I've been studying SMPSs for a while out of personal interest, and I stumbled upon an old schematic for a discrete component-based SMPS:
As I understand it, this is a self-oscillating SMPS operating in current mode. T302 is the main switcher, Th301 is the turn-off SCR, T301 acts as the voltage error amplifier whose output controls the maximum current on the primary coil. In particular, the cathode of Th301 is subject to a sawtooth negative voltage due to the primary coil current flowing through R317, while the gate voltage is controlled by the error voltage amplifier T301; the main switcher is turned off on a cycle-by-cycle basis when Vgk on Th301 reaches the turn-on threshold, and this determes the duration of Ton.
The general principle of current mode operation is that a lower than expected voltage on the secondary should result in a longer Ton. My problem is that in this case it seems that the opposite is true! In fact, the secondary rectified voltage drives the base of T301: a higher secondary voltage results in a lower current on R304 and thus in a lower Vg on Th301. This, in turn, means that a higher magnitude of primary coil current is needed on R317 to turn on Th301, resulting in a longer Ton. In short: the higher the secondary voltage the longer the Ton, positive feedback, BOOM!
I haven't got the real thing on hand so I cannot check the waveforms. I assume it's been built and it works, so I'm obviously missing something here.
Can anyone shed some light?
Thanks in advance,