Convert -80dBm at 1GHz impulse into a base band impulse

Hello,
I try to explain my problem: i've to convert a RF impulse ,width about 1us or less with frequency up to 1GHz or 2GHz into a baseband impulse viewable with my scope (100MHz).
My first idea was to use a not biased schottky detector and two mmic with +40dB of amplification with very simple design, but the sensivity ( the lowest signal measured with the scope 2mV/DIV) was about -45dBm.
The second step was a biased detector but the sensivity was about -50dBm
Then the last step is use an opamp to convert the high impedance of the detector (10K) to a 50ohm but the sensivity increase at -70dBm.

Can anyone give me some suggestion to increase the sensivity? There is a way to use an active detector instead a passive detector, i've tried with a precision rectifier but it work up to 500MHz.

thanks in advance.

I would think that would work (over a limited dynamic range) given you have a filter to limit the noise power. Typically pulse detection is done with a logarithmic detector/amplifier, probably down converted and filtered first. -100 dBm, 1 MHz pulses should not be a problem given a low noise down conversion and adequate filtering. However keep in mind the fundamental limit of -174 dBm/Hz + 10*LOG(BW), and you need some S/N above that for fidelity.

https://www.analog.com/en/rfif-components/log-ampsdetectors/products/index.html

A broadband preamp seems to be a general and feasible solution.

Thanks for the replies,
my first experiment was using ad8318 but it have the log amplifier ( over seven or lot decades) and any impulse will be confused with noise, infact the schottly diode is almost linear for small signal but for high signal it work with the square law.
For example i used a ZX47-60-S+ of mini-circuits and inside there is an ad8318 with the same capacitor/resistor explained in the evaluation board of analog devices.(!!)

The solution about use another preamplifier is dangerous, with a large gain the circuits became instable.

Any comment will be much apreciated!

thansk again

You have not mentioned how much filtering you have, if any. If you use the full 3.5 GHz bandwidth of the AD8313 then you will not be able to detect anything under -174 dBm/Hz + 10*LOG(3.5 GHz) = -79 dBm. Even then your pulse is the same level as the noise, and you should have 10 dB of S/N so you are talking at least a filter off 350 MHz bandwidth. Ideally you want only a few MHz wide filter so you must down convert since you can't easily get that filter at 1 -2 GHz. A quick way of doing it would be to tap the IF off a spectrum analyzer (usually 70 or 110 MHz), and feed that to a filter, then into your scope or diode detector. You'll need to check the IF BW of your analyzer to make sure it is wide enough. You could even use a radio scanner as some have 10.7 MHz IF out, but usually not enough BW, however narrow band ceramic filters are cheap. If you are trying to cover the whole 1-2 GHz band at once then that requires $$$$$.

Actually my sensivity is about -75dBm i've reached this feature with three mmic with 20dB each , one schottky diode from avago and a video buffer.The video buffer is the last step, because the diode is terminated on 10Kohm then the signal is decoupled with this video buffer (bandwitdh about 100MHz).

I've keeped in mind that i'm working on the edge of noise an signal but with a mixer i've to well know the frequency of the rf pulse, if the frequency is 1.5Ghz instead 1GHz and LO is fixed , this solution doesn't work.

Let me know if i'm wrong

I suspect your mmic amplifiers are feeding back and oscillating on you! I would do something like this.

If the LO is tunable, I would make the IF filter something like 200 Mhz or so to limit the noise bandwidth.

After the detector diode, you could add an opamp video amplifier to boost the signal if desired. Maybe it will not be needed. You probably DO want a lpf in front of the oscilloscope, assuming it is a digital oscilloscope(anti anliasing).