Nixphe
Junior Member level 1
Hello,
i'm trying to design a ringoscillator working around 10GHz. I'm quite (very) unexperienced, and i wonder how to start dealing with this design problem.
What i think ...
- i would use bipolar CML invertors
- i would load the outputs of the invertors with 2 known caps + input of next stage
- i would choose the resistors such that 0.69*timeconstant equals half a period (50ps)
- how should i choose current in this case? any reflections?
- any reflections on how to choose R and C?
- minimal die area would be an idea
- chosing R small to have little voltage noise, yet chosing current high for enough swing ...
Part 2 ... changing oscillation frequency by changing current
- i repeat the previous for lowest frequency
- i choose tail current such that i have maximum swing at this frequency
- when i increase tail current, swing can't change, so invertor gets faster, so frequency increases
- anyone thinking this is a good way of thinking? any reflections?
Last part ... how to interconnect different CML invertors. When i would just connect the outputs of one to the inputs of the next i think i might encounter following problems
- normally 100mV is enough to drive CML invertor, more might cause saturation and slow down the circuit.
- can i just limit the swing? so choosing little R/little current
- level shifters might be required if there is some Vcb <> 0 on bjt terminals.
- i don't know how to estimate whether this Vcb will be 0 or not
- can i somehow avoid level shifters?
thanks for your help, i appreciate it a lot. it's difficult to get started ...
Nixphe
-
i'm trying to design a ringoscillator working around 10GHz. I'm quite (very) unexperienced, and i wonder how to start dealing with this design problem.
What i think ...
- i would use bipolar CML invertors
- i would load the outputs of the invertors with 2 known caps + input of next stage
- i would choose the resistors such that 0.69*timeconstant equals half a period (50ps)
- how should i choose current in this case? any reflections?
- any reflections on how to choose R and C?
- minimal die area would be an idea
- chosing R small to have little voltage noise, yet chosing current high for enough swing ...
Part 2 ... changing oscillation frequency by changing current
- i repeat the previous for lowest frequency
- i choose tail current such that i have maximum swing at this frequency
- when i increase tail current, swing can't change, so invertor gets faster, so frequency increases
- anyone thinking this is a good way of thinking? any reflections?
Last part ... how to interconnect different CML invertors. When i would just connect the outputs of one to the inputs of the next i think i might encounter following problems
- normally 100mV is enough to drive CML invertor, more might cause saturation and slow down the circuit.
- can i just limit the swing? so choosing little R/little current
- level shifters might be required if there is some Vcb <> 0 on bjt terminals.
- i don't know how to estimate whether this Vcb will be 0 or not
- can i somehow avoid level shifters?
thanks for your help, i appreciate it a lot. it's difficult to get started ...
Nixphe
-