I have modelled DBPSK Transmitter in QUCS. It consists of one shift register, one D flip flop and analog part with two transformers to manipulate the phase of the carrier wave according to the binary data. My test binary pattern is 1010. If I simulate digital part alone it works as needed, and analog part also works as needed. But if I do transient analysis of both parts combined then QUCS hangs. I wanted to simulate the circuit during 4 clock cycles of the shift register. If I simulate during only one cycle it works without hanging. Please, could somebody help me or give some workaround for this problem.. The project is in attachment-- Thanks in advance
Your transmitter.sch seems to work fine on my side. Is the attach figure what you expect?
Which version of Qucs and platform are you using?
Anyway, With transient simulations sometimes it helps to reduce the timestep (or increase the number of steps).
Thanks for the reply,
I am using 0.0.0.17 for Windows. I noticed 0.0.18 is available till 4 days. Yes in your attachment one can see it works fine for you, but for me it does not still work. I also noticed that if I set start time to 0 then for the most simulations initial DC simulation fails, that's why I also give start time more than 0.. Could we ask the developers if the simulation hangs, maybe some diagnostic messages be given in simulation box, there are no messages and one cannot be sure if everything is ok..
In Ubuntu with 0.0.17 the same problem. Increasing number of steps has worthened the problem, I havent tried with 0.0.18 since i am using it in my univercity PC and don't have admin permission to install new programs
I've just compiled qucs 18 for ubuntu, but the problem still there. I tried to simulate only 1 clock cycle and it worked, but with 4 clock cycles it still hang. Please, could smbdy check it for windows or ubuntu. As I noticed it works without any problems for Guilherme under MacOS. Please, if somebody knows some workaround for this problem, help!
Have you tried to increase the number of steps? Change it on the setting of the transient simulation?
Is your system 32 or 64 bits?
Note that 0.0.18 is a snapshot, not a final release.
Confirmed on 32bit system.
It hangs around 20us.
I just run your transmitter.sch on a Debian 7, 32bit and it hangs with the message:
WARNING: Ignore simulation component in subcircuit "piso_shift_rg.sch".
WARNING: delta rejected at t = 1.999e-05, h = 1.729e-08 (no convergence)
WARNING: delta rejected at t = 2.000e-05, h = 5.369e-09 (no convergence)
WARNING: delta rejected at t = 2.000e-05, h = 2.039e-11 (no convergence)
WARNING: delta rejected at t = 2.000e-05, h = 6.554e-12 (no convergence)
WARNING: delta rejected at t = 2.000e-05, h = 4.287e-11 (no convergence)
WARNING: delta rejected at t = 2.000e-05, h = 2.147e-11 (no convergence)
WARNING: delta rejected at t = 2.000e-05, h = 4.571e-12 (no convergence)
You can find it on ~/.qucs/log.txt
This problem might be related to bug 128: https://sourceforge.net/p/qucs/bugs/128/
I will try out a few things to see if I find a workaround.
Here is one workaround.
For some reason the solver is not being able to converge with the sharp transition of 'data' at 20us. It is similar to the bug I mentioned above.
If you put "sort of a low-pass" to smooth 'data', you can help the simulator to step over the non-convergence.
See attached schematic.
My Windows is 64(ubuntu 32), but by default the qucs was installed as 32 programm(I have not paid attention to it).. Thanks for your comments, I'll try to do the suggested workaround
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.