generation of gDNA Std curves in qPCR - ABI 7300 qPCR (Mar/14/2006 )

I'm trying to make a std curve from 20K to 1.25K copies. ABI has a calibration plate using RNAse P doing the same thing. with the calibration plate the equation to the curve comes out like
(y = -1.416Ln(x) + 38.889, with an R2 = 0.9949).

i am using Genomic DNA purchased from ABI with a final Conc of 0.1 ug/ul. using their literature in designing std curves i determine the mass of gDNA needed per haploid genome to be 3.3 pg. i next determine the mass of gDNA needed for a 1:1 serial dilution starting from 20K copies to 1.25K. i divided the mass needed by the amount of volume i want to add to the reaction mixture (5ul) to get the final conc of gDNA.

to make 40 ul of 20K copies gDNA -
stock [0.1 ug/ul gDNA]
20000 copies x 3.3 pg/1 copy = 66000 pg
66000pg / 5 ul = 0.0132 ug/ul
WS of 20K copies = 5.28 ul stock + 34.72 ul adH2O.

i run the dilution in triplicate and i get a std curve of
(y = -2.4771Ln(x) + 56.138, R2 = 0.982) which is quite a bit flatter than the calibration plate.

anyone know what i'm doing wrong or how i can make the curve better?

please and thank you.

i've increased the amount of stock gDNA, thus decreasing the error. my pipettes are due for service in a couple of months but i think i'll call them early.

my new equation is - y = -1.3566Ln(x) + 45.613
R2 = 0.9964

not bad