The stability of radiation emulator on cloud microphysics changes is essential for utilization in operational weather-forecasting models with frequent updates. This study examined the effects of 15 microphysics schemes on a radiation emulator for real and ideal cases. In the real case, although the forecast errors (compared to a control run) were higher with different microphysics schemes compared to those with the trained scheme, the forecast error for the 2-m temperature rather improved by 0.9-5.4% compared to observations. The radiation emulator for the real case was applied to a two-dimensional ideal simulation to test the universal applicability of the emulator; the resulting forecast errors in heating rates and fluxes for 14 microphysics schemes increased by 8.6-41.3% compared to the trained scheme. The errors were reduced by 26.5-50.4% by utilizing compound parameterization. Therefore, the stability and accuracy of the radiation emulator were confirmed for various microphysics schemes.