At $1.5 a liter in Moscow, [Michail] couldn’t resist buying some liquid nitrogen for himself. He thought that because Arduinos were quite popular among geeks, he’d try to overclock one while bringing its temperature down to -196°C/-320°F.
To check the ATmega was still working correctly, [Michail] designed several stability tests: SRAM read/write, flash read, arithmetic math and program flow tests (code with some conditionals). He used a standard HD44780 LCD to view the tests results but also an LED, blinking the number of the test it would have failed. The Arduino was externally clocked by a TTL-logic based square signal generator he designed, which can produce a clock between 16 and 100MHz. It turns out that you can run an Arduino at 65.3MHz when it is cooled with liquid nitrogen!
[Michail]’s article also explains what happens to the different on-board components when cooled with LN2: electrolytic capacitors becomes virtually non-existent, X7R capacitors’ impedance drop by 2/3, silicon diodes voltage drop increase by 50% and LED’s colors change. Check out the video below:
Dude has some guts! “Little liquid nitrogen in your backpack never hurt anyone” . Then, on the video, he’s practically dipping his fingers into the vat …
Why is LED color changing under low temp?
Leidenfrost-effect?
It’s probably because the bandgap of the semiconductor changes with temperature (delta E = kT) and therefore so does the energy and colour (E = hv) of the photons.
Same reason why the silicon diode voltage drop changes. The bandgap of the diode determines the forward voltage drop, which determines the energy of the photon that gets released when a charge crosses the gap.
It’s actually a really cool quantum effect. The extreme cold of the liquid nitrogen changes the bandgap enough to visibly change the energy, and thus the color, of the photons released.
Ref: http://www.ap.smu.ca/demos/index.php?option=com_content&view=article&id=181&Itemid=78
Correct.
Source: Me, as this is my career.
So does this mean that you can make a monochromatic light source at any wavelength by taking a “close enough” laser diode and then shifting its temperature?
Not exactly. Although it’s important to stabilize the diode temperature for this very reason, varying the temperature isn’t a great way to tune a laser. The laser will experience a phenomenon called “mode hopping” which produces discontinuities in the wavelength/temperature dependence.
Fig. 1 of this paper shows mode hopping in a diode laser: http://assets.newport.com/webDocuments-EN/images/AN08_Mode_Hopping_Laser_Diode_IX.PDF
http://www.ap.smu.ca/demos/index.php?option=com_content&view=article&id=181&Itemid=78
http://www.youtube.com/watch?v=gjsMV1MglA4
I figured out that for this to work liquid nitrogen should be “fresh”, without floating ice crystals. Ice makes it much less fun :-)
,
You can pour a spoon of liquid nitrogen into your mouth without hurting you, in fact the nitrogen will evaporate almost instantaneously our your mouse and nose. Trust me, I’ve done it.
People have died doing that, they suddenly swallow from reflex and freeze/rupture their internal organs.
not to beat a dead horse, but as plenty of other people have cited, “Leidenfrost-effect”. Even if you immediately tried to swallow it, you’d be swallowing a gas. I’d be more worried about asphyxiating from the relatively large amount of nitrogen you have just filled the air immediately inside your mouth and nose with.
feel free to keep beating that dead horse. -but it doesn’t make you any less wrong.
http://www.bbc.co.uk/newsbeat/19866191
When swallowed liquid nitrogen, (can and documented does) cause cold burns and tissue death to the mouth oesophagus and stomach.
in cases like the one I’ve linked above necessitating the complete removal of the stomach…
LN2 is actually pretty harmless despite what most the people say. your body is so hot compared to the nitrogen that it will just vaporize on contact so unless you really put you hand in there for a few minutes noting bad is ever gonna happen.
“X7R capacitors’ impedance drop by 2/3” should read as “X7R capacitors’ capacitance drop by 2/3”
Impedance is the technically correct term actually…
http://en.wikipedia.org/wiki/Electrical_impedance
Than it should have been “X7R capacitors’ impedance increased by a factor of 3″, as impedance increases when capacitance drops.
*then
[R]un an Arduino at 65.3MHz!
A DUE has a stock CPU Clock of 84Mhz
It doesn’t take a brain surgeon to see that he’s using an ATMega328 here.
Obviously. But he’s liquid N2 cooling it and it’s still worse than a DUE. Upgrade the processor, not the cooling.
It’s cooled by LN2 not because I needed to make some heavy duty computations, but because I was curious how far I can push it :-)
I surely had both “small” (STM32) and “big” ARM (Qualcomm and iMX6), FPGA devboards which are faster but several orders of magnitude. But there is no fun using these without LN2 :-)
This is …Cool
(Sorry, had to make the pun)
im blown away that it still works down at those temperatures!
Why would you believe it would not?
There’s a thing called the freeze-out zone in semiconductors, electronics just don’t have enough entropy to function correctly anymore. You’d think lowering the temperature to extremely low temperatures enables you to overclock to infinity, but unfortunately other parameter come into play.
Here’s a website that explains it well: http://www.extremetemperatureelectronics.com
exactly as [solenoid] says … not to mention the possibility of cracks in chips and on the silicon from shrinking and some active components slow down and change there tolerances … also i believe internal resistors lower their value and you can risk blowing things
and i can guarantee you the internal voltage reference is skewed dramatically
but the fact that this chip lived that low is amazing to me seeing how it was never designed to put up with temperatures below 0! (i think)
Actually, the chip is specified for industrial temperature range (-40 to 85 degrees Celsius). Military chips are specified for range of -55 to 125 degrees. Only commercial ICs are specified for 0 to 70 degrees, but most of the time they’ll work below that temperature.
Are you talking C or F?
The temp ranges (except for the -40) are in C. The -40 is in F.
Extremely cool experiment, tnx for doing this. Always good to get stuff benchmarked. I’d at least wear some goggles cause if a cap blows you might end up with some eye damage there.
For the rest… now someone has to go and test it under heat. I guess it’s done when the solder starts to melt.
It is. Iused to do that with 68HC11 boards in a programming class. Put them in a special oven that would push the boards until just before the solder starts to shimmer then run some HSF instruction or some other stress test IIRC.
Hm, a reflow oven and some heat resist wires would do the deal… :)
yes, because it’s the same silicon from the mil to the commercial version
I don’t think that you need to cool it with ln2 to push it to 67mhz..
At room temperature it was stable only until 32.5Mhz (5V) and 37Mhz (8V).
In 21st century Russia, Arduinos run faster!
was the ATMEGA328 completily covered witn LiN
What I enjoy the most about this video is how excited he gets when he sees results!! so much so that he does that thing where he can’t connect the crocodile clips quick enough!! awesome!!
To me the interesting part is that the ‘328P itself doesn’t seem to be “boiling” much at all, but rather the 16U2 chip is going to -town-. Fast-forward to ~9’50” on that video to see what I mean.