Ok. I fixed it. I simply removed the first three lines of code from the bin_to_bcd routine where r2, r3, and r4 were (needlessly) initialized to 0.

However, I still don't understand why this fixed it. Even though there is no need to initialize r2, r3, r4 to 0, I don't see why it would prevent 1's in the LSB once those registers are re-assigned values.

Also, what I said about the previous code working in my counter circuit still stands, which confuses me even more.

Thanks.