What is the best scheme for 10GHz x10 multiplication?

Hello!
For 1.25GHz obtaining 10GHz x8 multiplication is easy using 3 FET doublers (1.25GHz x2 x2 x2 = 10GHz) or 2 FET doublers + diode doubler
But for 1GHz x 10 = 10GHz multiplication doubler can be used only one time. I used x5x2 scheme, but have poor results and big power consumption (2xERA3 for x5 multiplication, 2xCEL 10GHz FET for x2 multiplication and amplification).

Input is 0dBm - > ERA3 standard biasing -> ERA3 saturated -> microstrip bandpass -> CEL FET doubler -> CEL FET amp -> far below 0dBm!!! Just weak signal above noise on receiver, so i guess it is -60dBm or so. Total waste of 4 active devices. Still searching problem in my design.

Recently i found info about **broken link removed** (frequency marker using LNB schottky diode).
Author says:

At the Rutherford Appleton Laboratories there was an opportunity to look on a spectrum analyser to 47G. The results were very good. With a small horn on the Tektronix mixer the signal was above the noise floor up to about 4 ft away. Since there was no coupling into a guide the best estimate of output was -60dBm to -40dBm. These are respectable signal levels! It makes you wonder what it would be like on a bit of Duroid and coupled into a guide.

Click to expand...

Yes, it makes me wonder. And similar constructions: http://www.g3pho.free-online.co.uk/microwaves/marker.html
for 10GHz: http://www.qsl.net/iw1fyv/marker_en.html

So i am wondering can i do x10 using less active devices/stages?
For example: 0dBm input -> ERA3SM near saturation -> diode (x10) -> band-passfilter -> FET amplifier -> Can i get 0dBm of that?

I can try, but very tired and disappointed constructing x5x2 multiplier.

If it were me, I would either do a PLL, or use a Hittite X8 HMC444LP4

Hope the following Link helps.

**broken link removed**