Dispensers failing to activate when triggered via redstone. Mostly consistent and repeatable, but conditions are unkown.

I have a redstone construct that uses stacked arrays of Dispensers in two distinct patterns; First is a vertical stack of four facing sideways with an adjacent stack of blocks that can be activated by redstone Each of these four blocks are activated bottom to top by a repeater for the first and third and a single unconnected patch of redstone dust for the second and fourth. The single patches are activated by an adjacent repeater to the direct repeater>block. Basically a 2x2x3 modular vertical activation tower. Works for Lamps and TNT, so it should work for Dispensers. The bottom Dispenser in this setup - which should have identical input to the 3rd going up - fails to activate most attempts. Only about 1 in 15 attempts are successful. The second pattern was at first a horizontal row of 5 Dispensers aimed down with a row of blocks on top of them and a line of redstone on top of those blocks (because while you can place redstone on a dispenser it breaks when the dispenser is triggered - another bug/issue to be sure) and the inbound activation coming in over block #3 in the middle. This block #3 would trigger without fail, but the other 4 would only activate maybe once in about 15 attempts. I read a "solved" issue here on the topic suggesting to use side-activation in which the thread was not actually solved, but took to testing it anyway. I created a subset row of blocks beside the row of Dispensers and a row of redstone on top of them while leaving the original blocks and redstone in place. This did little more than make Dispensers #1 and #5 (the ends of the row) slightly more reliable. I then removed the blocks and redstone on top of blocks 1, 2, 4, and 5 while leaving the input where it was as that was not an issue. This had the effect of making Dispensers #1 and #5 100% reliable, but #2 and #4 remain at best 1 in 15 attempts. Again this construct would work for activating Lamps and TNT from all sources. There are no viable sources of residual "active" states and these dispensers should be inactive when their adjacent blocks are.

Side note: When first using the construct there was noticeable block update latency, which I promptly corrected by reducing view distance. Another bug/issue involving the single-player internal server design. Optifine can't fix the internal server unfortunately, but that shouldn't be an issue with a PC like mine (8-core CPU).