Designer’s Guide Community

umberabbas Junior Member Offline Posts: 13	LC VCO Simulation Help needed !! Sep 13^th, 2017, 8:14pm I need help in LC VCO simulation with differential outputs in Cadence spectre. I have designed a simple circuit but it isn't oscillating at all. I have run a transient analysis but it isn't showing any results. Kindly help.
Back to top	Screenshot_014.png IP Logged

cheap_salary Senior Member Offline Posts: 162	Re: LC VCO Simulation Help needed !! Reply #1 - Sep 13^th, 2017, 10:47pm Delete V0.
Back to top	IP Logged

Ken Kundert Global Moderator Offline Posts: 2384 Silicon Valley	Re: LC VCO Simulation Help needed !! Reply #2 - Sep 14^th, 2017, 8:34pm Oscillators generally do not self-start in SPICE. To start this oscillator you should place a damped sinusoidal current source across the resonator. -Ken
Back to top	IP Logged

umberabbas Junior Member Offline Posts: 13	Re: LC VCO Simulation Help needed !! Reply #3 - Sep 15^th, 2017, 1:02am It is cadence Spectre. I have tried to perform transient analysis but it isn't working.
Back to top	IP Logged

cheap_salary Senior Member Offline Posts: 162	Re: LC VCO Simulation Help needed !! Reply #4 - Sep 15^th, 2017, 3:49am umberabbas wrote on Sep 15^th, 2017, 1:02am: It is cadence Spectre. I have tried to perform transient analysis but it isn't working. Code: The following branches form a loop of rigid branches (shorts) when added to the circuit. V0: p (from vdd to 0) V0 and Vpulse consitute rigid loop. Delete V0.
Back to top	IP Logged

Geoffrey_Coram Senior Fellow Offline Posts: 1998 Massachusetts, USA	Re: LC VCO Simulation Help needed !! Reply #5 - Sep 15^th, 2017, 7:59am umberabbas wrote on Sep 15^th, 2017, 1:02am: It is cadence Spectre. I have tried to perform transient analysis but it isn't working. Ken meant Spice-like simulators (analog/transistor-level), meaning also Spectre. Sometimes you can start an oscillator with a well-chosen initial condition (ic) statement.
Back to top	If at first you do succeed, STOP, raise your standards, and stop wasting your time. IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2384
Silicon Valley

Re: LC VCO Simulation Help needed !!
Reply #6 - Sep 15^th, 2017, 10:35pm

Quote:

I have tried to perform transient analysis but it isn't working.

You should explain what you mean by "it isn't working". Certainly if it is printing an error message you should give it.

I think Cheap Salary has identified the problem. You cannot have two ideal voltage sources in parallel. But I disagree with him that you should keep Vpulse. Vpulse is clearly there simply to start the oscillator, but that is a terrible way to start a differential oscillator. A differential oscillator is naturally immune to changes on the supply voltage. You are much better served by starting the oscillator with a short differential stimulus.

-Ken

Back to top

IP Logged

umberabbas Junior Member Offline Posts: 13	Re: LC VCO Simulation Help needed !! Reply #7 - Sep 19^th, 2017, 8:01pm I have removed V0 along with the resistors so it is giving me this type of output. I have also attached the netlist. i/p voltage is 1v. c1 and c2 are 37fF while L1 and L2 are 83pH. Ibias is 12mA.
Back to top	netlist4.png IP Logged

umberabbas Junior Member Offline Posts: 13	Re: LC VCO Simulation Help needed !! Reply #8 - Sep 19^th, 2017, 8:01pm This one is the output from the circuit.
Back to top	Screenshot4.png IP Logged

Ken Kundert Global Moderator Offline Posts: 2384 Silicon Valley	Re: LC VCO Simulation Help needed !! Reply #9 - Sep 20^th, 2017, 1:44am You should not use changes in the supply voltage to start an oscillator. Differential oscillators are designed to be immune to variation in the supply voltage. A better approach is to apply a differential transient stimulus directly across the resonator. I recommend a damped sinuosoid. -Ken
Back to top	IP Logged

umberabbas Junior Member Offline Posts: 13	Re: LC VCO Simulation Help needed !! Reply #10 - Sep 22^nd, 2017, 4:17am I did the same as Ken specified. The circuit is kinda oscillating but not at the desired frequency. It is oscillating around 46.1 GHz but this doesn't sound right because the oscillation is going down in negative values.
Back to top	Screenshot5.png IP Logged

umberabbas Junior Member Offline Posts: 13	Re: LC VCO Simulation Help needed !! Reply #11 - Sep 22^nd, 2017, 4:18am Netlist for the above generated waveform.
Back to top	Screenshot-5_002.png IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1998
Massachusetts, USA

Re: LC VCO Simulation Help needed !!
Reply #12 - Sep 22^nd, 2017, 4:47am

umberabbas wrote on Sep 19^th, 2017, 8:01pm:

I have removed V0 ....

But I still see
V0 (vdd! 0) vsource dc=1 type=dc
in the screen-shot. (BTW: if you put the netlist in a "code" block, then others could copy&paste, rather than having to re-type. I'm not sure if it would be useful, since we don't have the device models, but we might be able to use a basic model.)

Back to top

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2384
Silicon Valley

Re: LC VCO Simulation Help needed !!
Reply #13 - Sep 22^nd, 2017, 9:01am

Well, you did not indicate what you plotted, but I assume it is the differential output voltage, and I would certainly expect that to go negative.

I think your problem has been solved.

You might consider making the amplitude of your tickler source larger so your oscillation starts up faster. Doing so would speed up your simulations.

-Ken

Back to top

IP Logged

cheap_salary

Senior Member

Offline

Posts: 162

Re: LC VCO Simulation Help needed !!
Reply #14 - Sep 22^nd, 2017, 8:59pm

umberabbas wrote on Sep 22^nd, 2017, 4:17am:

I did the same as Ken specified.

No.
You invoke supply voltage pertubation(step pulse) for kicking oscillator start up.

You don't set "maxstep" in previous netlist.
http://www.designers-guide.org/Forum/YaBB.pl?num=1505358844/7#7

Now you set "mastep=(1/60G*20)" and "method=trap" in new netlist.
http://www.designers-guide.org/Forum/YaBB.pl?num=1505358844/11#11

However I don't think your oscillator work with your netlist, since you set "delay=2" in V1.

I think "delay" is around 5nsec.
Show us true netlist.

Geoffrey_Coram wrote on Sep 22^nd, 2017, 4:47am:

umberabbas wrote on Sep 19^th, 2017, 8:01pm:

I have removed V0 ....

But I still see
V0 (vdd! 0) vsource dc=1 type=dc
in the screen-shot.

"vdd!" is a lonley node.

Back to top

« Last Edit: Sep 23^rd, 2017, 5:07am by cheap_salary »

IP Logged