Mixed Signal BIST: Fact or Fiction

Depending on your expectations or claims, mixed-signal BIST could be either fact or fiction. Recognizing facts from fiction is key for the industry to successfully deploy mixed-signal BIST potentials. Is Mixed-Signal-BIST Feasible? Despite the general consensus that considers mixed-signal BIST as a relatively new technology, it has been used as part of high-performance analog design process for several years. As shown in Karim triangle in Figure 1, mixedsignal BIST is used as an essential part of most autocalibration techniques and feedback loops. Feedback loops and auto-calibration techniques are used widely to enhance the resolution and stability of analog circuits. Most commercial PLLs, ADCs, DACs and filters have some kind of embedded measurement mechanism (i.e. BIST) based on which the circuit performance is adjusted either at the power-up or continuously during the operation of the device. In most cases, this capability doubles or triples the size of the original circuit. The area overhead is accommodated as it adds a real value to the device. It simply helps the device to meet specifications. Is Mixed-Signal-BIST a Complete Solution? Most of today’s available mixed-signal BIST solutions are not complete and do not solve the whole test problem. They could be extended to cover most of required tests, but the resulting area overhead and test time will definitively be way beyond reasonable and tolerable limits. Measuring circuit specifications inside the chip rather than outside the chip is another inherent disadvantage of mixedsignal BIST. What we really sell to customers and try to guarantee by test is the specifications that are perceived outside the chip. There could be a noticeable performance disparity between a signal inside the chip and outside the chip due to the signal degradation in the pad circuitry.