← Back to the chapter · Previous detail

Technical detail · Low-Power Design: Sleep-State Energy Accounting

Derive the CC2650 wake and retention costs

At 48 MHz, the active-core model gives

Iactive=1.45 mA+(31 μA/MHz)(48 MHz)=2.938 mA.I_{active}=1.45\ \text{mA}+(31\ \mu\text{A/MHz})(48\ \text{MHz}) =2.938\ \text{mA}.

A 151 μ\mus standby-to-active transition spans

48 MHz×151 μs=7,24848\ \text{MHz}\times151\ \mu\text{s}=7{,}248

clock periods. Approximating the transition at the active current gives 0.444 μ\muC, or about 1.33 μ\muJ at 3 V, before useful instructions. That fixed wake term is why batching several tasks behind one wake can save more energy than shaving a few cycles from each task.

Retention uses more sleep current but may avoid cache refill or reinitialization. Compare two states over sleep duration TT:

QA=Isleep,AT+Qwake,A+Qwork,A,Q_A=I_{sleep,A}T+Q_{wake,A}+Q_{work,A}, QB=Isleep,BT+Qwake,B+Qwork,B.Q_B=I_{sleep,B}T+Q_{wake,B}+Q_{work,B}.

If cache retention costs an extra 1.7 μ\muA but avoids 1,500 active cycles at 48 MHz, the avoided active charge in this simplified model is

ΔQwork=2.938 mA150048 MHz=91.8 nC.\Delta Q_{work}=2.938\ \text{mA}\frac{1500}{48\ \text{MHz}}=91.8\ \text{nC}.

The break-even sleep interval is 91.891.8 nC/1.7 μ/1.7\ \muA\approx54$ ms. With shorter sleeps, retention wins in this hypothetical workload; with longer sleeps, the lower-current state wins. Replace 1,500 cycles with a trace from the actual firmware, include oscillator and regulator transitions, and repeat at the required temperature.

← Back to the chapter · Previous detail