Rad-Hard ARM Microcontrollers, Because Ceramic Components Are Just Cooler

If you’re building a cubesat, great, just grab a microcontroller off the shelf, you probably don’t need to worry about radiation hardening. If you’re building an experiment for the ISS, just use any old microcontroller. Deep space? That’s a little harder, and you might need to look into radiation tolerant and radiation hardened microcontrollers. Microchip has just announced the release of two micros that meet this spec, in both radiation-tolerant and radiation-hardened varieties.

The new devices are the SAMV71Q21RT (radiation-tolerant) and the SAMRH71 (rad-hard), both ARM Cortex-M7 chips running at around 300 MHz with enough RAM to do pretty much anything you would want to do with a microcontroller. Peripherals include CAN-FD and Ethernet-AVB, analog front-end controllers, and the usual support for I2C, SPI, and other standards. This chip does it in space, and comes in a ceramic quad flat package with gold lead frames. These are beautiful devices.

Microchip has an incredible number of space-rated, rad-hard hardware; this is mostly due to their acquisition of Atmel a few years ago, and yes, it absolutely is possible to build a rad-hard Arduino Mega using the chip, space rated.

Of course, there are very, very, very few people who would actually ever need a rad-hard microcontroller; I would honestly expect this to be relevant to only one or two people reading this, and they too probably got the press release. If you’ve ever wanted to build something that goes to space, and you’d like to over-engineer everything about it, you now have the option for an ARM Cortex-M7.

57 thoughts on “Rad-Hard ARM Microcontrollers, Because Ceramic Components Are Just Cooler

    1. Microsemi is really cool, but I am hoping that the microchip acquisition will relax some of the just plain weird restrictions and NDA requirements for datasheets and information.

      Ethernet switches and PHY’s don’t really seem that special as to require an NDA for anything more than the summary datasheet. :/

      1. Accidentally reported this post.
        Those chips are a thing of beauty!
        I wonder if these were used by in the remote robots used to explore the Fukushima Daiichi reactor disaster.

        1. If you would use a chip like this exposed to a nuclear reactor it would still fail very fast. The radiation intensity is orders of magnitude stronger than in outer space.

      1. In contrast to what some documentaries would have you believe, magnetic pole reversal will not be much of an event, even if it happens quickly. In fact, it’s very unlikely it will do much more than a regular solar storm might. The earth’s magnetic poles are always moving and that process seems to be accelerating. But moving is not the same as weakening.

        Geological conditions may cause the field to weaken somewhat, but the atmosphere is more than thick enough to pick up the slack.

        In the end, even the ‘nightmare scenario’ is not likely to cause more problems than a regular solar storm.

  1. Here we should probably point out that radiation tolerant and hardened chips are useful for more then just deep space applications.

    But also for those situations when one deals with other more direct sources of high radiation. Like near a Nuclear reactor. (like in nuclear power plants, or in application around handling the cleanup of failed reactors. (Or just for applications that should work in a nuclear wasteland if such a war ever breaks out…))

    Though, I wouldn’t be surprised if radiation hardening could also be needed around particle accelerators, synchrotron light sources, and other similar high energy devices.

    But even these down to earth applications might still be a bit out there as far as the everyday electronics tinkerer might be concerned.

    1. I run an experimental fusor – which makes a lot of neutrons, and a pretty good amount of EMI.
      While I don’t use rad-hard stuff in the data acquisition, I have found that things near the core do eventually fail, and that the digital stuff with larger geometries handles it a lot better.
      Things like an Intel NUC – there’s no point having within some feet of the thing, they fail right away.
      A raspberry pi (older models) with external storage rather than SD card…pretty decent, generally killed by some EMP that makes it past all the faraday-cage type shielding, at least so far. Arduino Unos seem to be everything-proof in this environment – I have taken strong measures around RC filtering inputs that go outside the box and Arduinos are what is used to initially collect all the data – the pi is just there to be a linux machine, run linuxy stuff (like a control GUI and VNC, database drivers) and do networking, in this case with a fiber optic translator for better immunity to electrical noise.
      For those who didn’t think of it – there are some differences between cosmic rays and a neutron flux. The main one is transmutation in the case of silicon, not just injecting some charge and flipping a bit with the other damage being incidental. Cosmic rays generally are low flux but far higher energy each versus high flux neutrons at a “mere” couple of MeV.
      Transmuting in place of silicon to phosphorus is what seems to kill smaller geometry transistors the quickest here.

  2. The problem is that for commercial satellites FPGA is way more valuable since you can implement redundancy and 3oo1 voting for functions. Microcontroller will therefore stay a nieche in the low-end market for satellites.

    They seem to be ITAR free. Big plus.

    1. FGPAs for rad environments is not that great. All that 3 for 1 redundancy goes out the window when you realize that the chip can reconfigure its itself in unpredictable ways. The only FPGA qualified for rad hardness (no upsets) are crazy expensive, $16K for a device barely able to hold an 8 bit processor. Some rad tolerant (meaning they don’t latchup and burn themselves out but do freak out) need external devices to keep them going and the upset rate can be unacceptable for some applications. We ended up using the hard devices for mission critical stuff and the tolerant devices for communications.

