<< Click to Display Table of Contents >> Navigation: The graphical user interface > Building a circuit > Tricks and pitfalls |
See FAQ.
Known problems:
Issue |
Fix |
Floating sub-network |
Happens with transformers with open terminals (especially ideal). Solution is to add large resistors (>1 Mohm) to ground. |
Small resistors |
Some users tend to add small resistors more measurement purposed. Add Switch|Measuring instead. Also set EPSILON under Sidebar|Simulation to a much lower value than your smallest resistance. |
Timestep |
This must be smaller than the smallest traveling time in distributed parameters transmission lines (and smallest time constant in circuit). Reduce timestep of switch to PI-eqivalent for short lines. In many power electronic studies with frequent switching, the timestep must be reduced. |
Overflow |
Can happen with long transmission lines and small time steps. There is no need to use a line longer than Tmax*c/2 since far end response will not show up in the simulation. It could help to use the giga-version (tpgig.exe) of the solver in some cases. |
Inductor switching |
This will result in numerical oscillations since ATP uses the trapezoidal rule of integration. A fix is to add snubbers to get a smooth transition or damping resistance. This is why the standad inductor (and capacitor) comes with damping options. |
Dependent switches |
Often in power electronics there are dependent switches that should react at the same time instance. Dependent diodes can be captured with turning the GIFU option on. |
Component sequence |
In a few cases the component sequence matters. A MODEL producing output used by other MODELS must come before these in the data case. Use Sidebar|Project to sort the sequence in the object tree. Diodes to be controlled by GIFU must come after this switch and before other GIFU switches. The external field source of an synchronous machine must come after the automatic source for initialization. |
WYSIWYG problems |
In a few cases, the circuit looks properly connected by is actually not. Inside BCTRAN and XFMR the user can check Ext. neutral connections. In this case the neutral point is not automatically connected, but the user has to do this manually externally. The option was added to allow measurement of neutral currents. In effect the node becomes 3-phase in this case and if the user adds a single-phase switch, only phase A will be connected. |