On 07/02/17 15:04, Seth Forshee wrote: > On Tue, Feb 07, 2017 at 08:52:30AM +0100, Johannes Berg wrote: >> On Mon, 2017-02-06 at 14:50 -0600, Seth Forshee wrote: >>> + (5150 - 5250 @ 80), (23), AUTO-BW >> >> For 20 MHz this is right, since the document says: >> >> The maximum e.i.r.p. shall not exceed 200 mW or 10 + 10 log10 B, >> dBm, whichever power is less. B is the 99% emission bandwidth in >> megahertz. The e.i.r.p. spectral density shall not exceed 10 dBm in >> any 1.0 MHz band. >> >> However, for < 20 MHz, like 5 or 10, which we in theory do support, you >> can't use that much due to the spectral density requirement. It seems >> possible that that's where the 17 dBm came from, since 10 + 10 log10(5) >> is about 17. > > Yes, thanks. I had forgotten to consider bandwidths less than 20 MHz. > > Seth > > _______________________________________________ > wireless-regdb mailing list > wireless-regdb at lists.infradead.org > http://lists.infradead.org/mailman/listinfo/wireless-regdb > According to .../include/net/cfg80211.h >static inline int >ieee80211_chandef_max_power(struct cfg80211_chan_def *chandef) >{ > switch (chandef->width) { > case NL80211_CHAN_WIDTH_5: > return min(chandef->chan->max_reg_power - 6, > chandef->chan->max_power); > case NL80211_CHAN_WIDTH_10: > return min(chandef->chan->max_reg_power - 3, > chandef->chan->max_power); > default: > break; > } > return chandef->chan->max_power; >} This looks to me like it already handles the reduction of power according to bandwidth. BR Matthias
