Assigning weight and bias values for a Network

Hello all.. My network has 3 inputs, 5 neurons in a single hidden layer and 1 output. I observed that ANN automatically assigns weight and bias values for my network and thus training and simulation of the network 5 times will always generate 5 different output values. So after training/simulation I extracted the weight and bias values of the network using the code below

    InputWeight = myFunc.IW{1};
    LayerWeight = myFunc.LW{2};
    b1 = myFunc.b{1,1};
    b2 = myFunc.b{2,1};

I assigned these values to my network's IW, LW, b1 and b2 using the code below... (InputData is an (3xN) matrix and OutputData a (1xN) matrix)

            NormInp = minmax(InputData);
            myFunc = newff(NormInp, [5,1],{'tansig', 'purelin'});
            myFunc = configure(myFunc,InputData, OutputData);
            myFunc.IW{1,1} = IW;
            myFunc.b{1,1} = b1;
            myFunc.LW{2,1} = LW;
            myFunc.b{2,1} = b2;
            myFunc = train(myFunc, InputData, OutputData);
            myOut = sim(myFunc,InputData);

Running this code about 5 times, generated exactly the same output. But when I change, the activation functions to;

    myFunc = newff(NormInp, [5,1],{'purelin','tansig'});

normalize input/outputdata to [-1 to +1] and de-normalize the output, each element of the output matrix turns out to be the same i.e having a (1xN) matrix in which each of the elements is equal to 77. Please does anyone know what is responsible for the wrong output??