Question 9.7: Latch-up in CMOS Use the circuit in Figure 9.35 to calculat...

Latch-up in CMOS

Use the circuit in Figure 9.35 to calculate the power-supply Current I_{DD} as a function of the current I_W in the well. the current I_X in the substrate, and the well-substrate current source I_0 at the well junction. Assume that both transistors are active, and find conditions on the transistor common base gains α that cause I_{DD} to become unbounded. Assume that any voltage changes occur slowly.

9.35
The blue check mark means that this solution has been answered and checked by an expert. This guarantees that the final answer is accurate.
Learn more on how we answer questions.

Because the voltage changes occur slowly, the capacitor C_{PS} need not be considered.
By applying Kirchhoff’s Current Law to the circuit of Figure 9.35, we have

I_{DD}=I_X+I_{PE} \\ I_{DD}=I_W+I_{NE} \\ I_W=\alpha _PI_{PE}-(1-\alpha _n)I_{NE}+I_0 \\ I_X=\alpha _nI_{NE}-(1-\alpha _P)I_{PE}+I_0

Eliminating I_{PE} and I_{NE} we can write

I_{DD}=I_X+\frac{(I_W-I_0)}{\alpha _p} +\frac{(1-\alpha _n)}{\alpha _p} \times\left[\frac{(I_X-I_0)}{\alpha _n}+\frac{(1-\alpha _P)}{\alpha _n}\times(I_{DD}-I_X) \right]

Solving this expression for I_{DD} we have

I_{DD}=\frac{I_0-\alpha _PI_X-\alpha _nI_W}{1-(\alpha _n+\alpha _P)}

From this expression we see that I_{DD} tends toward infinity (and the circuit becomes latched) when the sum ( α_n + α_p ) approaches unity. This condition on the transistor αs can be compared to the latch-up constraint expressed through Equation 9.3.4. Equation 9.3.4 was obtained by applying the small-signal equivalent circuit for bipolar transistors, and thus represents a condition on the circuit gain during the transient build up to the latched condition. In this example, steady state is considered, and the r_π resistor-divider terms are not relevant. If these terms are not considered, Equation 9.3.4 simplifies to the condition β_n  × β_p = 1 , or

\left(\frac{\alpha _n}{1-\alpha _n} \right) \left(\frac{\alpha _p}{1-\alpha _p} \right) =1

which reduces to ( α_n + α_p = 1 ) as concluded above.

G_L=\beta _n\times\frac{R_X}{r_{\pi pnp}+R_X} \times\beta _p\times\frac{R_W}{r_{\pi npn}+R_W}                                                      (9.3.4)

Related Answered Questions