Airgap in transformers?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
to gap or not to gap

It depends on the topology. The five most common topologies for transformer-isolated smps are : the two single-ended, flyback & simple forward, and the three symmetrical, push-pull center-tap, half-bridge, and full-bridge. The single-ended circuits must have air gaps in the transformer core in order to prevent core saturation. Since the cores are driven in one direction only, there is a dc component to the core flux, and saturation is inevitable, unless the core has a sufficient air gap. With the symmetrical circuits, no air gap is needed. These transformer cores are driven in both directions, and the average flux *should be zero*. There are issues however, where the core flux could creep or "walk" upward or downward into saturation. With symmetric converters, any unbalance in the PWM drive applied to the transformer accumulates with each switching cycle. This unbalance could be due to unequal voltage drops in the power FETs, unequal delays in the PWM control IC, etc. Using an IC with *current-mode control* alleviates this issue for push-pull center-tap & full-bridge topologies (true current mode control, not merely pulse by pulse current limiting). The difference in volt-microseconds is compensated by the current control loop. With half-bridge, however, current-mode control corrects the volt-microsecond unbalance, but introduces an amp-microsecond unbalance, which causes the capacitive half-bridge to walk towards one or the other rail voltage. A series cap, or an additional primary winding plus two ultrafast diodes corrects this problem. With all symmetrical converters, an intentional air gap can prevent saturation, but the magnetizing current increases as a result, & efficiency is lowered. Also, there are losses associated with this gap itself. Aren't you glad you asked? Best regards.
 
Yes, this is intended to be the main power transformer.
The design is a push/pull or half bridge, which is to prefer?
As i said, this is going to be an SMPS for a car amp, based on 4*gainclones, 4*40-50W. the power needed thus at least 200W.
At first i had plans using a PWM regulation, but is this really necessary? what more should i think of?
the wandering issue could be a problem, could you explain a bit more, "series cap" where? in series with rails, input voltage or what?
 
smps issues

At the 200 watt level, the choices come down to a simple forward, push-pull center-tap, or half-bridge. The simple forward must be duty cycle limited to around 50%. The one-switch topology may place high stress on the MOSFET, so a two-switch design may be better. This will require a high side gate driver. A half-bridge is a good choice at higher voltages (lower currents). This application is probably best served with a push-pull center-tap topology.

I'd recommend 55 volt MOSFETs (assuming the input supply, the car battery, is 12 volts), a current mode control IC, such as the Texas Instruments (formerly Unitrode) UCC3808AN-2. I recommend a fully regulated circuit, including an output inductor. Also, Schottky diodes are good for output rectifiers. You must compute the peak reverse voltage stress on the Schottky's based on the transformer turns ratio.

The series capacitor is only for the half-bridge network. It goes in series with the transformer primary. I hope I've helped.
 
Re: smps issues

Claude Abraham said:
I recommend a fully regulated circuit, including an output inductor.
Put the inductor(s) in both positive and negative rails just after the rectifier but before the first filter capacitor. Wind it with two pieces of wire, preferably bifilar, and have *exactly* the same amount of turns on each winding. This will help the positive and negative rails to be sort of "locked" to each other.

Important -> Feed the rectifier pos output to the start of one choke winding and the rectifier neg output to the finish of the other winding.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.