The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> How to design a negative-gm oscillator?
https://designers-guide.org/forum/YaBB.pl?num=1225122628

Message started by Julian18 on Oct 27^th, 2008, 8:50am

Title: How to design a negative-gm oscillator?
Post by Julian18 on Oct 27^th, 2008, 8:50am

Hi, there
I am newbie in designing a oscillator, but now inevitably involved in a VCO project, I am building a negative -gm oscillator, which is shown below in the figure, but when I simulate this using Transient Analysis(Spectre), circuit simply doesn't oscillate, even after the initial condition is given, it just assumes it DC value, without any oscillating observed. I think maybe the start up condition is violated, but couldn't find how to get it satisfied. So I am stuck here.
Anyone can help me out there ? maybe some design equations specific to this circuit has to be satisfied involving the tail current and the gm of the differential pair. etc.
Thanks

Title: Re: How to design a negative-gm oscillator?
Post by didac on Oct 27^th, 2008, 1:02pm

Hi,
I suppose that R mimics the expected tank loses right?, so the input impedance of the differential pair should equal that losses for oscillation(in practice a safety factor is used), Itail will determine if you are operating in a current limited regime or a voltage limited regime so it will control(amongst other things like Phase Noise and power consumption) the output amplitude. But given your description that the oscillator is "dead" maybe you have a problem with the tolerance from transient, are you using the default ones?, I suggest to go an force a maxstep of 1/10 of the expected period and see what happens.
Hope it helps,

Title: Re: How to design a negative-gm oscillator?
Post by Tareeq on Oct 27^th, 2008, 2:21pm

Hi Julian,

1- Start with the Pi model of the inductor modelling all kind of parasitic effect of the inductor in terms of capacitance and resistance. And give it reasonable values so that, when you shall optimize the inductance with asitic things would be just ready (Documentation of asitic and Pi model available here http://rfic.eecs.berkeley.edu/~niknejad/asitic.html)

2- The fact that you dont see anything could means two things:

2a :That the gm of the transistor is not big enough , therefore increase either or both the width and the bias current

2b :Not any transient parameter can make simulation work. You have to be careful. Set: reltol to 10-5, vabstol 10-8, iabstol -13
Select gear2only for the integration method, select "alllocal" for increased acuracy, and gmin reasonably low (10-14).
Also you can use nodeset for initial condition and put an initial voltage higher than the expected common mode of the output signals.

Also, what is the value of the potential VSSD ?

Title: Re: How to design a negative-gm oscillator?
Post by loose-electron on Nov 7^th, 2008, 12:01pm

Putting an initial condition current in the inductors is a good way to get this started.