From 78599de6a83317c86b29b5c8364d4a211a8ff01f Mon Sep 17 00:00:00 2001
From: Sam Calisch <sam.calisch@cba.mit.edu>
Date: Mon, 28 Aug 2017 14:56:33 -0400
Subject: [PATCH] Update README.md

---
 README.md | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/README.md b/README.md
index e0288e2..1d49e47 100644
--- a/README.md
+++ b/README.md
@@ -7,6 +7,8 @@ This is documentation for an energy monitoring device built around the nRF52832
 
 This device uses the <a href='http://www.ti.com/lit/ds/symlink/drv5053.pdf'>DRV5053VA</a> hall effect sensor to produce around 0.1 volts/amp out of the sensor (or ~1 mV per Watt through the plug).  The nrf adc has a +-0.6V reference, 4x gain, and 12-bit conversion, which might resolve 100mW (maybe finer with oversampling).  Certainly not precision measurement, but good enough for monitoring appliances.
 
+Actually, in this first design, I forgot that the hall effect sensor has a 1V output baseline (when field is zero).  This means in this design can't use the ADC gain in the BC832.  To fix this, I should use a sensors near each conductor and measure differentially between them.  Not only will this eliminate the 1V measurement offset, but it should double the signal-to-noise ratio.
+
  The nRF52 radio+microcontroller draws 5mA in transmit and receive, so it could operate well within the specs of a coin cell, or harvest from the power line (say, using http://www.linear.com/product/LTC3588-1) to charge a capacitor for power.  
  
 Below, we see the flux guides registered into pockets on the pcb. 
-- 
GitLab