Reconciling “8 Years” With “20-40 mAh/day”

Ada re-derives this chapter’s own numbers step by step, at full precision

foundations
math-foundations
calculation-audit
http-web
Ada ADA · CALCULATION AUDIT

Reconciling "8 Years" With "20-40 mAh/day"

The chapter carries two polling-energy figures that look contradictory: the headline says HTTP polling costs 20 to 40 mAh/day, while a connection-overhead panel derives about 8 years of battery — barely 1 mAh/day. Both are the chapter’s own numbers for 144 connections a day. This audit reconciles “8 years” with “20-40 mAh/day” by finding the one hidden variable: active radio time per poll.

Companion to the chapter HTTP Connection Pitfalls — every number here comes from that chapter.

Ada: This chapter carries two polling-energy numbers that look contradictory — the headline says HTTP polling costs 20 to 40 mAh/day, while the connection-overhead panel derives a battery life of about 8 years, which is barely 1 mAh/day. Both are the chapter’s own figures, so let me show they describe the same device at two levels of fidelity.

  • Polling every 10 minutes gives 86400 / 600 = 144 connections/day — the shared anchor for both numbers.
  • The idealized handshake model uses E = 80 mA x 270 ms + 20 mA x 180 ms = 21.6 + 3.6 = 25.2 mA·s per connection (its stated ~450 ms of active radio), i.e. 25.2 / 3600 = 0.007 mAh. Daily: 0.007 x 144 = 1.008 mAh/day, so 3000 / 1.008 = 2976 days ≈ 8.15 years. That reproduces the panel exactly.
  • But the pitfall text itself states each poll holds the radio awake 3 to 5 s at 50 to 100 mA. Taking 5 s x 100 mA = 500 mA·s = 0.1389 mAh per connection: 0.1389 x 144 = 20 mAh/day — exactly the headline’s lower bound.
  • The two models differ by 500 / 25.2 = 19.8x, which is precisely why 8.15 years collapses to 3000 / 20 = 150 days once real radio wake time is counted.

The audit conclusion is that neither number is wrong; the one hidden variable is active radio time per poll, and it swings the battery estimate nearly twenty-fold — so the figure you must actually measure on the device is seconds-awake-per-wake, not handshake byte count.

Every number above is taken from the chapter’s own material and re-derived step by step.