3 ms·

Now I'm curious what other parts of this device could last 'over 200 years.' Whats the next weak link on the chain? Is contact corrosion the ultimate limiting

by cdchn 2y ago

Now I'm curious what other parts of this device could last 'over 200 years.' Whats the next weak link on the chain? Is contact corrosion the ultimate limiting factor?

vlovich123 2y ago
> It incorporates a Raspberry Pi RP2040 microcontroller, making it suitable for various secure storage applications in case it can fit into a tiny amount of storage space. I'm going to make a safe bet that there's no way any of the capacitors involved survive anywhere near that long. Now of course, technically the data is still there, but I doubt it in 200 years it's going to be all that easy to replace the microcontroller while retaining that data. A competitor in this race would be the connector: > The Blaustahl includes 4MB of NOR flash for firmware and a USB Type-A male port that requires no additional drivers on most operating systems. Something tells me that USB type-A is unlikely to survive the next 200 years.
- AshamedCaptain 2y ago
  Ironically by having its firmware in its NOR flash they are now subject to the same lifetime limitations.
  - chrisjj 2y ago
    I guess we should be grateful they eschewed the RPi preference for critical firmware in SD card :)
  - cdchn 2y ago
    I'd be really cool to have the RPi firmware in FRAM as well.
- nottorp 2y ago
  Are you sure? You^H^H^H your grand-grand-grand-grand children will probably be able to get an usb-g to usb-f to usb-e to usb-d to usb-c to usb-a adapter chain. Hey, what happened to usb-b?
  - freeone3000 2y ago
    A is the “host” side and B is the “device” side of that connector. A full-size USB-B connector is most commonly seen on printers; micro-B for old android phones.
  - LeonidasXIV 2y ago
    USB-B is the thing that's on the other side of cables with USB-A. A large, square-ish port that got even larger when it was extended for USB 3, with a connector that got even larger. That's the port that mini-USB and micro-USB are mini and micro versions of. You can find USB-B ports on e.g. laser printers (where size doesn't matter) or on some older USB drives (where you'd want USB 3 speeds).
    - nottorp 2y ago
      Oh right. I've had it for printers but I already forgot. I think i even had a sata to usb box using the huge usb 3 version. Those were fine-ish, but their micro versions can't die in a nuclear explosion fast enough...
  - kibwen 2y ago
    USB-B is the other end of a USB-A cable that plugs into the peripheral rather than plugging into the computer. But peripherals all started getting so small that they switched to USB-mini/USB-micro, and the only large peripherals left are printers, which is pretty much the only place you'll see it used. https://en.wikipedia.org/wiki/USB_hardware#Connectors https://en.wikipedia.org/wiki/USB_hardware#Connectors
  - vlovich123 2y ago
    Yes. It's important to remember that we've already lost connectors from < 100 years ago. Now USB-A does have a lot more mass production and popularity for it today which does have something going for it. But remember that changeovers happen quickly and 200 years is enough time that anyone who cares about the legacy dies and there's no market to sustain the adapters. 200 years from now you'd expect all physical connectors to have disappeared with wireless variants taking their place.
- twarge 2y ago
  Ceramic capacitors should be fine, right? The pico microcontroller, on the other hand, has normal flash and won't last very long. Here's an interesting article: https://www.ni.com/en/support/documentation/supplemental/12/understanding-life-expectancy-of-flash-storage.html https://www.ni.com/en/support/documentation/supplemental/12/... Maybe keep it in the freezer? Otherwise I would expect tin whiskers or some other process to dominate reliability at long periods.
