Surety is performing system maintenance this weekend. Electronic date stamps on new Prior Art Database disclosures may be delayed.
Browse Prior Art Database

Estimating Circuit Age with NBTI Sensor

IP.com Disclosure Number: IPCOM000234151D
Publication Date: 2014-Jan-14
Document File: 3 page(s) / 116K

Publishing Venue

The IP.com Prior Art Database


A method for indicating an age metric of a circuit is disclosed. The method relies on a physical sensor co-located with the circuit. The sensor data is translated into an age metric that is simple and understandable to the computer system user or operator.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 40% of the total text.

Page 01 of 3

Estimating Circuit Age with NBTI Sensor


It is well known that CMOS transistors suffer degradation over their lifetime which eventually leads to slower switching speeds. One such form of degradation is known as Bias Temperature Instability (BTI). The effect of BTI is that the threshold voltage required to switch a CMOS transistor on rises as the transistor ages. This increase in threshold voltage slows the time it takes for the transistor to turn on and thus reduces the maximum switching rate. This reduction in switching rate may eventually cause the circuit to fail once it reaches a critical point, since microprocessors typically operate within a clocking range that is fixed at design time.

Problem Statement

There are many ways of detecting or estimating CMOS circuit degradation in the literature. However, the indicators for degradation are typically measurements of threshold voltage (in millivolts) or slowdown of ring oscillators (in megahertz). Such raw measurement values are only meaningful in the particular context of an individual microprocessor, its operating environment, and its designed lifetime goal. These indicators have little meaning for a typical user of a computer system who wants to know how close a microprocessor is to the end of its functional life. The invention provides a human-usable aging metric, called effective age , for a microprocessor that is relative to its designed lifetime objective.


The invention employs a physical degradation sensor which is capable of providing a relative degradation compared to the initial circuit operation at the beginning of its lifetime. An example of such a sensor, is a dual-ring oscillator apparatus embedded in a microprocessor [1]. This sensor measures degradation by comparing the frequency of a ring-oscillator which is always under a stress load to the frequency of a reference ring-oscillator. The reference ring oscillator only experiences stress load when it is turned on and measured periodically. Since it is not subject to a load most of the time, it has almost no degradation from its manufacture time. The difference between the ring oscillator rates is measured and called "delta f". The effective degradation is the ratio between "delta f" and the reference oscillator rate. Thus the equation for effective degradation is

         effective_degradation = (reference_oscillator_rate - stress_oscillator_rate) / reference_oscillator_rate = "delta f" / reference_oscillator_rate (Eqn. 1)

where delta_f is the measured difference between the reference and stress ring oscillators and reference_oscillator_rate is the reference ring oscillator measurement.

Figure 1 shows an example of measuring effective degradation for 5 years of a microprocessor's use (dotted line). In this example, the microprocessor was designed to exhibit no more than 4% effective degradation in 10 years. The model (solid line) in the figure is an example mathematical model of the expected worst-case...