    2. See, I’m confused by Microchip’s choice here too. I’ve always felt that the Cortex M7 series of processors was a little bit out of place. It’s got an architecture that can be clocked faster than a Cortex M4 but typically much worse performance per Watt to go with it (at least how they’re normally optimized they draw more power while idle). Most applications that would be mission critical and where power doesn’t matter may be more suited to the redundancy provided Cortex R series processor.

      I feel like an M4 core or the newer M33 (which can be clocked faster and have lower power) would have been a better fit for this. Maybe they started the design on this before the M33 was ready? Maybe they had a partnership with a company taking the chip to market that specified the features in the chip?

      Can anyone chime in with an application where the radiation hardening would be beneficial yet where redundancy and power aren’t critical? It’s my understanding that a more efficient chip would be more suited to space since dissipating waste heat is such a pain. A missile? Hardwired terrestrial applications?

      1. I’m betting that it was the best (or best-supported) core available when they started development. They don’t have to be terribly aggressive about optimizing for power to beat the hell out of the legacy parts on the market. This thing is quoting 90 mA at 1.2 V to the core in active mode at 300 MHz, compared to multiple watts for older parts that it might replace.

      2. This chip has been in development for a long time by the French Atmel team.

        My team were considering it for an Airbus flight actuator controller. Passenger jet applications that are SIL4 criticality always need known radiation characteristics and extensive analysis was undertaken for a number of devices in that Project. It was found to be incredibly robust but was down selected because it was overkill for price.
        You would be surprised at the amount of SEE possibilities there are at 35000ft.

        Just for a laugh, we calculated that if you took a 64GB memory card full of data and stuck it to the window of an aeroplane, you would likely have around 100 bytes corrupted after a 4 hour flight. This is when using the Airbus standard of calculating bit-flips per flight hour vs silicon nm technology.

        As for core choice, I used to almost cry at the prospect that when running a DO-178 process, we’d have to turn off code optimization both at compiler and processor to avoid any risk of side effects. Making the ARM core run so slow, usually TI/Freescale power architecture win out.
        All great fun :-)

        1. I am curious on your calculations. ‘100 bytes’, was that 800 corrupted bits, or 100 upsets in total? Also did your math use just the raw upset rate, or did you also model the Error Correcting Code used by the card’s NAND controller?

        2. Hey,

          Your post is very interesting since I’m the project leader on this project, it’s true we’ve been developing it for years, first this was an automotive project and we worked hard at every level to make it radiation tolerant. I’m very impressed by all the comments here, rich and full of great explanations.


  3. I was out backpacking once. I came across an abandoned Uranium mine. Of course I had to do a little spelunking. My GPS was going bonkers after that. I had to buy a new one.

  4. Just for kicks I tried to find pricing for the SAMV71Q21RT. On microchip direct I got “This Device is currently not available for Purchase Online.” I hate it when companies do a “contact us” and not openly display the price, although to be fare in this case, this is a bit of a niche product.

    1. That’s pretty normal for rad-hard parts, and it’s awfully annoying when you just want to do a quick “does this make sense to pursue” survey for your cubesat design. The only microcontrollers I’ve found public pricing for are from Vorago (Cortex M0 with the weirdest peripheral set I’ve ever seen, shitload of timers and some serial comms but no internal clock, nonvolatile memory, or analog anything for about $700 each, and distributed on Mouser) and TI (hardened MSP430FR5969 for around $3000, minimum quantity 10).

    2. Because it is rad hardened it would be an ITAR item, so they need you to sign on the dotted line if they can legally sell it to you before they can sell it to you and that price would change depend on the legal documentation required for the destination country.

    1. Genuine question: how much actual logic do you have dangling in reach of significantly raided radiation levels? I can understand a bunch of sensors but I’d expect most of the logic to read those (including anything that’s not the actual transducer or w/e itself) to just be well out of the way – how wrong am i!?

      1. Sounds like you are asking how much digital computer equipment is exposed to significant amounts of radiation. I can speak most affirmatively for modern nuclear plant designs:
        Under typical power plant conditions, all the logic happens well away from areas with significant radiation dose rates. Only instruments and their transmitters, and equipment deemed Environmentally Qualified (read: appropriately tested with a restricted service life under a specific range of conditions) are located in containment where most of the higher radiation dose rates exist. Logic usually occurs in computer rooms that also tend to occasionally have engineering and maintenance personnel, and so in an effort to keep doses As Low As Reasonably Achievable (“ALARA”) to people using the principles of Time, Distance, and Shielding (especially those last two here), modern plants are designed to maintain those rooms included among the low dose areas.

        1. Sort of what I was asking, sorry, I was a bit ambiguous.
          Thanks for taking the time in any case!
          I understand the control logic is kept well away from ‘hot’ areas of course. My question was if it’s commonplace for such instruments and equipment as you mention to have their “logic” (sorry, I meant any digital logic like these microcontrollers there; I can see how that was ambiguous!) located around those higher radiation dose rates. I kind of figured constructing something like a nuclear power plant would allow for simply shielding most of that stuff away as well. I gather that’s not usually the case and these uC’s would get used and simply be exposed to those higher doses as well? That satisfies my idle curiosity, so thanks again! ;)

  5. Hey HaD. Know anyone who can do a description of the IC’s and other devices on Voyager? I would guess we are talking about the old days with features measured in microns.

