I've been testing the booster boards to convert a 3.4V input to 12V while down converting the current. I was running the efficiency curves using a BK DC load while sweeping the load current and monitoring the supply current. At a certain load current, the voltage would suddenly drop from 12V to <3.4V and the supply current would match the load current (Figure below). The load current this happens at is different for each board but all below the Iout max of 1.5A. Some boards would start with this behavior. The load current at which the switch occurred was repeatable so it doesn't look like a complete failure of the board but maybe a feature that I don't completely understand.

My test set up is simple. A DC power supply is hooked into the Vin and ground pins on one side and a DC load is hooked into the Vout and ground pins on the other. I followed the test set up recommended in the Keithly Application note "DC_DC Characterization2600B AppNote" (Figure 4). 

Any suggestions on why the boards are acting like this would be appreciated.