From: <claudio.girardi@vi...>  20110412 10:56:26

Hello Mike, I didn't think at the function derivative, so I thought I should obtain sqrt(0)=0; besides, it seemed strange also that the other EDD branches failed, even if their functions were valid, but I understand the simulator might not handle this. Thanks and regards, Claudio >Original Message >From: mbrin72043@... >Date: Apr 12, 2011 11:59 AM >To: "Claudio Girardi"<claudio.girardi@... it> >Cc: <qucsdevel@...>, "Stefan Jahn"<stefan@... de> >Subj: Re: [Qucsdevel] Behaviour of EDD component ? > >Hello Claudio > >I downloaded your test schematic and ran it. The results obtained are much as one would expect. In your test network EDD I3 is a sqrt function: two things to note with this function: 1. if V1 < 5, the sqrt(V15) becomes a complex number, and 2. when V1=5 , sqrt(V15) =0 which gives a discontinuity in the partial differential of I3 as the partial differential is proportional to 1/sqrt or at sqrt=0 > infinity, causing Qucs to fail. When using the sqrt function with EDD check to make sure over the voltage range of interest that it does not become complex or go to zero. When using V1 to V8 arguments it is normal practise to add a very small constant to make sure the sqrt never becomes zero. Run the following netlist  it demonstrates adding a constant. > >I hope these notes help. > >Best wishes. > >Mike > 