formula in calculating attenuation in a series configuration?

Well I was using Sheldon's calculator ( It gives you attenuation for a particular step.

I gave the calculator the ff. values: 24 steps (I think he means positions here), 100k total resistance and total db of -64.

It thens gives me several calculated resistance/attenuation for each step. Since the calculated resistor values is different from what's available, the calculated attenuation also changed.

I found this site ( the gives a formula for calculating attenuation.

Here's the formula

Well, I calculated the db for step 23 which has 48.05 as the calculated resistance.

here's the calculation:

db=20*log(48.05/100000)= -66.xxdb. Sheldon's site gives be 64.xxdb.

When I calculated the db for step 22 with 87.94ohm as calculated resistance.

here's the calculation:

db=20*log(87.94+48.05/100000)= -57.xxdb. Sheldon's site gives be 59.xxdb.

I might be using an incorrect formula. So please guide me to the correct one. Thanks :)

Can you have forgotten that

log 20 scale always compare value
to zero attenuation?
While a "normal" scale have zero
at nothing,
the log20 have its zero at maximum output. zero attenuation.
A high value is set to zero.
Values below are-xxdB
values above are+xxdB

A normal scale have no negatives,
as there can not be anything that doesn't exist.

Also nothing can be added to what exist.
The bible says you can not take way
or add something.