I built this circuit, which is supposed to deliver currents up to 20 amps (when everything else works...). The purpose is (eventually) to build a constant current source (as opposed to a constant voltage source) which will be used for anodizing aluminum. To keep the current constant see the alternative feedback path with a shunt and an instrumentational amplifier. But set that aside for the moment.
Let the input voltage be 19V and keep the voltage divider variable (with a poti). Load about 4 Ohm.
The problem: Some output voltages work fine, i.e. the voltage stays constant and the current (drawn from the source) is as expected. But other voltages produce a strange phenomenon: The switcher is on (i.e. mosfet is on almost all the time) so that the current increases over a lot of periods (the voltage at the feedback pin goes way above the reference voltage) and results in a heavy current surge which results in current limiting of my voltage source and voltage drop. The switcher turns off and the current goes back down. This repeats a few times per second.
Here scope shots, when good and bad, zoomed in and out (in time): This is ok, voltage stable. Signal from former pic zoomed in. Caution, yellow signal measured at the input of the mosfet, therefore it is inverted (low=Mosfet on). Here the problem occurs. The voltage at the feedback pin rises far to high (note different volt/div) and the supply voltage drops (seen at the yellow signal). Again a zoom of the bad signal.
Why does this occur?
Next I provide some additional information which may be needed.
C some thousand uF
Mosfet - P-channel
Converters used were LM2576 and MC34063 (scope shots for the latter) with basically the same results.
(Forgot to draw the pull up resistor at base of T1.)
The strange looking use of a transistor wrapped around a diode is to discharge the gate capacitance faster (imagine the lower transistor T2 turns off, then base of T1 is immediately pulled up while the gate capacitance is low -> T1 turns on and pulls gate high with a low resistance). This seems to work all right.
(Also forgot to draw a speed-up Schottky diode for T2, but it's there, promise:-) )
An optional follow up question: When using the shunt circuit mentioned above, the same problem occurs all the time and no predictable behavior is observed.
Edit: There are also large input capacitors. Sorry for having forgotten half of the parts, I sketched the schematic quickly. With input I mean gate, where the yellow trace is taken from. When taking a scope shot on the output of the Mosfet you will see the same on and off signal plus ringing after the current has decreased so that the diode becomes more of a capacitor.
When starting a project, you just set some parameters without anticipating which choices you have to justify afterwards E.g. I took a long time to focus on and develop the needed inductor since this seemed to be the crucial part (at the time I started) while the buck converter just seemed to be not as crucial.
Furthermore, my Layout is a mess, because I have made a lot of changes until now. Give me some time to clean up.