SoylentNews is people
SoylentNews
Submitted via IRC for TheMightyBuzzard
DFI's Coffee Lake based "CS551" 3.5-inch SBC features an auto heat-up function that enables a -30 to 80°C range. The feature also appears on new, Whiskey Lake based, 2.5-inch "WL051" and Ryzen-based, 3.5-inch "GH551" SBCs.
[...] The heat-up feature enables the -30°C minimum by automatically heating the CPU when it drops below the chip's typical lower range of 0°C. On the high end, the systems supports up to 80°C instead of the usual 60°C in part by dynamically allocating computing resources between the CPU and GPU. On higher-TDP models, a fan option is required to achieve the range.
The auto heat-up function and -30 to 80°C support are also available on two other SBCs that are listed as new: an 8th Gen Whiskey Lake based, 2.5-inch WL051 and a Ryzen Embedded V1000/R1000 based, 3.5-inch GH551 (see farther below).
Now that is pretty damned nice. Beats the hell out of having to build and heat an enclosure.
(Score: 4, Funny) by DECbot on Tuesday March 30 2021, @04:14PM
Finally, a feature to keep my coffee at temp no matter what climate I might be at today. Now I just need to incorporate the continuous brewing module to keep the revivor topped off and a monitor to periodically dump coffee when it detects that it is too old/burnt.
cats~$ sudo chown -R us /home/base
(Score: 2) by bzipitidoo on Tuesday March 30 2021, @07:14PM (6 children)
The problem has always been too much heat. For years, I've had the impression that the electronics could safely be submerged in liquid nitrogen. It's not just that they can take it, it's that they like it. Your computer could run faster if it was in a nice cold bath of liquid N.
(Score: 5, Informative) by sjames on Tuesday March 30 2021, @08:28PM
Too cold has always been a problem, it just doesn't come up that often on things like desktop PCs. Tolerances for component values including resistance, capacitance, and switching times are only valid within the specified temperature range on the data sheet.
Consider what happens in a digital circuit if a latch meant to trigger in the middle of a square clock pulse is slow and instead latches at some point on the falling edge (it will essentially randomize the values).
Depending on the device, you might or might not get away with operating below the minimum temperature if you slow the clock.
The cooling the processor with liquid nitrogen stunt never actually cools the chip that cold due to limited thermal conductivity and dissipated heat from the chip. But it does make an impressive cloud of fog.
(Score: 2, Informative) by Anonymous Coward on Tuesday March 30 2021, @08:30PM
Your impression is quite mistaken.
Most electronic components have characteristics that depend on temperature. For example, the forward voltage across a silicon diode is normally described as a function of temperature and current. If the current remains constant, as temperature decreases, the forward voltage across the diode will increase. Especially for low voltage stuff, such changes could violate design assumptions elsewhere in the circuit, leading to incorrect operation. These kind of electrical problems are usually not immediately obvious and may not happen all the time, perhaps if you assemble 1000 "identical" boards you might find 950 of them will work correctly at -30°C yet all 1000 work at -10°C.
This is precisely why almost every electronics device is designed for use within a specified operating temperature range (usually quite prominently displayed on the datasheet). While suitable parts might exist, most stuff is simply not designed, tested or intended to operate anywhere near the approximately -200°C of liquid nitrogen.
Worse, mechanical stresses due to thermal expansion can be a big deal, and you often find storage temperature specified on components too. If you dunk the board into a -200°C nitrogen bath I would be very suprised if all the ceramic capacitors survive the experience. You're likely to crack some solder joints too.
As this is specifically about an entire board, you have to consider the temperature ratings of every component on that board. If you want something to run in a -30°C, you will either have to select components that are all rated for this temperature, or do your own measurements to use parts "off label" (usually not practical for mass production), or (as is the case in this article), use a heater to bring things into the expected range.
(Score: 1, Interesting) by Anonymous Coward on Tuesday March 30 2021, @10:32PM
You really don't want to use cryogenic cooling with newer hardware. Lead free solder slowly goes to powder when it gets cold due to a phase change in the tin.
(Score: 2) by crafoo on Wednesday March 31 2021, @12:05AM
all kinds of really bad things happen down at -30C and below. -55C is a test point limit on many aerospace applications. look at many datasheets for commercial components and the best you usually find is -40C. Capacitors have a bad time as well as a few other typical COTS components when it gets cold. don't know about processors though.
(Score: 1) by The Mighty Buzzard on Wednesday March 31 2021, @01:11PM
Industrial settings. Having a SBC that will keep itself above the minimum operating temp saves you having to build a heated enclosure, simply water/insect/ratproof will do if you've already got the max temp dealt with. This is a big thing to me since I'm monkeying with exactly that lately.
My rights don't end where your fear begins.
(Score: 2) by VLM on Wednesday March 31 2021, @03:06PM
Nobody gave a practical answer so I'll try
Ever since CMOS chips have been made people have been Fing around dunking them in liq N2 to see what happens.
From memory the ETA10 supercomputer generally made things run 4x faster, just dunk them and magically quadruple the speed.
Imaging chips don't have the slightest problem and perform much better at Liq N2 temps plus or minus the nightmare of water condensation.
If you dunk a schmidt trigger digital input chip into liq N2 as part of an experimental apparatus I sadly don't remember what happened exactly but generally speaking it F-d up somehow, like the hysteresis voltages all wandered around not just a speed change. I don't recall turning it into an oscillator but it definitely didn't do the hysteresis thing correctly anymore.
The biggest problem is you make logic 4x faster but delay lines and stuff remain the same so "marginally stable" designs are not any more.
