These high load transient responses are very usual with voltage regulators and not limited to switchers at all. If you have a look at the datasheet of LM317 for instance, a superb linear regulator, then you will see a 3V spike load transient response (yes, three volts!) on a load step change of 1.5A. So, the LM2576's 150mV spike on a load step change of 3A isn't so bad at all. The brother of LM2576, the LM2676, running at 260kHz, shows only a 100mV spike by the way.

In most cases I have an additional RLC-lowpass filter at the output of switcher, in order to suppress the switching noise. In situations when high load step changes are expected I cure the problem by using big electrolytics at the output of switcher. 4700µF can deliver 3A for a period of 15µsec with only 10mV dropping.

Take care, in applications with high load step changes the finite equivalent series resistance of electrolytics must be taken into consideration. 3A across 50mR causes a voltage drop of about 150mV! So, you need to take several high quality caps in parallel to kill this voltage drop.

Kai Klaas