Mass Memory Storage Devices in Vacumm

From CAPE Wiki

Most developers want to use commercial off the shelf parts. Unfortunately, there is very little data on specific COTS memory that can handle the CubeSat's environment.

Vacuum test results on SD cards would be great. The problem is that you're unlikely to be able to guarantee that the cards that you buy, based on the tests you find, are internally identical to the ones that were tested. The market moves very quickly.

NASA uses DSLRs in extravehicular operations (I'm certain they used film cameras like the F3 as well), and I'd be very surprised if said cameras are operated exclusively in a pressurized environment. The DCS 7xx series of cameras used PC Card CF, and the D2x series that are replacing them use CF. The difference between CF and SD should only be in the interface. The Flash memory and the charge pumps should be very similar. Developers want to know which of the Flash technologies is the least susceptible to radiation events. It's nearly impossible to buy microcontrollers these days that don't use Flash.

Luke Stras of Stras Space writes:

On projects that I have been involved in (UTIAS SFL, and later SpaceQuest) have flown both commercial flash (NAND and NOR), as well as commercial EEPROM; these have been in both our own custom-developed hardware (as processor memories), and in third-party hardware (again, for storage of embedded software). Unfortunately, we've seen Flash devices simply Go Away -- for example, we had strong evidence that a device with a two-stage bootloader (one part in either Flash or mask ROM, the other part in a separate Flash chip) was stuck in a continuous reboot loop, with the first-stage bootloader trying -- and failing -- to load application code from a Flash chip.

We've also had a fancy commercial solid-state drive simply erase itself, probably a radiation hit on the controller chip, which triggered the "secure format" functionality. The system limped along for a while, running out of cache RAM, but when we rebooted it to try to clear what we thought was a transient fault. Make sure that you've got some capabilities to switch to backup storage. You don't want your core spacecraft functionality running out of a single-string device!

Unfortunately, it is difficult to select devices for "reliability", as manufacturing processes can change drastically, with little notice to the customer. If you have the capability, time, budget, and, most importantly, inclination, to do this, you can try to do lot qualifications.

This may

(a) take some time, and

(b) be inconclusive; in our cases, the failures didn't occur until at least 6 months on orbit.

All that having been said, I know people who have flown standard SD cards as mass storage, with no problems observed after orbit-years. Once you settle on a design, buy enough parts to last you for the duration of the project.