Remember from high speed bus design problems that the biggest problem with bus reflections and stuff is the rate of change of voltage not the clock speed. So simply making everything half a dozen times snappier means your high speed bus has to be better terminated.
I bet if you dunked a computer mouse in liq N2 the on board microcontroller would continue to work just fine but it would no longer meet USB spec WRT but performance issues. I bet you could dunk a bluetooth mouse and it would work. I almost guarantee a 1980 S-100 Z80 computer would stop working the bus would be too noisy they barely worked under ideal conditions. I bet a ISA legacy AT 16 bit motherboard bus dies a LONG time before a PCI express motherboard.
Searching youtube for "liquid nitrogen ttl cmos" found me NOTHING. I should record some stuff. Its not expensive, that's for sure. It would be amusing to hook up a 74hct000 nand gate up as an inverter and just beat the hell out of it using my function gen and scope and then see how performance changes when dunked in dry ice and liq n2.
One interesting problem is cryogenics don't have the heat capacity of something like distilled water. A 100 watt pentium 4 at 60 C is still going to be 60 C no matter if its in air or vaporized N2. I think you'd need a firehose or rocket pump to keep a server class processor all the way down to liq N2 temps.
(Score: 5, Funny) by dltaylor on Tuesday March 30 2021, @08:33PM (4 children)
Many years ago, I did some work for an In-Flight Entertainment company. Because airliners often sit overnight in places like Alaska, -30C was a requirement for all of the electronics aboard, flight-related or not.
When management opted for a less-robust processor in one of the subsystems, at -30C, its clock didn't run. We were asked to "fix it in software".
(Score: 0) by Anonymous Coward on Tuesday March 30 2021, @08:44PM (3 children)
> We were asked to "fix it in software".
OK, I give up -- please supply the rest of the story. What did you (your team) actually do?
(Score: 2) by Freeman on Tuesday March 30 2021, @10:10PM
Spent a week or two trying to do what the boss said, then spent the money to fix the problem, probably?
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 3, Funny) by DECbot on Tuesday March 30 2021, @10:34PM (1 child)
I hope it goes something like:
cats~$ sudo chown -R us /home/base
(Score: 3, Informative) by dltaylor on Tuesday March 30 2021, @10:58PM
Can't do that. A parked airliner is usually completely powered off by just throwing the big switch. On-board electronics are required to just "deal with it".
As far as what they finally did, that was about the time I was let go for once again pointing out the obvious to senior staff and management who didn't want to hear it.
(Score: 3, Interesting) by dwilson on Tuesday March 30 2021, @10:28PM (4 children)
And I live and work in Canada.
This winter, I put a RPi4 out in the barn I winter my livestock in (leafcutter bee cocoons), hooked to a few 1-wire temp sensors and recording data every ten minutes with rrd.
The insulated room with the sensors and the bees is heated, kept at about 4 celcius. The pi sits on a shelf outside that room, unheated except for leakage.
Ironically, when we got a cold snap and the outdoor temperatures dropped to the -40c range, the pi never missed a beat. It faithfully recorded temperatures, every ten minutes, like clockwork.
The outdoor-mounted, -30c-rated Wifi Bridge I'm using to link the pi to my lan, however, tends to stop working at about -15c.
Funny old world, sometimes. Maybe I just got lucky with that specific Pi.
- D
(Score: 2) by sjames on Wednesday March 31 2021, @12:28AM
The Pi I have idling right now is running about 30C above room temperature. That probably helped, though some luck was probably also involved since that would still leave it at -10C.
Do not power the Pi off when it's that cold or you may have to wait for things to warm up to start it again.
(Score: 2) by Marand on Wednesday March 31 2021, @01:32PM (2 children)
Out of curiosity, why a Pi4 for that use case? Seems overpowered and unnecessarily expensive for what you did with it. The Pi Zero (W) boards are perfect for that kind of thing, since they're cheap as chips (heh) and still powerful enough to handle most workloads of that sort. I'd rather put a $5/$10 (depending on wifi needed or not) chip in an electronics-hostile environment in case something happens to it.
(Score: 2) by dwilson on Wednesday March 31 2021, @11:43PM (1 child)
Used an RPi4 because I had one kicking around being unused. If I'd had to buy all the bits and bobs up front, I'd have used with a Pi Zero, yeah.
- D
(Score: 2) by Marand on Thursday April 01 2021, @07:55AM
That makes sense. Even though I don't have an immediate use for them, I bought a Pi Zero W and a dirt-cheap ESP32 microcontroller a while back just to keep around so I'd have them if I need them. I figured I'd rather have and not need right now than need them later but not have them around. I know how I am, and if I don't have what I need handy I'm more likely to either not bother or do some stupid kludge instead of waiting for an order to ship. Whenever I eventually put one of them to a specific task I intend to immediately replace it with another so that the cycle can continue, always keeping a spare microcontroller and Pi Zero available.
Initially I was using the Pi Zero as a sort of general testing ground for RPi stuff, but I decided I was better off with something faster for that kind of thing and got a Pi 400 I keep at my desk and running. Bit more convenient to do things on the Pi400 instead of moving files to the Zero W after compiling them in a loopback mounted, chrooted raspberry pi disk image running on my desktop (using qemu-user-static to translate the ARM binaries in the chroot to amd64 on-the-fly), which is what I was doing before that.
(Score: 0) by Anonymous Coward on Wednesday March 31 2021, @01:18AM
"undisclosed pricing"
No price. || if you have to ask...
Sad.