I'm having some problems with AD8368 working in AGC mode. The component is at the input of a RF chain, and the issue is that for an input power variation of 28 dB, I get an output power variation around 3.6 dB. I'm testing the unit with tones, and I'm using the external clamp to prevent AGC overload shown in page 15 of the datasheet. I think the DETO voltage is not correctly generated, because I monitor it to obtain RSSI. The result is that for the lower input (-16 dBm) RSSI result is -18 dBm (due to input losses caused for previous passive components, that could be OK), but for higher input (12 dBm) RSSI result is 7.5 dBm. So difference between input power and measured power increases from 2 dB to 4.5 dB. I've also measured Vdeto when I increment input power 1 dB, and Vdeto variation is 24mV, instead of 26 mV needed to obtain 38 dB/V specified at the datasheet, moreover, output power increases around 0.1 dB for each 1 dB increment of input power.
I attach a pdf file with the schematic of the AD8368 implementation.
Thanks for your help.