From: Stefan J. <st...@gr...> - 2005-03-14 12:30:42
|
------------------------ Ursprüngliche Nachricht ------------------------- Betreff: Re: [Qucs-help] toying digital Von: "Stefan Jahn" <st...@gr...> Datum: Mo, 14.03.2005, 13:28 An: "Raimund 'Raimi' Jacob" <ra...@lk...> -------------------------------------------------------------------------- > hello *, hi there! > Florian and me were playing around, trying to start a digital > simulation. Attached are NOR, OR, NOT, NAND, AND in TTL (wich seem to work) and an attempt of a NOT in CMOS. the CMOS version does not work at all and we dont understand why. > > symptom is: > ERROR: TR1: Jacobian singular at t = 2.000e-03, aborting transient analysis > > does anyone have an idea what may cause this? is the circuit wrong or is that a shortcoming of the simulator? i had a look at the example and found a small bug in the mosfet model and fixed it. this has no influence on the result of your specific simulation but on others which now work much better (faster convergence). thus at least one problem has been fixed while looking for another... i keep the example in my "NotWorking" project... thanks for trying qucs, stefan. |
From: Stefan J. <st...@gr...> - 2005-04-20 06:20:05
|
Hello! Some news regarding "toying digital": > ------------------------ Ursprüngliche Nachricht ------------------------- > Betreff: Re: [Qucs-help] toying digital > Von: "Stefan Jahn" <st...@gr...> > Datum: Mo, 14.03.2005, 13:28 > An: "Raimund 'Raimi' Jacob" <ra...@lk...> > -------------------------------------------------------------------------- > >> hello *, > > hi there! > >> Florian and me were playing around, trying to start a digital >> simulation. Attached are NOR, OR, NOT, NAND, AND in TTL (wich seem to > work) and an attempt of a NOT in CMOS. the CMOS version does not work at > all and we dont understand why. >> >> symptom is: >> ERROR: TR1: Jacobian singular at t = 2.000e-03, aborting transient >> analysis >> >> does anyone have an idea what may cause this? is the circuit wrong or is > that a shortcoming of the simulator? > > i had a look at the example and found a small bug in the mosfet model and > fixed it. this has no influence on the result of your specific simulation > but on others which now work much better (faster convergence). thus at > least one problem has been fixed while looking for another... > > i keep the example in my "NotWorking" project... The 'cmos-not.sch' has definitely some convergence problems. With the changes I made recently in CVS (improvement of equation system solvers and some changes in the device models as well) the example is now kind of working, if you: 1. set "IntegrationMethod" to "Euler" (property of transient simulation box) AND 2a. set abstol to 1nA and vntol to 1mV OR 2b. put a 100k-100G Ohm resistor between out and gnd Cheers, Stefan. |
From: Raimund 'R. J. <ra...@lk...> - 2005-05-15 13:58:20
|
Stefan Jahn wrote: hello! > The 'cmos-not.sch' has definitely some convergence problems. With the > changes I made recently in CVS (improvement of equation system solvers > and some changes in the device models as well) the example is now kind > of working, if you: > > 1. set "IntegrationMethod" to "Euler" (property of transient simulation > box) AND > 2a. set abstol to 1nA and vntol to 1mV OR > 2b. put a 100k-100G Ohm resistor between out and gnd 2a is enough, but also doing 2b with a 100k resistor makes the simulation faster. so... the cmos-not works now in principle, but i am too dumb to make it work with the correct voltage level. florian: we have to work on this again :) thanks for the investigation, Raimund -- ___ ___ _____________ / /| / /_ / ____/ ___/\ Nothing useful for / / / / _ / / __/ / __\/ more than a decade / /_/_/ \/ /_/_/ /_/ {www.|raimi@} /______/__/\._\.____/\.____/\ .org \._____\._\/\._\.___\/\.___\/ |
From: Stefan J. <st...@gr...> - 2005-05-17 08:33:12
|
On So, 15.05.2005, 15:39, Raimund 'Raimi' Jacob sagte: > hello! hi there, >> The 'cmos-not.sch' has definitely some convergence problems. With the >> changes I made recently in CVS (improvement of equation system solvers >> and some changes in the device models as well) the example is now kind >> of working, if you: >> >> 1. set "IntegrationMethod" to "Euler" (property of transient >> simulation >> box) AND >> 2a. set abstol to 1nA and vntol to 1mV OR >> 2b. put a 100k-100G Ohm resistor between out and gnd > > 2a is enough, but also doing 2b with a 100k resistor makes the > simulation faster. > > so... the cmos-not works now in principle, but i am too dumb to make it > work with the correct voltage level. florian: we have to work on this > again :) this mail hung sometime in your queue, didn't it :-) good luck with further "tyoing digital". > thanks for the investigation, thanks for testing qucs, stefan. |