Controlling high power LED with pnp transistor.

[Pariston]

Hi,
Recently I've designed a circuit where current through High Power LED is controlled with pnp transistor. I know there are more efficient ways of controlling current, but unfortunately I was not able to use any of these here. After one day pnp transistor died. This transistor was rated up to 0.5W (150C) while I used 0.25W(77C Max). Any suggestions why this could happen? I will attach circuit diagram and used components.

PNP - https://www.digikey.ca/en/products/detail/FMMT596TA/FMMT596CT-ND/92668?itemSeq=341787914
NMOS - https://www.digikey.ca/en/products/detail/BSS123/BSS123NCT-ND/244665?itemSeq=346061027

[james_s]

How did you calculate the dissipation of the transistor? Keep in mind that the maximum temperature is the temperature of the die, which will normally be considerably higher than the case temperature. This is also the *maximum* temperature, not a target you want to operate at. I like to keep components under about 60C, if you can't hold your finger on it for a second or two then it's too hot.

[Pariston]

I measured temperature directly from one of transistor's legs. Maximum observed voltage drop on this transistor was around 1.75 volt at 145 mA (0.254W)

[gcewing]

77°C is pretty darn hot! I'd be putting a heatsink on it.

Also, have you actually measured the current, or are you just going by the rating of the light? Is there anything in the light to limit the current? If not, it could be passing more current than you think.

[Pariston]

Yes, i measured current directly on the led line. Current is limited by potentiometer on the gate side of pnp transistor. Current on the line was set to 145 mA, which resulted in around 1.75 volt drop across pnp

[Ian.M]

I'd bet on thermal runaway.  See the h_FE vs I_C graph in the datasheet: https://www.diodes.com/assets/Datasheets/FMMT596.pdf
With I_C in the range 100mA to 200mA, as the junction temperature increases from 25 deg C to 100 deg C, h_FE will increase by approximately a third, so I_C will increase proportionately.  Meanwhile as the LED array warms up its Vf will be decreasing, probably by about 0.3V for every ten deg C, increasing the voltage across the transistor.  It wont be long before its dissipating twice as much power, which would put it right on the bleeding edge  of failure.

[Pariston]

This could be it. I will perform some tests tomorrow to confirm.

Thanks

[aheid]

Note that the observed voltage drop and current is right at the edge of the Safe Operating Area for DC operation. Also note the SOA graphs are not always that accurate and so you probably should have more margin than that.

[golden_labels]

Pariston
Aside from what everyone else has already said, you are not controlling current. You are sinking constant current from the base of the PNP transistor and, I assume, you expect that the collector current will be approximately hFE · I_B. But the problem is that hFE varies wildly, by more than an order of magnitude. See the table on page 2. And not only between specimens. I see that there is a potentiometer to adjust that, but even if you do, the junction temperature and collector current affect hFE. Even worse, there is a positive feedback loop between temparature and the aplification you get, which is why you get thermal runaway mentioned by others.

You might as well use a 12Ω, 0.5W resistor there to control the current and a transistor as a switch.

[Doctorandus_P]

As said before, Hfe of a BJT is an unreliable parameter, and the trick is to use it in a way that it regulates itself.

In the attached circuit a single BJT is used as a current sink to drive the LED (string, presumably).

How it works:
I assume "LED4" is a 5V uC output, and when it is high, the BJT will be cut of if it's Emitter rises above around 4.3V. and this will happen at a current through R3 of around 4.3/32 = 134mA.
Most of that current will go through the LED.

R2 limits the Base current and also protects the uC in case the LED string is disconnected or the +48V power supply is absent.