Speaker
Description
After nearly a decade of prototyping and designing on XH180 at TESAT, our first design RUCA has finally reached FM-Level and is ready for launch. The RUCA is a mixed signal ASIC, build to bias external RF-components. For this purpose, DACs, ADCs, on-chip power supply and biasing, regulation loops and a huge digital part have been implemented. All these functionalities showed sufficient performance in our lab experiments. Before qualification could start however, the last step was to eliminate issues, which had been discovered during radiation tests of the last prototype device. These effects included inhibited operation due to drifts during total dose tests, single even transients on analog outputs and device destruction under high energy Xenon irradiation with 62.5 MeV/(mg/cm²). All destructive effects have been successfully eliminated and transients have been mitigated down to an acceptable level.
Here, we will describe the journey from the prototype to the FM considering radiation. This includes radiation test setups, tests carried out and their outcomes, methods of analysis and finally the solutions to eliminate destructive effects and to mitigate nondestructive effects.
For TID-Testing, we have developed a very flexible setup consisting of several small boards, which can be arranged in equivalent distances to the radiation source and a test system for screening the devices. Screening is done automatically with minimum user intervention and can be done by none expert users. The same setup can be used for burn in and life tests. Data of each run has been evaluated directly and drifts have been calculated. As for all our tests, a python test framework is used. During prototype testing, a single circuit failed earlier than expected, so the complete device could not stand high doses. Further irradiation of the prototypes was possible, however.
After the tests, we developed an analysis method using a circuit simulator. We added voltage sources in series with the gates of each single transistor and looked for the devices, which failed first or had most influence on the output. This way, the complete circuit failure could be traced to a single MOS device, whose channel did not close anymore. The easiest and working solution was to use the affected high voltage device for protection only and to work with a low voltage device for functionality.
For single event testing, a completely different approach is necessary, as all data has to be collected very fast during a very short period of irradiation. For this purpose, we developed an FPGA-based measurement platform. The setup can on the one hand observe several observables in parallel and count the crossing of adjustable thresholds. On the other hand, some observables can be switched onto coaxial outputs for measurements outside the chamber with the oscilloscope. This way single event rates can be measured while the exact figures of transients can be observed, too. A very fast latch up protection compares the current flowing through a shunt resistor with an adjustable threshold and cuts off the device it is crossed.
We had three major outcomes of the first test: One circuit part showed a single event functional interrupt requiring a power cycle of the device. As this circuit was not necessary for the FM functionality, it has been completely removed. Moreover, the single event transient behavior of some analog outputs has been worse than expected. It has been possible to improve those outputs. The worst outcome however has been a destructive behavior at Xenon (62.5 {MeV/(mg/cm²)}). A massive raise in current has been observed triggering our latch up protection. This current remained static even after a power cycle and seemed to be additive. There could be several of those events.
Several test runs and analysis methods could track the failure to a single circuit in the device. Interestingly it is most probably no latch up, but a gate or diffusion rapture. However, the high current of a latch up might have caused a destruction leading to a high current, so it cannot be proven if it was a latch up, or a direct destruction. The finally easy and secure solution was to remove the bad circuit from the device, as it has not been necessary for operation.
After solving all issues of the RUCA prototypes, it has been possible to finally produce radiation hard RUCA FM-devices.
