Link budget template for any point-to-point radio link

Worth recording for the next person who hits it.

Short answer: Start with transmit power, add both antenna gains, then subtract all the losses.

In more detail:
1. Free-space path loss is the baseline; add margins for rain, foliage, and multipath.
2. Receiver sensitivity depends on the required data rate, bandwidth, and noise figure.
3. The margin you calculate is meaningless unless you also account for interference and fading statistics.

Will post an update once we have the next batch measured.

This template is a good starting point, and the part most often missed is keeping the different kinds of loss in separate columns rather than folding everything into one margin figure. Free-space path loss is only the baseline. Once it is mixed with rain, foliage, multipath, cable loss, and antenna misalignment, it becomes impossible to tell which assumption failed when the measured link comes in short.
Two items deserve their own lines. First, use realized antenna gain, not the peak gain from a datasheet. Mounting, nearby metal, polarization mismatch, and pointing error all reduce the gain that actually appears in the link, and on a point-to-point link a small pointing error can cost more than a generous rain margin. Second, state receiver sensitivity at the required modulation and error rate, not at the lowest rate the radio supports. A receiver that is sensitive at a slow robust rate may need several dB more signal at the rate the link is actually planned to run, and that difference should appear explicitly in the budget.
Fade margin also needs a stated availability target. A margin that is adequate for 99 percent availability can be far too small for 99.9 percent, especially where multipath or rain dominates. I would record the margin together with the fading model and the time percentage it is meant to cover, so a later reader does not treat it as a universal safety factor.
Finally, keep a measured column beside the predicted column. Log transmit power, cable and connector loss, antenna gain as installed, and received signal level for each link, then compare prediction with measurement. When the next batch is measured, that comparison will show whether the template needs a larger multipath allowance, a pointing correction, or a different sensitivity figure, rather than just a larger overall margin.

Thanks for sharing, very thorough. We learned this lesson the hard way too.