AnsweredAssumed Answered

xRC: How to extract parasitics w.r.t local substrate?

Question asked by dontpanic on Jan 19, 2016
Latest reply on Jan 29, 2016 by dontpanic

Hello all! Is possible to have Calibre extract parasitic capacitances with respect to multiple ground nodes, according to which local substrate the parasitics are located in?


I have an ADC module with 2 power domains: vdd_A/vss_A (amplifiers' "quiet" analog) and vdd_SW/vss_SW (sampling switches' "noisy" analog). Each domain is implemented in its own deep-Nwell, which means that all the devices in each domain see a "local" p-substrate that is isolated from the local substrate of the other domain, and also from the "global" p-substrate area lying in between the 2 deep-Nwell regions (which in my layout is tied to a node called vss_SUBSTRATE). Thus both in layout and schematic "vss_A", "vss_SW" and "vss_SUBSTRATE" correspond to separate, electrically isolated nodes.


The problem is that in my extractions all the parasitic capacitors are tied to whatever node I specify in the field "Ground node name" in Calibre Interactive: if I specify "vss_A", then the parasitics of the devices in the vdd_SW/vss_SW deep-Nwell region are also tied to "vss_A"; if I specify "vss_SW", then the parasitics from the vdd_A/vss_A deep-Nwell are also tied to "vss_SW"; if I specify "vss_SUBSTRATE", then all the parasitics caps are tied to this node. (BTW, If I specify more than one net name in the "Ground node name" field, then Calibre creates a fictitious net name by concatenation of the names provided, and ties all the parasitics to that fictitious node!). Thus all these extracted netlists are erroneous because they imply a coupling between domains that in reality doesn't exist, due to the extracted parasitics being tied to the same electrical node!


Is there a way I can instruct Calibre to tie all the parasitics in the "vdd_A/vss_A" region to the vss_A node, the ones in the "vdd_SW/vss_SW" region to vss_SW, and only the interconnection parasitics of the routing between these regions (lying above the global substrate) to "vss_SUBSTRATE"?


Thanks in advance for any help!







P.S. I referred without success to the section "Extracting With Multiple Substrates" on the xRC User Manual, but I think it's more related to developing SVRF rules to handle layers for actual physical diffferent substrate materials. In my case the physical layers are the same, the only difference is their geographical location (and their electrical isolation by means of the deep-Nwell walls). Also, the workaround of creating separate layouts and schematics for the various domains and extracting them separately is not really a solution, because of its poor scalability to larger modules in the ADC, which contain even more power domains!