Fix a thinko in computing the log10 table.

(This shouldn't actually affect the resulting SNR since it's just a constant offset, and the errors in signal and noise will cancel out)
This commit is contained in:
Oliver Jowett 2014-09-23 14:05:25 +01:00
parent 76c958b03e
commit 83d256e984

View file

@ -189,7 +189,7 @@ void modesInit(void) {
// Prepare the log10 lookup table.
// This maps from a magnitude value x (scaled as above) to 100log10(x)
for (i = 0; i <= 65535; i++) {
int l10 = (int) round(100 * log10( (i + 365.4798) * 258.433254) );
int l10 = (int) round(100 * log10( (i + 365.4798) / 258.433254) );
Modes.log10lut[i] = (uint16_t) ((l10 < 65535 ? l10 : 65535));
}