Dear Morteza,
No, if correct your leadfield matrix elements should have a magnitude of more or less this. Don't scale your leadfield matrix.
I can see the communications background, in which you would convolve using an impulse response. In this case however you can see it as a static gain. The lead field matrix does not contain time.
The leadfield matrix is in the order of #electrodes*(3*#sourcepoints), in your case 64*(3 * 15670), if it is unconstrained. (In constrained sources the 3 becomes a 1 since the dipole orientation is fixed then) Your source matrix thus should be (3 * #sourcepoints) * time if you want to end up with a matrix that is #electrodes*time
Which probably means a matrix full of zeros and at the rows corresponding to the vertices you put in the time series of your source.
Hope that answers your questions 
Kind regards,
Steven