Total square sum with proportional error model

Dear Sir or Madam,

I am doning nonlinear regression with "fitnlm". The result is a "1x1 NonLinearModel" which contains a lot of information. I am interested in the determination of the value "SST" (totla sum of squares) in case of "'ErrorModel','proportional'". My question is how "SST" is calculated, what is the formula for "'ErrorModel','proportional'"? Thank you very much for you support!

Yours Sincerely,

Josef Denk