@jgardona, Without your schematic, I am not exactly sure why you are getting a singular matrix and timestep error. However, the solution is NOT to add an extra GND. If you add a GND at R1, this will create a short circuit path for one of the diodes, resulting in half-wave rectification.
Your circuit is a full-wave rectifier; you can either have the GND at the source OR at the output, but not both. Here is a replication of your circuit from @macsky’s work, showing how to obtain a full-wave rectified output with these two GND locations.
Thank you. I really think it was strange, because in other simulators it will work. But qspice has better performance, and I was trying to solve this problem. Very nice examples you gave.
If you cannot identify what went wrong in your original schematic and still need help with it, you can consider uploading the schematic for review. If that is no longer important, this case can also be closed here.
It seems the simulation encounter convergence issue in V(N03) during first few timesteps. You are calling a diode “D” directly without a physical or behavioral model. During a DC sweep test, the default “D” exhibits an extremely small reverse current. The discontinuity of the diode or extremely small reverse current seems to make it very difficult to solve this circuit.
There are several options you can try to get this simulation to run (any one of these options can work in my verification)
Add .option gmin=1e-10 , this add minimum conductance in parallel to diode and help solving in reverse direction
Add .option fastmath=0 , force to use Qspice80.exe for more precision math
Add .model D D Ron=1m Roff=100Meg Vfwd=0 , to give a model for D
But, the rule of thumb is to prevent discontinuity in the I-V curve. I don’t recommend just call a D as diode for simulation.