0
\$\begingroup\$

I have designed amplifiers, using potential division method.

What are the steps to design differential amplifier using inversion coefficient based design methodology?

Please provide links/references of designs with this technique.

\$\endgroup\$
2
  • \$\begingroup\$ I believe you're talking about the feedback coefficient (β). Am I right to assume so? \$\endgroup\$ Jul 19 '17 at 14:10
  • \$\begingroup\$ No. I want to start from designing common source amplifier, differential amplifier and opamp using this technique \$\endgroup\$ Jul 20 '17 at 14:52
1
\$\begingroup\$

The paper by Su, et al gives a specific example of designing a high temperature OTA, using design methodologies around IC parameter. They generally point out that inversion coefficient design can start by targeting weak, moderate, or strong inversion, ideally independent of process. And so one can start out by finding which region optimizes a designer's given constraints. You can plot out a curve of some desired parameter (gm/Id, ft, etc) vs IC, to get a sense of what value of IC corresponds to what region. Then use that IC as a starting point.

According to the authors, "In [13]*, moderate inversion optimizes the tradeoff between gain, speed and power consumption." So right away, you can zoom in on designing around this region to optimize the constraints mentioned earlier.

The authors also recommend modeling using EKV, instead of BSIM models for accuracy in moderate inversion.

Here's a blog with a more concrete realization of the above steps, based upon a design seminar by Willy Sanson. In it, Sanson shows one only needs the process independent IC (based on region as above) and channel length, to optimally design an OTA with gain and bandwidth constraints.

enter image description here

enter image description here

*The source they refer to in 13, is David Binkley's text "Tradeoffs, and Optimization in Analog CMOS design." Binkly is a proponent of using IC modelling for porting over designs in different processes (esp. sub micron), and discusses it at length in his text.

\$\endgroup\$
0
\$\begingroup\$

Assuming this is limited to CMOS RF design; but you must do your own homework to find links e.g. Researchgate, Google Scholar, Microsoft Academic.

a priori

  • Choose the SPECS for BW(f), Linearity, 3OI, Power Gain (Gmin), Noise Figure, (NFmax), Power (Pmax) standby power, etc.
  • get select all spectral values of each active and passive component after selecting Inversion Coefficient(IC) vs channel length(L),
    • simulate IC vs {Vgs, Cgs, Id, fT, gm, gm/Id,...}
    • define IC, Id, L,
    • compute W, gm, gds, Cgs, Cgd,
    • verify specs and repeat until converged or all tradeoffs are met.

Design Optimization Methodology Based on Inversion Coefficient Parameter for CMOS RF Circuits. Available from: https://www.researchgate.net/publication/268219100_Design_Optimization_Methodology_Based_on_Inversion_Coefficient_Parameter_for_CMOS_RF_Circuits [accessed Jul 19, 2017].

\$\endgroup\$
1
  • \$\begingroup\$ Thanks for the advice. The design specs are gain, ft and power dissipation. I started with determining the technology current for nmos and pmos. \$\endgroup\$ Aug 5 '17 at 6:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.