LDO SPX3819M5-L-3-3 does not handle high Vin

Question

I am testing an SPX3819M5-L-3-3 LDO in SOT-23-5 pkg. Its max. Vin is from 2.5 V to 16 V. I need 300 mA, so as a load I connected Vout to a 10 Ω, 10 W resistor to GND.

When I input Vin more than 6 V, the output voltage Vo begins to drop. According to the data sheet spec. table I should get about 0.6 V drop at around 300 mA. According to the charts I should be around 3.3 V.

Vin    Vo                        Io
9 V    drops from 3.3 to 1.8 V    173 mA
8 V    drops from 3.3 to 2 V      200 mA
7 V    drops from 3.3 to 2.6 V    250 mA
6 V    stays at 3.3 V             311 mA

If I cool it with an air can, Vo goes back to 3.3 V. Any ideas why at higher Vin, Vo drops so much?

Answer 1

You cannot use a 1W load on 3.3V with 16V input as 300 mW equates into 4.5W with only 1W going into the load. The thermal resistance is given as θja SOT23-5 191°C/W. You can do this with a better supply and if needed a buck regulator even if it is just to drive 3.5V into the LDO if you need low ripple.

Answer 2

From the SPX3819 data sheet:

Other key features include reverse battery protection, current limit, and thermal shutdown.

Thermal shutdown depends on how you have it heat-sunk. Some package options conduct heat away from the chip better than others.
By reducing Vin, less heat is generated, because heat dissipated is proportional to: \$ (V_{in}-V_{out})\times I_{load}\$

As the chip temperature rises, a limit is reached where it begins to starve the supply of output current...which causes output voltage to slide downward. The data sheet graphs top-out at 120 degrees C, so the temperature threshold for thermal shutdown is in that region.

In addition to thermal shutdown, this chip has a negative temperature coefficient - as chip temperature rises, output voltage falls (graph from data sheet). The temperature axis is likely chip temperature rather than ambient temperature:

I suspect there's a typo in that graph: \$V_{IN}=4.3V\$, rather than \$ V_{IN}=4.3uF\$