+ core chip in no time.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by bob_super on Wednesday February 01 2017, @06:28PM
That's a problem already solved by diode+ADC, which gives you dynamic control over where you set your trip point based on part testing and use case.
Not sure I see the benefit.
(Score: 2) by butthurt on Wednesday February 01 2017, @06:32PM
P6 family processors introduced a thermal sensor that acts as a catastrophic shutdown detector. This catastrophic shutdown detector was also implemented in Pentium 4, Intel Xeon and Pentium M processors. It is always enabled. When processor core temperature reaches a factory preset level, the sensor trips and processor execution is halted until after the next reset cycle.
The mechanism is builtin to the processor, not under the control of BIOS and not dependent on other sensor outside the processor.
-- https://software.intel.com/en-us/forums/software-tuning-performance-optimization-platform-monitoring/topic/282144 [intel.com]
(Score: 2) by The Mighty Buzzard on Wednesday February 01 2017, @10:37PM
I was thinking more selective-like. Clocking down the cores based on heat and outright shutting down bits you could shut down and not bugger up anything running on it.
My rights don't end where your fear begins.
(Score: 2) by takyon on Wednesday February 01 2017, @11:07PM
I'm pretty sure they already do that.
https://www.pcgamesn.com/amd/amd-zen-release-date-specs-prices-rumours [pcgamesn.com]
Pure Power and Precision Boost are part of a 5-feature package AMD calls "SenseMi".
Maybe a heat transistor could do the job better than the sensors, but I doubt the sensors have a huge power draw.
(I expect Intel includes similar features, but I didn't see anything noted for Skylake or Kaby Lake)
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by butthurt on Thursday February 02 2017, @06:31AM
> (I expect Intel includes similar features [...]
Someone (aznricepuff) on another Web forum wrote:
TCC activation temperature: The temperature, measured at the CPU core(s), at which a PROCHOT# signal is sent. This is the signal that activates throttling of the CPU. In essence, the CPU will rapidly step down voltage and clock speed until temperature drops below the TCC activation temperature. If this doesn't work, the CPU will cycle the clock on and off repeatedly (effectively only remaining active a fraction of the time) to reduce temperatures.
-- http://www.webcitation.org/6ny96bCxK [webcitation.org]
(Score: 2) by VLM on Wednesday February 01 2017, @04:07PM
Yeah that's too slow. The Buzzard already commented on CPU apps but the first thing that came to mind for me was on the die of a RF power transistor you could put something like this in series with one of the leads and when it starts VHF oscillations or goes into thermal runaway usually in about a second the transistor blows. I've seen packages actually fracture although its usually not that exciting. Anyway 20 to 200 seconds is at least one or two orders of magnitude too slow for transistor thermal protection circuitry but Maybe with continued development etc etc.
In the past I looked into patenting the idea of a shitty grade microcontroller speaking I2C on the same die as say a MMIC or RF power amp transistor and it could talk to the microcontroller of the radio or whatever and even with crappy technology the transistor could talk to the main controller to tell it to cut it out if it was overheating. Also it was not rocket surgery (although it added capacitance) to monitor the leads of DC switchers and the "smart transistor" could tell the controller if it was burned out or testing normally. The problem ended up being that there's nothing on the marketplace but there is prior art and prior patents. Maybe when the patents expire we'll see "smart transistors" flood the marketplace, it seems like a pretty obvious win. Why shouldn't my high power MMICs have a I2C interface and an onboard monitoring system? (Obviously low power receive path this doesn't work for digital noise reasons)
Anyway the second paragraph is strangely on topic because it means in a couple years the application of the first path would be obsolete. So they best license their "thermal transistor" and ship it fast or it'll get runover by newer more advanced on board monitoring.
You know when I was a kid playing with retro at that time vacuum tubes and retro tube testers I never thought the day would come when I'd connect something like a contemporary "bus pirate" (a cool product BTW) to a transistor to monitor and test it, very star trek world.
(Score: 2, Insightful) by Scruffy Beard 2 on Wednesday February 01 2017, @05:23PM
And that is how patents stifle innovation.