I’m gonna swallow my pride. I have a very simple TPS56121 design. I cannot figure out why it consistently trips its over-current protection (OCP) circuit at about 10A.
Anyone want to take a shot? Let me share the specifics:
DESIGN:
Vin: 12V
Vout: 6V
Max Output Current: 15A
Switch Frequency: 500kHz
Output Inductance: 1.5uH
Output Capacitance: 590uF
Controller: TPS56121 (Constant Conduction, Voltage-Mode Control)
SYMPTOM 1: False tripping of over-current protection (OCP)
At around 10A output current, the OCP on the high-side MOSFET trips, shutting down the converter. This happens regardless of the output voltage (adjustable from 3V to 9V) or the switch frequency (selectable 250kHz, 500kHz, or 1MHz). This high-side OCP is hard-wired to 34A (according to the datasheet). Before the trip, there are no signs of problems, no variation in the duty cycle, no transients on the output voltage. The FET just cuts out and the chip turns off. I can tell it is the high-side FET by watching the pulse-limiting happen on my scope.
Increasing the low-side OCP set resistor (R1A) has no effect. Removing R1A disables all OCP and the converter runs fine at its rated 15A with no stability or thermal problems.
SYMPTOM 2: Small oscillation after high-slew current steps
I don’t know if this is related, but I have noticed that after large, high-slew current steps of (roughly) at least 5A magnitude and 1A/us slew rate, a small oscillation appears. On the output, this is only about 100mV peak-to-peak. The frequency is about 45kHz, which is near the crossover frequency of my loop transfer function. I’ve done a lot of tweaking of the compensation network to try and fix this, increasing the phase margin and decreasing the crossover frequency, but with no visible change in behavior. Otherwise, transient response appears to be fine.
Any thoughts? For what it’s worth, the board is already working in my application driving about 22 low-power servos with no problems. OCP is turned off and the slew rates aren’t high enough to cause oscillations. I want to understand the problem though.
Best,
- Daniel
----
Schematic:
Normal, Un-triggered Output:
Output with 45kHz oscillation:
Duty-cycle variation as a result of 45kHz oscillations:
