Last time (Jupiter: The IC Danger Zone, Part 1) we looked at Total Ionizing Dose and Displacement Damage and how they relate to testing a waterproof fabric and the environment that the Juno probe will encounter in orbit around Jupiter. This month we’ll dive into the details of Heavy Ion Radiation, which is a different kind of radiation testing. Some of these failure modes are deal-breakers, some of them are not.
Heavy Ion Radiation
Sometimes, you may want to bring your high-performance waterproof fabric into a water gun fight. Let’s face it, being soaked can be a cold experience, so keeping yourself dry is important! And how terrible would it be if that fabric didn’t stop the intense pressure of being pelted with a Super Soaker!? Or, assuming you’re wearing a full body suit, what if you get shot in the eye or get the wind blown out of you? If a high-pressure water stream was enough to cause the fabric to fail and leak water to your skin, you’d be in for a cold battle. Heavy Ion testing puts ICs through a similar test.
In any, particularly CMOS, there are parasitic bipolar transistors which in normal operating conditions cannot ever be excited into latch-up on earth unless there is some sort of other catastrophic failure. However, in space there are free energetic particles floating around which could trigger a latch-up if they happen to hit in just the right place. As the particle punches through the device, it can ricochet off of the crystalline structure. Often latch-up occurs through a combination of the lateral and vertical parasitic transistors as the particles bounce around. Let’s look at a CMOS inverter as an example.
Latch-up current flow in a CMOS inverter
The parasitic BJTs are shown in red, and the latch-up current flow is shown by the blue arrows. It’s just like latch-up in any other situation, except now you have to worry about ions which can hit at random. If any high energy particle were to hit in just the right spot, it could trigger these BJTs to turn on in such a way that normally wouldn’t be possible. Since they connect VDD to VSS, the only way to stop this failure is to turn the inverter completely off. This is of course assuming that these kinds of huge currents don’t cause any fatal damage to the device! Since these high energy particles will hit at random times and on random parts of the die, any space application device must be checked that these latch-ups do not occur. And of course, this effect can be exacerbated by any lingering charges lodged in the IC from other ionizing radiation.
This is another important parameter to check for in any of Juno’s electronics. Much of the radiation which exists around the Red Giant is very high energy particles moving close to the speed of light. If any one of these particles were to barrel into a sensitive device, it could be catastrophic. That titanium box protects these sensitive devices from what their design modifications cannot.
Moving slightly away from our waterproof fabric example, let’s say you’re, again, in a water-gun fight. Someone nails you right in the face. Agh! That hurt! Your face probably got contorted for a few seconds, but then you were fine and back at it and ready to go. Interruptions to device function like this caused by radiation are referred to as Single Event Effects, or SEE. They go by different names depending on the device and interruption in question, but they are all caused by similar radiation events.
SEFI, or Single Event Functional Interrupt, is a term used for disruptions in functionality of a device, usually a data converter, where a register bit gets flipped by an injected ion. In this kind of interruption, an ion is injected into a register and gets trapped there, changing the register state, and thus the device operation. It could be as simple as putting a data converter into a different mode of operation, or it could affect whether the device is enabled or disabled. It is considered a soft failure if a device reboot or a register re-write clears the error. There are many methods used to prevent these kinds of soft failures in the field, including redundant memory and checksums, which are used to trigger a device reset or register re-write so that the IC can return to normal functionality. The equivalent term used for memory devices is referred to as a Single Event Upset, or SEU.
On any analog device, such as an amplifier, these ions can end up in the output-bound path of charge flow, causing a blip seen on the output as the injected ion is “flushed out” with whatever other charges are flowing with it. These are considered transient effects. They usually don’t require a reset of the device, and simply time will fix these errors, sort of like wringing out a water-logged shirt.
Failures of these types are not deal-breakers, but they should be highlighted and specified to customers so that they can be prepared. They are also more common in low supply voltage situations than in high supply, as there is less energy required to excite enough electrons to change a register state or cause the output to fluctuate. This is actually one of my favorite failure modes to look at because they are recoverable out in the field, and are not considered permanent damage. These kinds of blips in operation are one of the big reasons any satellite destined for space is designed with so much redundancy. A device which might be used in few numbers for an earthbound application could be used by the dozens in a satellite.
What Could These Transients Do to Juno?
SEFI and transient events, as mentioned, are not deal-breakers at all. However, the system designer must keep them in mind when coming up with a system to monitor the engines for instance, or read from a sensor. In many cases, simply having a duplicate (or several!) of the memory or registers or sensors in question stored somewhere and running a checksum or average frequently is usually enough to alert the control circuitry that there’s a problem. In addition, a simple power-cycle of both the device and the duplicate register memory and a re-program into the desired mode will correct the problem and get the sensor back up and running. System designers for space probes usually implement layers upon layers of redundancy in case something catastrophic happens. After all, it’s way easier to replace a defective device on Earth than it is in orbit or worse, out in deep space. If anything were to become damaged on an earth orbit satellite, it could be years before it gets fixed, if it ever gets fixed at all. And out in the solar system? Forget it!
Even for Juno, however, these kinds of events could even be considered data in themselves. Any time one of these transient faults happens, it of course means there is a lot of radiation in the atmosphere where Juno is performing scans. In fact, I wouldn’t be surprised if NASA intends to somehow collect this kind of information to further characterize the radiation belts around Jupiter. In fact, I recall reading somewhere that NASA will use the noise in photographs taken to help characterize the radiation belts. Now that they have years of data from Juno’s journey from Earth to Jupiter, they have a baseline noise floor without the intense radiation around the Red Giant to compare to. If you ask me, it’s quite an ingenious use of what is normally considered unwanted data.
Juno is approaching its final jet burn next month, which will put it into the 14 day science orbits. I’m excited to see what we can find out about the King of the Solar System!
Radiation Testing Documentation
Analog Devices offers radiation reports for many devices in our space portfolio. A good example is the ADA4077-2S, a dual low offset and drift, high precision amplifier that was recently released. For an example of a Single Event Effects report, check out the product page here . And be sure to take a look at the full space products portfolio.