Voltage*time product of pulse transformer

Question

I am designing a circuit to excite a piezo-electric crystal with 1MHz bipolar square wave at 30V (Vp-p=30V, Vb-p=15V) consisting of 10 pulses, 50% duty cycle.

The idea is to use a pulse transformer whose inputs shall be driven with +15V from either directions to achieve bipolar excitation. First, IN_A will be driven with +15V while IN_B is at ground and vice versa.

1) The voltage * time product in this case should be 15V * (1/1MHz) = 15V-us. Is this correct?

2) Or should it be 30V * (1/1MHz) = 30V-us since the excitation is bipolar?

3) Finally, since the excitation consists of 10 pulses and not just a single pulse, should the calculation of V * t value take that into account as well?

Answer 1

The limitation on transformer volt.seconds is the core flux.

A 1 MHz square wave spends 0.5 us in one state, and 0.5 us in the other.

During the first 0.5 us at 15 V, and assuming it starts from zero flux, the core will go to a flux representing 7.5 Vus. During the second phase, the flux will go back to zero. Subsequent repeats will cycle the core through the same flux swing. If operated like this, you therefore need a 7.5 Vus core specification.

If you can contrive to deliver 'half' a pulse first, so 0.25 us to swing the core to 3.75 Vus, then full pulses of 0.5 us to swing it between +/- 3.75 Vus, finally finishing with half a pulse to get it back to zero, you can use a 3.75 Vus transformer.