Transformer secondary: grounded/1Meg/or nothing?


All three cases a), b), and c) gives in the end the same results.
Case a) gives a warning, but at least in this case signals can be displayed. In case b) and case c) no warning/error are received.

My question: from all these 3 cases, what case is the correct one and on what case should I rely on?




If you probe a node, for example, V(a), you are actually probing the voltage difference between node a and GND. When you say 1V, it can either be a differential voltage between two nodes or a potential between a node and a reference (0V node), which is generally referred to as GND. (In Qspice, this GND symbol is 0V node)

So, in the case of (a), think about what what to expect V(c) or V(d)? Since the primary and secondary are transformer isolated, and without a definition of a 0V node in the secondary, V(c) and V(d) become floating. This means you cannot define the potential of V(c) and V(d) to 0V.

In this case, Qspice gives you a warning and adds gshunt=1e-13, which represents the addition of a shunt conductance from every node to ground (refer to .option gshunt). Now, it forces every node with a 1/1e-13 ohms resistor to GND and establishes a relationship with GND for nodes C and D. This is how the circuit looks like after gshunt is added.

On what case should I rely on?

In short, you should define a ground relationship in the secondary using any method you prefer. Personally, I find a node where practically my differential probe negative terminal is normally connected and force that to be the secondary ground in the simulation.

I have noticed that people don’t have a problem understanding that a circuit must have a ground (GND) for the simulation to work (all of us are ok to add a ground, as shown in left figure). However, when working on transformer simulation, many people struggle to accept the concept that the secondary side must also have a ground connection. It is similar to how you have to provide a ground connection in the output side of an E-source (as shown in right figure).

In above, case (b) and (c) both OK! There is no right or wrong in where you define the GND in secondary, as secondary is floating entirely, and you need a node connect to GND for simulation purpose. If not, you get a warning message as spice have to calculate absolute potential of those floating node.