+ Post New Thread
Results 1 to 2 of 2
  1. #1
    Newbie level 6
    Points: 435, Level: 4

    Join Date
    Jan 2017
    0 / 0

    Difference between foundry model and ADS simulation result of inductor

    Hi,there! Recently I have been working on the design of inductor. My method is, first drawing an inductor in Cadence Virtuoso, and then getting its characteristics in ADS after import.
    To verify the validity of my ADS simulation method, I'd like to compare the foundry model(simulated in Cadence, 55nm GlobalFoundry)with ADS simulation results and see if they match.
    The comparison involves S-parameters(S11,S21),inductor, and resistance. Now my problem is, S21 and resistance between the two are a bit far from identical.
    Take the following case for example(please refer to Inductor.JPG), it's a differential inductor made of three top metal layers, the patterned ground shield is down under made in a lower metal(i.e. M1).
    The comparison results are shown in Result_Comparison.JPG: it seems like the faster growth of parasitic resistance has caused a bigger insertion loss(i.e. S21), is that right?
    In light of the difference, there are two settings I am really doubting about:
    1)Substrate. Since I haven't found out the dielectric constants in the process document, I get these "average" values by calculation upon post-simulation capacitance extraction in plate capacitor model.Please refer to Substrate.JPG.Are there any apparent mistakes?
    2)Ground connnection of patterned shield. I finish the ground connection by simply putting a pin over M1’s left edge and then connecting it to ground in the schematic view. I've found out the pin position can slightly affect the simulation results. Should there be any special handling on this?

    •   AltAdvertisement


  2. #2
    Advanced Member level 5
    Points: 13,414, Level: 27

    Join Date
    Apr 2014
    844 / 844

    Re: Difference between foundry model and ADS simulation result of inductor

    Does the extracted inductance show reasonable agreement?

    Your ADS model has less DC series resistance, but faster growth from skin effect and/or substrate losses. The expected DC resistance can be checked by manual calculation using process spec resistances for metal layers and vias. For high frequency response, you really need the correct substrate conductivity (and for more accurate results consider possible EPI layer with different conductivity).

    For the patterned ground shield, connecting the shield to GND and evaluating the differential mode results is correct. You get differential mode results if you place a "floating" port and connect it between the two inductor terminals. ADS then uses a symmetric excitation for that port, so that you shield connected to ADS ground symbol is a mid potential, as desired.

--[[ ]]--