  - vlovich123 2y ago
    Can't find info on what kind of capacitors they use but [1] suggests certain kinds of ceramics can only last ~10 years. More generally, all electronic components can fail [2] and there's a lot of such components in this device. This is applying component-level thinking to a systems level problem with no real way to verify their claim (they don't even claim to have done any aging test to try to simulate how long the device could last) and why this should be called out as nothing more than a marketing gimmick. [1] https://www.edn.com/class-2-ceramic-capacitors-can-you-trust-them/ https://www.edn.com/class-2-ceramic-capacitors-can-you-trust... [2] https://en.wikipedia.org/wiki/Failure_of_electronic_components https://en.wikipedia.org/wiki/Failure_of_electronic_componen...
- crote 2y ago
  > I'm going to make a safe bet that there's no way any of the capacitors involved survive anywhere near that long. Why? All capacitors on the flash drive are MLCCs, which unlike their electrolytic counterparts aren't exactly known for failing - let alone simply due to age. When push comes to shove you can just remove them without meaningfully impacting functionality anyways.
  - vlovich123 2y ago
    Both [1] and [2] suggests that the lifetime of ceramic capacitors is generally limited. I can't find details on whether the Pi uses type 1 or type 2, but my guess would still be the capacitor with other components within the CPU failing before the 200 years. The USB would definitely still fail (physically or logically) before the 200 years is up. [1] https://page.venkel.com/hubfs/1_Images/Resource%20Center/Cross%20Reference%20Images/Cross%20Reference%20Guide%20(PDFs)/Cheat%20Sheets%20(PDFs)/storage-and-shelf-life-mlcc-components-venkel-ltd.pdf https://page.venkel.com/hubfs/1_Images/Resource%20Center/Cro... [2] https://www.edn.com/class-2-ceramic-capacitors-can-you-trust-them/ https://www.edn.com/class-2-ceramic-capacitors-can-you-trust...
crote 2y ago
Possible failure modes: - ESD / supply voltage spikes. This device has basically zero protection against them. The RP2040 is well-known to be quite sensitive to them, so I'd expect it to kill the controller relatively quickly. The FRAM chip itself isn't as exposed, but could still be damaged. - USB-A becoming obsolete. I fully expect USB-C to replace it on new machines within a decade - and we'll probably switch to something completely novel within the next 50 years or so. You'd have to get a PC out of a museum to access it. - Flash failure. The data is stored on an FRAM chip, but the controller's firmware isn't. That chip is rated for 20 years (but I do expect it to last quite a bit longer). - Crystal oscillator failure. It's literally a crystal, which is vibrating. They can fail. - LDO fail-closed, exposing the MCU, flash, and FRAM chips to an overvoltage. - Tin whiskers. The device is almost certainly using lead-free solder, which means it is susceptible to the solder growing tiny "spikes", which can lead to shorts. We've gotten quite good at managing these over the years, but all bets are off when it comes to a 200-year scale. Most of those can probably be recovered from, as you'd still be able to attach debugging tools directly to the chip and bypass the controller. The biggest risk is probably losing the documentation, which makes accessing it an awful lot harder.
qwery 2y ago
I'd say mechanical damage would be, by far, the most likely cause of catastrophic failure. Loss of data will occur when the device is lost or discarded.
- cdchn 2y ago
  Assume the device is in a time capsule, not rattling around on someone's keychain for 200 years.

