Delta sigma transient simulation in Spectre

charge injection simulation spectre

I am running into a problem when I was trying to simulate my DSADC in spectre. The entire system is using ideal components (no transistors, but there are VCVS, R, C) and Verilog models. When I was doing transient sim using "Liberal" setting, for about 60us, it was running fine and giving correct output (noise shaping and all that) but then at about 60us, all of a suddent, there is a "Zero diagonal found in Jacobian at XXXX net and YYYY net". and then the output starts to just oscillate and I cannot figure out why.

One more thing is that my ADC will work ONLY IF I do the transient in "Liberal" setting, if I try to do it in "moderate" or "conservative", then it wont work at all (The output just stays at a constant voltage).

I was wondering if anyone out there can help me with this.......

Thanks

charge conservation in spectre

i have the same question in simulation of sdm which is modelled by ideal components. i think that the charge conservation law isn't hold due to the usage of ideal components, so adjust some components such as ideal switch instead by real MOS switch.

My DSM is continuous time, so since there is no switched capacitor, i dont think there will be conservation of charge problem.....

So did ur problem go away once u put real switch in replace of the ideal ones?

Thanks

yes, when i use real mos switches the simulation is normal. i'm not familiar with ct-sdm, but there are switches in loop, maybe you can try it.

Ya, so I switched out my ideal switches to real MOS switches but i m getting some really weird things....first there are spikes when it's not supposed to be....I swear they are NOT charge injection coz they are happening when the clk isn't switching.... and they are at much higher frequency than the clk.....

any ideas??

Thanks