diff --git a/implementation/spi-uwb.tex b/implementation/spi-uwb.tex
index cdda2bd3e497a2fedbd64566d535312fea486b51..a15c13efb10e2802ae3925c0229a005af41c76c0 100644
--- a/implementation/spi-uwb.tex
+++ b/implementation/spi-uwb.tex
@@ -74,11 +74,10 @@
             \item \makebox[2cm][l]{SPICLK:} \makebox[2cm][l]{PB10}
         \end{itemize}
 
-        Handling NSS (Slave select) in hardware would be faster and easier.
-        An undocumented hardware bug in the STMF7-series processor causes the hardware-managed NSS to malfunction if NSS Pulse was disabled. \cite{web:stm_nss_problem}
+        The NSS pin is handled in software like a standard digital GPIO pin.
+        Handling NSS (Slave select) in hardware would be faster and easier, however an undocumented hardware bug in the STMF7-series processor causes the hardware-managed NSS to malfunction when NSS Pulse is disabled \cite{web:stm_nss_problem}.
         An example of NSS Pulse mode can be seen in Figure \ref{pic:NSSP}.
         The DW1000 does not support NSS Pulse communication, so software NSS was required.
-        The NSS pin is instead handled like a standard digital GPIO pin.
         All other SPI is implemented using the HAL\footnote{Hardware Abstraction Layer} libraries supplied by STMicroelectronics for this processor.
 
         \begin{figure}[H]
@@ -212,7 +211,7 @@
         At maximum precision (\SI{0.25}{\degree} resolution and 3 byte per point) a total of $\frac{\SI{270}{\degree}}{0.25} = 1080$ data points, resulting in 3240 bytes of data.
         This gives 
         \begin{align*}
-            \frac{1080}{90 / 3} = 36\text{ full packages.}
+            \frac{1080}{90 / 3} = 36\text{ packets.}
         \end{align*}