The oscilloscope screenshot below shows an address followed by 22 contiguous DIM signals sent by a CM11A in the upper trace and the data output from a TW523 in the lower trace. The TW523 does not output in real-time. It delays its output for a minimum of 22 half-cycles and does not report every code on the powerline. The output code is what was on the powerline 22 half-cycles earlier.

The next screenshot shows a similar sequence but at slightly higher horizontal resolution. It can be seen that once the TW523 outputs a code, it will not output another for 22 half-cycles (or longer if the first 1110 comes after more than 22 half-cycles). What this means is that the TW523 only reports every third code for contiguous DIM or BRIGHT codes and cannot accurately report extended codes.

The next screenshot shows what happens when the same CM11A signal collides with a signal from another transmitter. The TW523 does not report codes if the bits which follow the startcode (1110) are not valid Manchester codes. In most cases, collisions will result in no valid code but there are cases where the collision does output a valid code that differs from both colliding codes. I suspect this most often happens when the address/function bit (the last two half-cycles) gets changed from address to function.

_{Copyright: No part may be reproduced except as authorized by written permission. This restriction extends to reproduction in all media.}