It depends on the unit of your time. Often we think of it as an “amplitude”, but if your time series is a series of voltage amplitudes versus time, then your PSD estimate will be Volts^2/Hz . This is because the PSD is the Fourier transform of the autocorrelation of the original signal: autocorrelation has units of Volts^2 , and its launch through the Fourier transform decomposes these units in frequency instead of time, resulting in units of Volts^2/Hz , This is usually called Watts/Hz , but the conversion from Volts^2 to Watts not very physically significant, like W = V^2/R
10*log10(power) will result in a unit of dB/Hz , but remember that decibels always represent a comparison between two power levels; you quantify the balance of power. The best decibel definition is 10*log10(P1/P0) , as described here . If you simply plug in the PSD buffer rating in this equation, you set your PSD bit to P1 and implicitly compare it with the value of P0 1. This may be what you want, and it may not. For visualization purposes this is pretty typical, but if you have a standard reference power with which you must compare, you should use it instead of P0 .
Assuming that you are trying to construct an estimate of the power spectral density of dB to convert from Hz to MHz , you simply rescale the x axis of your frequency graph. Remember that MHz is only 1 million Hz, so the only difference is that 240000Hz = 0.24MHz
EDIT The question raised by mtrw is very important; if you are dealing with large amounts of data and averaging FFT vectors, I highly recommend the Multitaper method; This is a much more statistically sound way of sacrificing frequency resolution for greater confidence in evaluating your PSD.