[SOLVED] VCO no oscillation at the output

[rty94]

Hello,
I am trying to design a simple vco with lc tank but when I run transient analysis it seems like an offset at the vdd and oscillation doesn't start at all ..How could this problem be solved?

Thanks!

 

[FvM]

The decaying oscillation during the first nanosecond suggests that circuit has too low gain respectively the LC too low Q to maintain oscillations. Operation point looks fine.

You could measure the oscillator loop gain to check how far it's from fulfilling the oscillation condition.

 

[BigBoss]

Try to inject a time limited current into circuit to excite the VCO. ( look at rflib of Cadence ADE for inspiration-Diff_osc )
The initial conditions are very important in oscillator designing and there are some options ( ramp-up,DC,transient assisted etc.) to start-up the oscillators in SpectreRF simulator.
Read carefully the SpectreRf manual and try to tickle the oscillator.There are other possibilites of course ad FvM has mentioned.
Have you ever checked the "negative resistance" of the VCO under this DC biasing condition ?? I think it's a bit "poor"..

 

[rty94]

It was an initial approach just to understand how an oscillator works.I will try to improve it according to your suggestions.Thanks for your replies

 

[frankrose]

Hi,
Set integration algorithm in the tran analysis options to 'traponly'. Default is 'gear2' which supress oscillation. And let more time for the oscillation to wake up.

 

[BradtheRad]

The schematics I've seen provide a supply wire to the node between the inductors. Your schematic appears to have wires overlapping at that location, not connected.

- - - Updated - - -

Also:
To get oscillations going, it helps to create asymmetry of component values. The aim is to start current flow through one mosfet, until volt levels change to the point where the imbalance causes a change of state.

 

[rty94]

Hi again, I 've used ipulse from analoglib for starting the oscillation + in transient analysis I've setted the algorithm as traponly so it seems working .. Is there any better way to start-up the circuit? cause ipulse is ideal..

 

[frankrose]

sure, add initial condition for the inductor or the capacitor. for the inductor it means initial current, for the capacitor initial voltage. you can set at the properties for both. some uA/mV should be enough... other way to add transient noise but it will slows down the simulation more.

 

[BigBoss]

Hi again, I 've used ipulse from analoglib for starting the oscillation + in transient analysis I've setted the algorithm as traponly so it seems working .. Is there any better way to start-up the circuit? cause ipulse is ideal..
View attachment 150258

In simulations, initial values V-I of capacitors,resistors and other elements come from DC simulation therefore oscillators cannot start in such some cases ( equilibrium situation).
It's a trick to start-up an oscillator in simulation environment but noise or any transient regime will start up the oscillator.
That Current Source should be piece-wise type so it will be off after oscillator reaches to steady state.