How can I get my Pi 3 to work in a vacuum?

[FlimFlam]

Im trying to do some data logging, using a raspberry pi 3 would be super easy as it has a built in WiFi and Bluetooth receiver as well as an SD card. However when the vacuum gets to about -30inHg the Pi shuts off and the red light starts flashing as if there is no OS. Temperatures arent crazy high but I don't see why this is happening.

 

[f95toli]

Im trying to do some data logging, using a raspberry pi 3 would be super easy as it has a built in WiFi and Bluetooth receiver as well as an SD card. However when the vacuum gets to about -30inHg the Pi shuts off and the red light starts flashing as if there is no OS. Temperatures arent crazy high but I don't see why this is happening.

It is probably overheating. There is no convection meaning there is virtually no cooling.
Could you somehow thermally link the microprocessor to say the metal walls of the vacuum chamber?

 

[FlimFlam]

It is probably overheating. There is no convection meaning there is virtually no cooling.
Could you somehow thermally link the microprocessor to say the metal walls of the vacuum chamber?

I don't know if temperature sensors get messed up, but the web interface says the processor is at 92F before this happens. I could attach a heatsink to the processor and the walls of the chamber.

 

[Vanadium 50]

For the same reason you can't move heat off the chip, you can't move it to the sensor. This is almost certainly a cooling problem, and you need to do what f95toli said.

 

[rbelli1]

the processor is at 92F before this happens.

If that is the core junction temperature then it is something else overheating. Maybe the power supply?

BoB

 

[FlimFlam]

If that is the core junction temperature then it is something else overheating. Maybe the power supply?

BoB

Well I'm powering it from mains current trough USB. The power supply isn't inside the chamber.

Sent from my SM-G900P using Tapatalk

 

[Rive]

If that is the core junction temperature then it is something else overheating.

If it can produce an error message of a kind, then the CPU is working.

I think it will be the SD-card, since that's the one with the OS.

However, I would not rule out that it is something caused by the missing pressure. Some micro-SD cards are sealed but not filled, so it might deform.

 

[Vanadium 50]

If it can produce an error message of a kind, then the CPU is working.

but the web interface says the processor is at 92F before this happens

(emphasis mine)

 

[Rive]

Can't see the point, sorry.
According to the OP:

However when the vacuum gets to about -30inHg the Pi shuts off and the red light starts flashing as if there is no OS.

So AFTER 'this' happens the CPU is still working since it can produce an error message of a kind.

 

[f95toli]

So AFTER 'this' happens the CPU is still working since it can produce an error message of a kind.

This does not mean anything; microprocessors tend to have separate "watchdog" circuitry built in that monitors the actual CPU and shuts it down if discovers that it is about to overheat (more complicated watchdogs can be used to e.g. turn a sleeping CPU back on it detects that a key been pressed)

Hence, it is perfectly possible for e.g. a LED to blink while the actual CPU is completely turned off.

 

[Vanadium 50]

There are motherboards with LEDs that blink if the CPU isn't there at all.

 

[Rive]

Onboard watchdog or service chips usually does not check the availability of the OS. (Except some server boards, but that level of stuff gives you a serial terminal and tells you exactly where the problem is.)

If the OP is right with his statement that it's flashing like no OS present, then that flash is a quite good indication of the still working CPU.

 

[FlimFlam]

If it can produce an error message of a kind, then the CPU is working.

I think it will be the SD-card, since that's the one with the OS.

However, I would not rule out that it is something caused by the missing pressure. Some micro-SD cards are sealed but not filled, so it might deform.

Ive never had a problem with an SD card before. I use an SD camera and it works perfectly.

Onboard watchdog or service chips usually does not check the availability of the OS. (Except some server boards, but that level of stuff gives you a serial terminal and tells you exactly where the problem is.)

If the OP is right with his statement that it's flashing like no OS present, then that flash is a quite good indication of the still working CPU.

Raspberry pi stays in post code if it doesn't detect an OS> I've tried 3 different SD cards and still can't find the issue.

 

[Rive]

Raspberry pi stays in post code if it doesn't detect an OS...

Just out of curiosity I've tried to find something about the 'flashing red' and R-pi, but practically every source takes this as a far more basic (indeed: low level watchdog/supervisory class) error like missing or low power or such. Although these sources were not directly about R-pi 3, but none of them connected any red flashing to missing OS. Are you sure what you see has this meaning?

The most interesting source I found were blaming the polyfuse on the incoming power line to malfunction at slightly higher than average temperatures, resulting in a voltage drop and insufficient voltage feed for the onboard PSU, triggering an undervoltage error (error LED flashing red). However, the CPU is expected to be fine up to 70-80 C core temperature, far higher than mere 92F.

 

[Vanadium 50]

Is this just the usual PF quibbling? Or has someone else actually worked with electronics in vacuum, and has reason to suspect that the overheating answer is wrong?

 

[Rive]

has reason to suspect that the overheating answer is wrong?

The suspected component had a temperature measurement (92F) far below the known limits (over 200F).
By my comprehension in such situation to blindly push for more cooling of the mentioned component is the exact case of the 'law of the hammer'.

I don't know about quibble.

 

[FactChecker]

Is there any vibration from the pump (or airflow) that might be causing an intermittent connection problem? Try a vibration test. Also, check any power connections to it in the test stand. Other than that, if it works reliably when it's not in the test stand, I would suspect a cooling problem as others have suggested.

 

[CWatters]

Try heating it to the same temperature out of the chamber to see if it behaves the same?

 

[RonL]

Im trying to do some data logging, using a raspberry pi 3 would be super easy as it has a built in WiFi and Bluetooth receiver as well as an SD card. However when the vacuum gets to about -30inHg the Pi shuts off and the red light starts flashing as if there is no OS. Temperatures arent crazy high but I don't see why this is happening.

Not sure what you are interested in, but here's a link that with it's 207 pages, might answer or give a clue about how the Pi-3 is different from what is used in space (or vacuum conditions)

http://snebulos.mit.edu/projects/reference/International-Space-Station/SSP30312RH.pdf

 

[halfapizza]

Is this just the usual PF quibbling? Or has someone else actually worked with electronics in vacuum, and has reason to suspect that the overheating answer is wrong?

This is years late, I know. Your convection comment is spot on. I work in vacuum often and this is a common challenge.

 

[Vanadium 50]

So hoe does one fix this? A thermal pad connected to a block of copper that pipes the heat out to where there is air? (Or my preferred solution - put the Pi ouside and run any cables through feed-throughs.)

 

[pbuk]

So hoe does one fix this? A thermal pad connected to a block of copper that pipes the heat out to where there is air? (Or my preferred solution - put the Pi ouside and run any cables through feed-throughs.)

I'd start with a microcontroller with a lot lower thermal demand: if a Pi Zero W is still too hot then ditch the wireless, you've already got power running in so add data, the more copper you have attached to the thing the better. Or maybe try a Seeed XIAO.

Then yes, thermal coupling to something: with a sensibly low powered board I'd be surprised if you need to go to the outside, unless the walls of the chamber are not thermally conductive.