The radio light curves of SN 1993J are discussed .
We find that a fit to the individual spectra by a synchrotron spectrum , suppressed by external free-free absorption and synchrotron self-absorption gives a superior fit to models based on pure free-free absorption .
A standard r ^ { -2 } circumstellar medium is assumed , and found to be adequate .
From the flux and cut-off wavelength , the magnetic field in the synchrotron emitting region behind the shock is determined to B \approx 64 ~ { } ( R _ { s } / 10 ^ { 15 } { ~ { } cm } ) ^ { -1 } G. The strength of the field argues strongly for turbulent amplification behind the shock .
The ratio of the magnetic and thermal energy density behind the shock is \sim 0.14 .
Synchrotron losses dominate the cooling of the electrons , while inverse Compton losses due to photospheric photons are less important .
For most of the time also Coulomb cooling affects the spectrum .
A model , where a constant fraction of the shocked , thermal electrons are injected and accelerated , and subsequently lose their energy due to synchrotron losses , reproduces the observed evolution of the flux and number of relativistic electrons well .
The injected electron spectrum has dn / d \gamma \propto \gamma ^ { -2.1 } , consistent with diffusive shock acceleration .
The injected number density of relativistic electrons scales with the thermal electron energy density , \propto \rho V ^ { 2 } , rather than the density , \rho .
The evolution of the flux is strongly connected to the deceleration of the shock wave .
The total energy density of the relativistic electrons , if extrapolated to \gamma \sim 1 , is \sim 5 \times 10 ^ { -4 } of the thermal energy density .
The free-free absorption required is consistent with previous calculations of the circumstellar temperature of SN 1993J , T _ { e } ~ { } \sim ( 2 - 10 ) \times 10 ^ { 5 } K , which failed in explaining the radio light curves by pure free-free absorption .
Implications for the injection of the relativistic electrons , and the relative importance of free-free absorption , Razin suppression , and the synchrotron self-absorption effect for other supernovae are also briefly discussed .
It is argued that especially the expansion velocity , both directly and through the temperature , is important for determining the relative importance of the free-free and synchrotron self-absorption .
Some guidelines for the modeling and interpretation of VLBI observations are also given .