TPS2490 does not latch off when fault occurs.

Please send waveforms showing that the behavior is the same as Figure 17 in the datasheet. We need to see the voltage on the TIMER pin especially. If it is a close match, please pull the date code from the device and return the device to TI for failure analysis. To eliminate the possible effect of droop on the EN pin, I recommend simply connecting EN to VCC. Then retry the lab experiment.

In each scope shot, the following connections are shown:

Channel 1 (yellow plot) is the gate voltage as measured on the gate pin of the FET at 10V/div
Channel 2 (cyan plot) is the Vout of the circuit as measured on a test point on the circuit board at 10V/div
Channel 3 (magenta plot) is the voltage on the EN pin of the TPS2490 at 2V/div
Channel 4 (green plot) is the voltage on the timer pin (Pin 4) of the TPS2490 at 2V/div

The scope is triggered on the rising or falling edge of the gate voltage depending on the circuit condition so that it captures the image when a change occurs.

Here are the waveforms I captured when I measured the device:

The first plot, tek0555.png, shows the waveforms after the input power supply is turned on.  You can see the EN voltage rise to overcome the UVLO condition and the circuit starts up
The second plot, tek0556.png, shows the waveforms that occur when a short is applied to the output.  You can see the gate voltage drops to ~2.8V, the output drops to a few mV, the EN pin stays in the operate condition, and the Timer voltage rises to 4V.  Per the datasheet, if I read it correctly, when the Timer voltage reaches 4V, the gate voltage should drop to 0V and the circuit should latch off until either the power is cycled or the EN voltage is cycled.  This does not happen in the data I have captured.
The third plot, tek0557.png, shows the waveforms after the short is removed from the output.  The gate voltage starts to rise, the output voltage starts to rise, and then the timer voltage starts to fall, and the EN voltage stays steady in the operate mode.  The circuit has restarted after a fault that should have latched it off.
The fourth plot, tek0558.png, shows the waveforms that occur when I remove the power by turning off the input supply.  All of the voltages decay to 0V.

Attached is the schematic for the circuit I am using.  I started out using the TPS2491 to get the automatic restart, but decided 2 weeks ago that I would rather have operator intervention in the event of a short circuit and to have the circuit latch off until the fault was removed and the EN pin cycled.  So I ordered some of the TPS2490 parts which are pin compatible and replaced the TPS2491 with a TPS2490 part.  So the circuit I am using is the same as in the schematic other than the TPS2491 which is now a TPS2490.  When it didn't latch off when I applied a short to the output, I thought perhaps they had sent me some TPS2491 parts that were mislabeled as TPS2490 parts.  I ordered another lot of TPS2490 parts from a different vendor.  The new parts were labeled differently and had a different lot code.  I tried them with the same results.  They seem to act like TPS2491 parts - with the exception that even the TPS2491 gate voltage should drop to 0V until a restart after the Timer pin reaches 4V if I read the datasheet correctly.  Looking at the internal functional diagram for the TPS2490/91, it would seem that the switch that closes and connects the gate output (Pin 8) to ground through a 2 mA current source is not shutting off the gate voltage like it should.

I look forward to your insight on the problem I am seeing.TPS2491_Ckt.pdf

Dave,

We can rule out mis-marked devices (TPS2491 vs. TPS2490). We have seen this behavior type before when the EN pin voltage is very close (maybe just a bit above) to the turn on threshold and noise/droop on that pin due to VIN droop (caused by output loading) causes EN to go below (momentarily) the turn OFF threshold. From the plots CH3 seems to be quite close to 1.35V (Ven_h on pg 4 of the datasheet) and I suspect that when the short is applied EN drops below Ven_l causing the non-latching behavior.

In plot 2, CH4 (TIMER) holds at 4V but if we zoomed in on both CH3 and CH4 you would see some low level oscillation which keep the GATE from fully turning OFF.

To prove this (brute force method), connect EN directly to VCC (short out R8) and try the OUT short test. A softer method is to reduce the circuit turn on threshold (~20V) by changing R4 to a 17.4kohm). In many applications, where VIN noise may be present it is also a good idea to put a filter capacitor on the EN pin (10nF would be fine with R8=240k). You could also quickly to prove this by increasing the VIN supply to maybe 26V rather than just 24V before performing the short test.

Lastly, make sure the VIN supply will not foldback when the output is shorted (3.33A current limit). The VIN supply should be able to remain stable with > 2x the current limit or the same effect might be seen (if VIN droops, then so will the EN pin voltage). 2x can account for the fast trip threshold of the TPS2490 which is ~2x the current limit.

Good luck.