When given multiple 1-redstone-tick pulses, separated by 1 redstone tick, from certain sources, observers will trigger late on the second pulse, causing a corruption of the intended signal. This only occurs in certain orientations from certain sources. This will in some cases result in the loss of entire pulses (for example, if 1 is on and 0 is off, a signal sent as 1010101 would result in a signal of 1001001) (13.png)

The delay might come from the observer being on cooldown (unable to receive input) for one tick too long after pulsing the first time, and detecting the redstone/rails/etc turning off rather than detecting them turning on the prior tick as would be expected.

I use a resource pack in the pictures to show on all sides when an observer is powered.

• Observer chains facing north/south trigger as expected (1.png)
• Observer chains facing east/west don't trigger the expected result (even though they power the repeaters fine) (2.png)
• Repeater chains in either orientation trigger incorrectly (3.png and 4.png)
• The bug seems to be based on the orientation of the chains themselves, not the actual observers that power the redstone (5.png and 6.png)
• Observer-chains that output vertically trigger incorrectly no matter the orientation of travel (7.png and 8.png)
• Observer clocks seem to work fine in either orientation (9.png and 10.png)
• Observer chains traveling vertically work as expected (11.png and 12.png)