When/where to open and close a transformers output
The arguments against zero crossing switching transformers applies to primary of secondary. The claim (largely borne out in practice) is that the core saturates when subject to the rising voltage while unmagnetised. This is generally not an intuitive outcome. It makes sense when you consider that in an inductor under steady state conditions voltage and current are 90 degrees out of phase. By switching the voltage on at the peak of the voltage waveform you have maximum voltage and zero current so the 90 degree relationship is automatically established as the initial condition. You still need to build the "magnetising current field" but you are at the best starting point.
Nicely summed up in the final reference below:
Instead, it would be better to close the switch at the peak of the input AC line voltage. Since the inductor's current is initially zero as before, switching this way puts the applied voltage and the inductor's current immediately in quadrature with each other (or really close to it) and there is no transient event or current settling phenomenon.
The current starts out already settled.
At a minimum what happens is that the current waveform is offset relative to zero so that a current sinusoid DC offset by Ipeak/2 initially occurs. This leads to about double expected peak currents if the core does not saturate - and some sources suggest even higher initial surges, due to saturation effects.
The subject is covered reasonably well in Rod Elliot's inrush current -
and especially section 4. "Inductive & Transformer Inrush".
Useful paper here - less severe claims
Effect of Switching angle on Magnetizing flux and Inrush current of a Transformer
(ie saturation effects not dealt with)
Open electrical wiki - transformer inrush
Some practical experiences reported here To Zero Cross or Not To Zero Cross.
He comments:
Some years back, I was the project engineer on a large component burn-in system. The system was supplied off of a 3-phase power transformer so that the loads could be as balanced as possible. I had designed TRIAC controllers to ramp up the line voltage to minimize the line surge with +90 degree phase shifting. Each power supply was a very large, 50lb, power transformer with multiple secondaries. To tell a long story in short, the 'chief' engineer insisted on replacing the three small power transformers I had specified with the one big monster and he also insisted on using zero cross over phase control to minimize the current surges despite my arguing for the other method. To him, zero voltage across the primary meant zero current....
The PC boards were finished per the Chief's changes and a few power supplies were built. When they were switched on, there was a snap and flash of light in each one of them, a PCB trace was vaporized due to the huge magnetizing current resulting from the zero cross over phase control. Since the boards had already been built, modifying them was out of the question, the only thing I could do was to place a large ballast resistor in the primary to limit the current surge. The Chief refused to accept the premise that it was his zero cross over and monster power transformer causing the problem.
Connecting the transformer primary to the AC line at the peak of the waveform prevents excessive core saturation and this applies with or without secondary load. First point to note.
Secondary current neither increases nor decreases core saturation\$^1\$ so whenever you decide to draw current from the secondary, the core flux will not change because secondary ampere-turns are exactly cancelled by primary ampere-turns due to the load.
\$^1\$ In fact a higher secondary current has a tendency to reduce core saturation due to small volt-drops in leakage components. There volt drops slightly lower the supply voltage reaching the magnetization inductance of the primary hence core saturation is slightly reduced.