This isn't a creation, but an idea. It doesn't have to be done, though.
So how the step by step adding machine works is to add large numbers with only a few or couple, if not, even a single full adder. Of course, you need to input something. It can be anyway you want it. Although it might be complex. The adding machine is limited to the amount of outputs it has.
Here are the steps:
1. Adding
The machine takes an input and sends it to the adder(s). If there are multiple adders, the inputs will be taken faster. The Cin will also be taken, synchronized with the next input taking of addition.
2. Sum Writing
The sum that comes from the machines will be written to the output. If there are multiple adders, sums will be written faster.
3. Write Shifting
The output piece that gets written (if it doesn't) will be shifted to the next output piece. Example: If a sum is written into Piece 2, after the synchronization clock sends a pulse for another calculation, the writable output is shifted to Piece 3.
The process repeats until all digits are calculated, or if memory runs out.
The sums can then be stored as temporary memory, which can be read until the next operation, where it clears.
If it doesn't make sense to you, I'll (try to) add more information about what confuses you.
I have something like this on FP right now. I was looking for a smallish adder a while ago for a bomb tester and someone said to check out serial adder. So, I did, then I made one, Then I upgraded it to do subtraction as well as addition. I've seen a few other users make them too.