level shifter with low phase noise

[HSSN]

I am looking for a level shifter with low phase noise which shifts from 1v to 0.75v input common mode and swing from VH=1.2v & VL=0.8v to VH=0.95v & VL=0.55v

 

[jiripolivka]

The phase noise is a feature of an oscillator. In amplifiers, contribution is by several orders lower. The only remaining thing for you is to define your frequency range- this directs to an opamp to be used. Then its noise can contribute in a defined manner but the phase noise does not.

---------- Post added at 19:59 ---------- Previous post was at 19:58 ----------

For your voltage values, a simple resistor voltage divider is the best device. No phase noise added.

 

[HSSN]

am working with CML dividers @ 8GHz so the simple resistor will limit the frequency and when i used a simple inverter it contributes about 90% of the phase noise ???!!!

 

[jiripolivka]

am working with CML dividers @ 8GHz so the simple resistor will limit the frequency and when i used a simple inverter it contributes about 90% of the phase noise ???!!!

Ok, I am working up to 110 GHz, and never heard about a resistor limiting a frequency response. Good pure resistor in a 50-Ohm network will reduce VOLTAGE as you wrote you need.
How do you measure VOLTAGE at 8 GHz? What is your line impedance?

I used frequency dividers at above 25 GHz and had no problem with phase noise. I know there is a very small contribution of amplifiers to phase noise but why a simple inverting amplifier can add phase noise? What is 90% of phase noise? We measure it in dBc/Hz for specified separation from a carrier.

 

[HSSN]

the phase noise profile before the inverter was (fc=15k and noise floor =-149dBc/Hz ) but after the inverter added (fc=100k & noise floor =-130dBc/Hz)
the simple resistors need also higher current to be injected ??
i didnot measure the line impedance

 

[jiripolivka]

the phase noise profile before the inverter was (fc=15k and noise floor =-149dBc/Hz ) but after the inverter added (fc=100k & noise floor =-130dBc/Hz)
the simple resistors need also higher current to be injected ??
i didnot measure the line impedance

I think you should at least know the line impedance across which you measure the voltage. At 8 GHz we usually measure rather the power.

Did you filter well the DC power for the inverter? Can you adjust inverter input level by varying the series resistor? I used SMD resistors up to ~10 GHz and did not see any effect on frequency spectrum.
If you measure the voltage by a fast oscilloscope (sampling I assume), its probe does affect the impedance. I guess the voltage is not the problem but some instability of the inverter, or, crosstalk through the DC line or signal passing by induction to places where it should not be.
It there a visible effect on the eye diagram through the system, with and without the inverter? What effect has connecting two or three inverters cascaded? Is then the phase noise floor growing by their number, or not?