MaudRacapé

Hello everybody,
I registered to bioforum because we are really struggling with a human IgG elisa that was perfectly working a few weeks ago.
So we coat our own engineered peptide in CCB O/N at 10µg/ml, block with 1X PBS 0.1% BSA for 30min, add our diluted human sera for 30 min (usual dilution is 1/100 in wash buffer 0.1% BSA, usually works). During this step we incubate the biotinylated goat anti human IgG (1/10 000) with the streptavidin-HRP (1/4000) in 1ml diluent. Then we add the 1ml to the desired volume of diluent, incubate the mix on the plate for 30 min, add TMB for 10 min and stop the reaction with 2M H2SO4.
As I was saying this protocol was perfectly fine a few weeks ago but now OD values for the same control samples (we have tested several of our positive controls) are extremely low and not reproducible from one experiment to another and from one person to another.
We have changed everything: making fresh CCB, fresh coating peptide, changing secondary antibody, HRP antibody, TMB, water to prepare the buffers, we even have checked the plate reader. We have also checked other dilutions of secondary and HRP, diluting less the secondary and more the HRP gave higher OD values but again we were not able to reproduce it. Trying to decrease or increase the concentration of the coated peptide also did not change anything. The HRP/TMB seems to work because adding TMB to 1µL HRP in diluent gives a strong blue color. Nothing that we tried works.
Two weeks ago, the same type of elisa with almost the same type of coating peptide also started to go off, and now it is not working any more. On another hand, the same type of elisa with the same peptide on mouse serum seems to work as usual (although I should repeat to be sure). The only difference between the human and the mouse elisa is the serum and the secondary antibody. Given that the secondary antibody has been changed it could come from the serum, but even if we had lost the reactivity of our sera (which would be surprising on all of them), why can't we go back to stable values even if they are low?
I hope somebody has an idea of what I could try now, thank you in advance!
Maud