LDO linear regulator output problem

1. the disabled output is a bit of a mystery unless you exceeded the Absolute Maximum Ratings.

2. Pulse test with over 1.5W power drop and up to 150'C Functional range and Thermal resistance, you are exceeding 150'C which could easily occur in 100ms with your scope scale as 1s/div. The device would cool down slowly then recycle.

The mystery is the Absolute maximum temp of 150'C is also the same as the maximum functional operating temp of 150'C and you were clearing exceeding it. A discussion with your Distributor to customer support would be advised after you repeat the test with a new part to be sure. If this OTP is unlikely to significantly reduce the lifespan in a short period, it may be just an isolated defect. Was it a sconnection failure and the EN input was just floating high?

The real problem is your design choice to use the excessive-high input voltage.

Reconsider your design specs and insert a 2W series resistor if necessary to drop the input to 5.5 min.

Problem #1

When the output finally turns off in your test, it follows a decay path that tells of an RC influence. There is some current being discharged through the high internal resistance (spec'ed at up to 2.6MOhm). The TLS715B0 datasheet also shows a fairly low threshold for EN, using the bandgap reference; a comparator may have some bias/leakage that counter-acts the pull-down.

Solution

Add an external pull-down (1.5k, 1/4W would be more than sufficient) if you want to control EN, otherwise just short-circuit EN to your input voltage so the LDO is always enabled.

Problem #2

You have an unstable feedback loop between the LDO's protection circuits and the software-controlled constant-current load.

You are pushing the edge of the regulator's thermal and current limits. The thermal protection circuit may be participating, but I think it is a minor player from how fast this is oscillating. The over-current protection recovers more quickly and is likely more sensitive at high temperature.

When you increase the input voltage, you increase the voltage drop across the LDO and thus its absorbed power. This starts to engage the protection circuit(s), dropping the output voltage. The constant-current load senses its voltage dropping and tries to maintain a constant current by decreasing its resistance. In the meantime, the protection circuit is successful and the output is again activated - with a stronger (lower R) load - causing the output current to max out. Lather, rinse, repeat. Each cycle happens fast enough that the software controlled load cannot react to stabilize the output current.

Experiment

Repeat your test with an equivalent resistive load; the thermal protection may still kick in (it is consuming (18V-5V)*150mA = 2W), but the output current will not spike over the limit and it should not oscillate so fast.

Regarding Oscillation: Assuming an input voltage of 16V, a output voltage of 5V and a load current of 150mA implies that the LDO has to dissipate around 1.6W. Following the calculation done in section 6.3 of the datasheet we end up with something close to 33 K/W for the maximum allowed thermal resistance. If this is exceeded the thermal shutdown will be triggered. According to section "5.1 Voltage Regulation" of the datasheet that can lead to oscillations (LDO overheats, turns off, cools down, restarts, overheats, ...).

Possible solutions if my considers are right: reduce the load or decrease the thermal resistance.

Regarding turn on behaviour: Can you confirm that behaviour over multiple devices? For me it looks like the EN is not properly connected/externally pulled down and slowly discharged over the internal pull down after turn on. Simply connecting the EN to GND might do the trick if it is currently not externally pulled down.

Try using a better oscilloscope. Higher Bandwidth. I have seen a case where a narrow bandwidth oscilloscope showed a DC current through a resistor when no current should be flowing.