Diode detectors have low sensitivity.
You can use one of the Analog Devices Log Detectors AD8313 or ADL5513, and place in front of them two or three MMICs from Minicircuits.
It can't work, unless you reduce the bandwidth considerably ( e.g. < 1 MHz). Otherwise the input referred noise power will be already higher than -100 dBm.
Unfortunately, bandwidth is 75Mhz, centered @ 1842.5 Mhz.
Such detections are common in Tower mount amplifiers. MMICs with inbuilt bypass are already available with Hettite, Avago, Toshiba, Freescale etc. But I've to do this with discreet components.
Is there any other way to bypass amplifier circuit when Rx signal power exceeds prescribed level ?
You can use BGA728L7 from Infineon which has high and low gain mode (+15dB and -5dB).
**broken link removed**
Beware that is specified for 100MHz to 1.7GHz frequency range but I use it successfully up to 2.1GHz.
The only issue for you could be the package.
Another option is to use a standard MMIC and an RF switch to shunt it for a through state (low gain). Check Infineon for parts. In this case is possible to need two RF switches in series to increase the OFF isolation, otherwise the MMIC could become unstable.
You can expect a noise level of about -85 to -90 dBm with 75 MHz bandwidth, if this is acceptable, you can use a
wide band detector. Otherwise, it has to be frequency selective.
1) So what you are proposing is - amplifying the signal and then comparing it with some reference. ? I'll check said LNA, but it'll take some time.
Or shall i first amplify the signal through LNA to be bypassed and check its level and switch on the bypass path if necessary ?
Further, I dont think noise will b a problem as all I need to do is detect signal power level and provide bypass if its above certain level. Noise may be tolerated in this case as application needs simple comparison.