    1. Given that they were launched in 1977, I’d say most of it was done with TTL level chips. According to this
      The computer control was from the Viking lander.
      Here is a description
      he Command Computer’s central processor contained the registers, data path control and instruction interpreter76. The machine was serial in operation, thus reducing complexity, weight, and power requirements. It had 18-bit words and used the least significant 6 bits for operation codes and the most significant 12 for addresses, as numbered from right to left. This permitted 64 instructions and 4K of direct addressing, both of which were fully utilized. Data were stored in signed two’s complement form, yielding an integer range from -131,072 to +131,071. Average instruction cycle time came to 88 microseconds. Thirteen registers were in the Command Computer, mostly obvious types such as an 18-bit accumulator, 12-bit program counter, 12-bit link register that pointed to the next address to be read, and a 4-bit condition code register that stored the overflow, minus, odd parity, and nonzero flags77.

      I am guessing features measured in inches…

    1. If you don’t work for a major aerospace company they won’t even return your calls. The market is such a tiny niche that only a handful of companies even bother (some only do it because the government pays them stay in). 30 years ago it was the complete opposite, vendors knocking on your door to get into the booming cold war business.

      1. I have purchased radiation hardened ICs before for work for an X-Ray project, it’s not all that difficult and you can use the regular order form to do it as long as your in the United States. Radiation Hardened parts don’t raise any alarm bells and i had them shipped to my house no problem.

        The market for them is much more significant than you think, space is one use and many small satellite companies exist, also universities do satellite projects and even in some cases its a course requirement. You can also use these chips for any generic radiation heavy environment from reactors to X-Ray based systems where there is simply not enough room to properly shield the electronics compartment or simply cant because the part is under the sensor.

        This is not something you would use for a weapon, they are too heavily shielded and don’t stay in space for very long so that’s not an issue.

    2. These packages are called CQFP
      here are some designs by TI http://www.ti.com/lit/an/snoa025/snoa025.pdf
      Its simply a QFP package with ceramic on the base and gold plated tin or nickel on the top that are still in the lead frame and not shaped.
      You can shape them how you please with your own jig, cut a hole in the PCB and put them in sinked in to the PCB, put them in upside down, however you please. Its also possible to sandwich the chip between 2 PCBs and bend some leads up and some down.

      These are parts you typically use in very low volumes at very high cost so manually shaping them is not a problem.

      There are other similar packages like CERQUAD

  6. I spent most of 2017 trying to buy a Atmel (now Microchip) rad hard microcontroller,
    the AtmegaS128. Really frustrating. The 128 was $4, the S128 was $2700. The nice
    people in Singapore always substituted the cheaper part for the more expensive part.
    I even had the tech guys in town (Denver) trying. No joy. We flew with what we could get.

    I later heard that someone was buying them all up and shipping them overseas labled
    as LCD controllers. Jail time there.

    Really wish we got the parts. FCC screwed us in the end anyway, but It would have been nice to have long life for more lawyer options.


  7. Saying “just grab any old micro” is dangerous. Sure, LEO has less radiation than other parts of the solar system but it’s still enough for a single event upset/latchup. Is a single COTS micro an acceptable risk? You need to decide what your mission is worth.

  8. That’s not strictly true, if you’re inside the reactor core then it is but equipment isn’t actually put in there. What you do have to deal with is a radiation profile that’s very different to what you encounter in space, which is a huge amount of high-energy gamma and not much else, at least that you need to worry about. In nuclear you’ve got much lower-energy gamma, but need to worry about thermal and occasionally fast neutrons, which are really nasty. So a space-qualified rad-hard part isn’t necessary what you’d need tor nuclear use, or at least it isn’t automatically the perfect fit for that. A lot of nuclear equipment is relatively COTs components in custom hardware builds with TMR software driving it and similar, because rad-hard is too much of a hassle to work with.

  9. As a followup to that, equipment used around reactors isn’t necessarily rad-hard but easily replaceable and duplicated. So instead of using a $100K RAD750 device you use a generic industrial-grade equivalent costing a few hundred dollars and assume you’ll need to replace it based on a bunch of calculations around device lifetime and radiation levels. Then you run duplicates of everything, again based on calculations about both failing at the same time. The “duplicates” bit is always fun because it really means two physical copies of everything, all the way up and down the line. For example plugging the ethernet cables from the duplicated devices into a network switch or router is a no-no, because then you’re back to a single point of failure. Other failures are a lot less obvious, you invariably end up with some common point of failure somewhere that got overlooked during the planning process.

  10. Shielding against gamma is (relatively) straightforward, but shielding from neutrons is more or less impossible because of the sheer mass of shielding material required, your equipment would need a heavy-duty forklift to move a basic device like an assay meter around. So the usual practice is to make them radiation-tolerant, see an earlier post above.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.