Three inputs and intermingling pipelines made it hard to get the periods of each pipeline in sync so that the machine wouldn't crash.

Two things that might be interesting about this solve:
1. Disposal.  Exactly one quicksilver atom is disposed.  This allows the first reactive water to be delivered earlier to the multibonder.  Before adding the disposal, I needed an arm + track to pull one partially-built product out of the stream due to timing constraints.  Disposal is free, and I happened to already have a perfect spot for it!

2. The biarm in the middle that delivers reactive water to the multibonder.  My original build order required performing this 2 cycles apart, which would have required much more than a biarm!  It was a lot of work, but I was able to make it so that arms 14 and 15 are offset time-wise from each other, so that the pulls were 6 cycles apart instead.  You'll notice over the course of the loop that the delivery of iron and quicksilver to projection varies in timing.