by cdchn 2y ago

Now I'm curious what other parts of this device could last 'over 200 years.' Whats the next weak link on the chain? Is contact corrosion the ultimate limiting factor?

vlovich123 2y ago

> It incorporates a Raspberry Pi RP2040 microcontroller, making it suitable for various secure storage applications in case it can fit into a tiny amount of storage space. I'm going to make a safe bet that there's no way any of the capacitors involved survive anywhere near that long. Now of course, technically the data is still there, but I doubt it in 200 years it's going to be all that easy to replace the microcontroller while retaining that data. A competitor in this race would be the connector: > The Blaustahl includes 4MB of NOR flash for firmware and a USB Type-A male port that requires no additional drivers on most operating systems. Something tells me that USB type-A is unlikely to survive the next 200 years.

AshamedCaptain 2y ago
Ironically by having its firmware in its NOR flash they are now subject to the same lifetime limitations.
- chrisjj 2y ago
  I guess we should be grateful they eschewed the RPi preference for critical firmware in SD card :)
- cdchn 2y ago
  I'd be really cool to have the RPi firmware in FRAM as well.
nottorp 2y ago
Are you sure? You^H^H^H your grand-grand-grand-grand children will probably be able to get an usb-g to usb-f to usb-e to usb-d to usb-c to usb-a adapter chain. Hey, what happened to usb-b?
- freeone3000 2y ago
  A is the “host” side and B is the “device” side of that connector. A full-size USB-B connector is most commonly seen on printers; micro-B for old android phones.
- LeonidasXIV 2y ago
  USB-B is the thing that's on the other side of cables with USB-A. A large, square-ish port that got even larger when it was extended for USB 3, with a connector that got even larger. That's the port that mini-USB and micro-USB are mini and micro versions of. You can find USB-B ports on e.g. laser printers (where size doesn't matter) or on some older USB drives (where you'd want USB 3 speeds).
  - nottorp 2y ago
    Oh right. I've had it for printers but I already forgot. I think i even had a sata to usb box using the huge usb 3 version. Those were fine-ish, but their micro versions can't die in a nuclear explosion fast enough...
- kibwen 2y ago
  USB-B is the other end of a USB-A cable that plugs into the peripheral rather than plugging into the computer. But peripherals all started getting so small that they switched to USB-mini/USB-micro, and the only large peripherals left are printers, which is pretty much the only place you'll see it used. https://en.wikipedia.org/wiki/USB_hardware#Connectors https://en.wikipedia.org/wiki/USB_hardware#Connectors
- vlovich123 2y ago
  Yes. It's important to remember that we've already lost connectors from < 100 years ago. Now USB-A does have a lot more mass production and popularity for it today which does have something going for it. But remember that changeovers happen quickly and 200 years is enough time that anyone who cares about the legacy dies and there's no market to sustain the adapters. 200 years from now you'd expect all physical connectors to have disappeared with wireless variants taking their place.
twarge 2y ago
Ceramic capacitors should be fine, right? The pico microcontroller, on the other hand, has normal flash and won't last very long. Here's an interesting article: https://www.ni.com/en/support/documentation/supplemental/12/understanding-life-expectancy-of-flash-storage.html https://www.ni.com/en/support/documentation/supplemental/12/... Maybe keep it in the freezer? Otherwise I would expect tin whiskers or some other process to dominate reliability at long periods.
- vlovich123 2y ago
  Can't find info on what kind of capacitors they use but [1] suggests certain kinds of ceramics can only last ~10 years. More generally, all electronic components can fail [2] and there's a lot of such components in this device. This is applying component-level thinking to a systems level problem with no real way to verify their claim (they don't even claim to have done any aging test to try to simulate how long the device could last) and why this should be called out as nothing more than a marketing gimmick. [1] https://www.edn.com/class-2-ceramic-capacitors-can-you-trust-them/ https://www.edn.com/class-2-ceramic-capacitors-can-you-trust... [2] https://en.wikipedia.org/wiki/Failure_of_electronic_components https://en.wikipedia.org/wiki/Failure_of_electronic_componen...
crote 2y ago
> I'm going to make a safe bet that there's no way any of the capacitors involved survive anywhere near that long. Why? All capacitors on the flash drive are MLCCs, which unlike their electrolytic counterparts aren't exactly known for failing - let alone simply due to age. When push comes to shove you can just remove them without meaningfully impacting functionality anyways.
- vlovich123 2y ago
  Both [1] and [2] suggests that the lifetime of ceramic capacitors is generally limited. I can't find details on whether the Pi uses type 1 or type 2, but my guess would still be the capacitor with other components within the CPU failing before the 200 years. The USB would definitely still fail (physically or logically) before the 200 years is up. [1] https://page.venkel.com/hubfs/1_Images/Resource%20Center/Cross%20Reference%20Images/Cross%20Reference%20Guide%20(PDFs)/Cheat%20Sheets%20(PDFs)/storage-and-shelf-life-mlcc-components-venkel-ltd.pdf https://page.venkel.com/hubfs/1_Images/Resource%20Center/Cro... [2] https://www.edn.com/class-2-ceramic-capacitors-can-you-trust-them/ https://www.edn.com/class-2-ceramic-capacitors-can-you-trust...

crote 2y ago

Possible failure modes: - ESD / supply voltage spikes. This device has basically zero protection against them. The RP2040 is well-known to be quite sensitive to them, so I'd expect it to kill the controller relatively quickly. The FRAM chip itself isn't as exposed, but could still be damaged. - USB-A becoming obsolete. I fully expect USB-C to replace it on new machines within a decade - and we'll probably switch to something completely novel within the next 50 years or so. You'd have to get a PC out of a museum to access it. - Flash failure. The data is stored on an FRAM chip, but the controller's firmware isn't. That chip is rated for 20 years (but I do expect it to last quite a bit longer). - Crystal oscillator failure. It's literally a crystal, which is vibrating. They can fail. - LDO fail-closed, exposing the MCU, flash, and FRAM chips to an overvoltage. - Tin whiskers. The device is almost certainly using lead-free solder, which means it is susceptible to the solder growing tiny "spikes", which can lead to shorts. We've gotten quite good at managing these over the years, but all bets are off when it comes to a 200-year scale. Most of those can probably be recovered from, as you'd still be able to attach debugging tools directly to the chip and bypass the controller. The biggest risk is probably losing the documentation, which makes accessing it an awful lot harder.

qwery 2y ago

I'd say mechanical damage would be, by far, the most likely cause of catastrophic failure. Loss of data will occur when the device is lost or discarded.

cdchn 2y ago
Assume the device is in a time capsule, not rattling around on someone's keychain for 200 years.