I have a circuit using the LM3409 with Iadj controlled by an analog voltage, and the on/off or PWM controlled by a digital (3.3V) signal on the EN pin. I have found some unusual behavior which effectivley locks the output. The load is such that occasionally, there may be a momentary short. The current is limited nicely in this situation, but the output voltage latches at the low Vout--within a few 10s of millivolts. In troubleshooting, I think this issue may be that Toff is determined by the RC circuit driven by Vout. Once the short is removed, there is not enough energy in the short pulses separated by the default Toff of 300uS to bring the output voltage back up if a "standard" load is present. The only way to unlatch the situation is to disconnect the load, giving a maximum impedance and thus Vout can be charged. We cannot easily disconnect the load, so we need to have this restart under load after Vout is momentarily dropped. This problem is worse at lower output currents and smaller load voltages.
Another point to emphasize is that the circuits regulate fine once the Vout is above some threshold. In other words, it starts if the initial Vout is greater than some small voltage (say 1 volt). It can approach from above, but not below.
Also, the output does not have to be shorted to see this effect. Just a relatively low voltage load, say 4-8 volts, will latch up near zero output by switching the enable line off. Turning the enable back on results in a much lower output with the 300uS timeout on the Toff. This will remain until the load is disconnected and reconnected.
The circuit is the relatively standard buck controller in the data sheet. We are just driving it using a microprocessor instead of the "standard" analog controls.
Does anyone have some suggestions? I'm a bit lost for things to try at this point.