In my last two blogs in April, Single Event Effects (SEEs) with High Speed ADCs: Single Event Transient (SET), and in May, Single Event Effects (SEEs) with High Speed ADCs: Single Event Transient (SET), Part 2, I discussed single event transients with high speed ADCs. In this installment we will talk about single event upset (SEU). Just as in the case of an SET the device does not require a reset to return to normal operation after a single event upset. The device is also fully functional after the event since the SEU is a non-destructive event. Technically an SET could also be considered as an SEU for a high speed ADC since it is likely that an ion strike could cause an SEU that results in a transient event that we would call an SET. For the purposes of this blog we will refer to these events separately to provide some differentiation in the types of single event effects that can occur with a high-speed ADC.
Once again, we will look at the case of the AD9246S to learn more about SEUs. The test setup for measuring the SEU performance of the AD9246S is virtually the same as the setup used when testing for SETs. Recall that the supply voltages are nominal for SET/SEU testing and are only elevated for the SEL testing. I’ve included the setup diagram here once again for your reference. In the figure the test conditions called out for SET also apply to SEU.
AD9246S Single Event Effects (SEE) Test Setup
Recall from my previous blogs in April and May that the SET test is generally performed at ambient temperature conditions and that the temperature is monitored and recorded. The SEU test is performed in the same manner. Just like the SEL and SET testing, the SEU evaluation for ADI products is typically performed up to an energy level of 80 MeV-cm2 /mg and out to a fluence of 107 ions/cm2 . In some cases, the energy level can be ≥ 120 MeV-cm2 /mg depending upon the customer and application requirements. The SEU evaluation would be performed out to a fluence of 106 ions/cm2 or up to 100 events, whichever comes first.
Once again, we are looking for the energy level where SEUs onset with the device and increase the energy level until the saturation point for SEU events is found. The ADC is exposed to increasing levels of radiation while the number of SEUs is recorded.
If you recall, in part one of the SET blog back in April we discussed that there were four different ions used when testing the AD9246S which allowed us to have five different energy levels thanks to our mathematical friend cosine. As a reminder the different energy levels are shown here again in Table 1.
Ions and LETs Used for AD9246S SET Testing
The same test procedure used during the SET evaluation is also applied for the SEU evaluation. Once again, I’ve included the test procedure for SEU (and SET) evaluation of the AD9246S just below:
- Power up the AD9246S.
- Select the desired ion and desired angle of incidence.
- Turn on the ion beam while observing, monitoring, and logging the power supply currents and recording any upset events.
- Turn off the beam when either the specified number of upsets is recorded or the fluence reaches 106 ions/cm2 .
- Repeat the procedure beginning at step 2 until the AD9246S has been irradiated across the desired range of LETs (∼2 MeV-cm2 /mg to 80 MeV-cm2 /mg).
- Test the remaining AD9246S devices using steps 1 through 5 until the desired number of units have been tested.
For the purposes of this blog an ADC a device configuration register upset is treated as an SEU. A device configuration register bit could upset when struck with an ion and disrupt the device performance in some manner. For the case of the AD9246S the device configuration registers are each 8 bits. It is possible that a single bit could upset as below:
SEU: Single Bit Upset in a Configuration Register
It is also possible that an ion or ions could strike the device and upset multiple bits simultaneously such as is shown below. In this case two bits are upset, but it is possible that more than two bits could upset simultaneously depending on the number and location of ion strikes.
SEU: Multiple Bits Upset in a Configuration Register
Either event, whether it be a single bit upset or multiple bit upsets, would be considered an SEU. In both cases the device functionality could be interrupted in some manner. Since the bit(s) could return to their expected state or could be reset to their expected state via a device reset or a write to the register the upset is non-permanent and is considered a single event upset. It is possible that an upset bit(s) could produce an SET in the device depending on what function the upset register bit(s) might control. Again, for the purposes of this blog we will consider the register bit(s) upset(s) to be SEUs.
The information, collected and calculated for the Weibull curve, can be used to project the SEU performance of the device in various orbits around Earth. Recall that this Weibull curve can be used to calculate a probability of an upset event, in this case an SEU. This is helpful information as it can be used to predict the probability of an SEU in a particular orbit. The onset threshold and saturated cross-section are derived to understand the LET at which SEUs begin and the LET at which the number of SEUs saturates. In the case of the AD9246S there were no configuration register upsets observed as reported on page 17 of the AD9246S Single Event Effects Test Report. Since there were no configuration register SEUs there is no need to plot a Weibull fit curve. As an example, the Weibull fit curve for the AD9246S SET performance is given below. If the AD9246S were to have had SEUs the Weibull fit curve would look similar to the plot below.
AD9246S SET Cross-Section and Weibull Curve
The goal with testing for these different types of SEEs is to predict the device behavior when placed into an application in space. SEE testing exposes the device to many more ions than the device would typically see in its service lifetime in space. Observing these SEEs while exposing the device to these numerous ions is in a way an accelerated life test. In a short amount of time a large amount of data on the SEE performance can be gathered and then subsequently used to predict how a device will respond in an actual application in space. It is imperative to understand as much as possible about the performance of the device before placing a device into service in space. Obviously, the opportunity to repair or replace device or system in space is not very feasible. It is perhaps possible, albeit costly, to replace something that is in orbit. However, think about Juno that was sent to collect data on the planet Jupiter. Juno must operate and withstand multiple ion strikes in its environment without failure. It is virtually impossible and completely impractical to try to repair the probe once in operation. This is one of the reasons why it is imperative to adequately test devices here on Earth to ensure they will operate for the desired mission life in space.