LP2951-33 - what can cause output voltage to be 2.5V?

"... The LDO has SENSE and OUTPUT connected together, as well as VTAP and FEEDBACK are connected together..."

That sounds correct.

Q1: What is voltage at VTAP/FEEDBACK connection?

Q2: What is the output capacitor value/type?

Voltage is 0.931V

Output capacitor is 10 uF, 25V aluminum electrolytic.

Voltage at VTAP/FEEDBACK should be ~1.2V. However, the measured 0.931V vs the expected 1.2V is roughly proportional to the Vout error (2.5V vs 3.3V).

With 5V input voltage, 3.3V output, and estimated 85mA load current the dissipation will be ((5.0-3.3) x 0.085)= 0.145W.

The 'D' package has nominal RθJA rating of 97°C/W. So, estimated device junction temperature rise above ambient will be (0.145 x 97) = 14°C.

It's seems unlikely that this is a thermal problem, and your observation "... the chip does not feel hot when I touch it..." is a good enough confirmation.

I suspect that the output might be oscillating due to low phase margin. While the 10uF capacitance value should certainly be adequate, the ESR of the 10uF aluminum electrolytic might be too high.

Try adding Cbypass to see if it helps. See section "8.2.2.3 Cbypass: Noise and Stability Improvement" in the LP2951 datasheet.

Other alternative is to use quality MLCC (ceramic) capacitor (X5R, X7R) for Cout with a 0.1 Ω to 2 Ω series resistor added to simulate the required ESR. See section "8.2.2.2 Capacitor Types" in the LP2591 datasheet.

I tried a 0.1 uF ceramic cap from pin 1 - pin 7.

Did not seem to help, voltage is still 2.5

I will scope the pins and provide the waveforms, to see if that gives a clue to what is happening.

It may also be worthwhile putting a MLCC cap in parallel with the 10 uF electrolytic on the output.

I replaced the 10 uF aluminum electrolytic with a 10 uF MLCC.

The output voltage is still 2.5V.

I observed the output voltage and feedback voltage with an oscilloscope.

The voltage looked smooth, and I could not see any oscillations.

Any other suggestions I should try?

Running out of options.

I see two scenarios left: 1) some subtle connection issue, which seems unlikely; 2) the measured 0.931V at the Vtap/Feedback pins might indicate that the LP2951-33D is damaged somehow.

If you can provide images of your schematic, and PCB layers, it might help. Additionally, please confirm the device marking is : KY5133

Here are images of the relevant sections of the schematic and layout.

I saved a screen shot t a PDF, and attached; I hope this works for you to view it.

On the layout, I did not show the ground plane layer for clarify. The vias from C38 and C39 connect to the ground plane.

The blue plane shown is VCC.

Here is the marking on the device:

496L

2951C

M33C

There is also a logo that appears to be a National semiconductor logo on the part.

Schematic and Board for LP2951 regulator.pdf

"... There is also a logo that appears to be a National Semiconductor logo on the part ..."

The marking information you provided would be appropriate for the National Semiconductor LP2951CM-3.3/NOPB.

Direct equivalent to LP2951-33D.

 

Make sure you measure to confirm that the vias from C38 and C39 (and LP2951 Gnd) really do connect to the ground plane on the PCB.

I don't see anything out of the ordinary in the schematic, or in the layout.

I'm out of ideas at this point.

Down to : damaged device.