# Transistor heating up problem

I have an LM317AHVT power supply with PNP pass transistor.

I noticed that the power transistor is heating up to about 64C after a few minutes. I am using an 130x80x30mm heatsink and an 92mm fan to cool the heatsink. The temperature of the heatsink is about 46C. I mounted the transistor using mica insulator and thermal paste.

I also made another test, using a bigger heatsink (170x80x35mm), and the temperature of the TIP36C transistor was the same (about 64C) even if I used a bigger heatsink. The temperature of the heatsink was about 40C. I also used the same fan to cool the heatsink and the transistor was mounted using mica insulator and thermal paste. In both cases, the transistor was mounted using an M3 screw. The fan is blowing air like in this image: This is the schematic of my power supply: The input voltage is about 39Vdc (measured on the filter capacitor), the output voltage is about 0.6V (the power supply works in constant current mode, because I want to test the power supply in the worst conditions) and the current throungh transistor is 1A. The current through LM317 is about 0.7A.

1. Is this temperature normal or it should be less?
2. When I used a bigger heatsink, why the temperature remained the same?
3. It is OK to use the transistor without mica insulator and only with thermal paste?
4. Will the transistor be damaged if I will use it at 64°C?
• You added a heatsink and a fan. Probably would have been easier to have replaced the regulator with an SMPS module. What is the input voltage, output voltage and output current? – Oldfart Feb 21 '18 at 15:30
• The input voltage is about 39Vdc(measured on the filter capacitor), the output voltage is about 0.6V (the power supply works in constant current mode, because I want to test the power supply in the worst conditions) and the current throungh transistor is 1A. The current through LM317 is about 0.7A. – beard999 Feb 21 '18 at 15:34
• You've asked 8 questions and I see some have produced good answers. Are you aware that if you get an answer that helps you you can upvote it and, if you get an answer that explains things to your satisfaction you can mark it as "formally accepted"? – Andy aka Feb 21 '18 at 15:50
• @Andyaka I upvoted now some of the answers I got. Thanks for your advice. – beard999 Feb 21 '18 at 16:06
• And there is the formally accepting of answers. If you cannot formally accept an answer it is better to raise a comment for deeper clarification rather than leave it. Then when you get the answer you understand you can formally accept it. – Andy aka Feb 21 '18 at 16:10

You transistor is dissipating nearly 40W of power. So in answer to question 1 - yes that temp is probably "normal".

Question 2 - The temp is the same because only so much heat can effectively pass from the transistor into the heatsink with your configuration.

Question 3 - Yes you can operate without the mica insulator - BUT, you must leave the heatsink ungrounded and it will have the power supply's output voltage on it. I would not advise doing that.

If you want to reduce the operating temps in your "worst case" operating condition, I would suggest you use two TIP36 transistors in parallel, giving you twice the thermal contact to the heatsink, doubling your effective heat dissipation..

• Thanks for the reply. @Norm If I will use the power supply (with the mentioned transistor temperature of 64C) the transistor will be damaged ? – beard999 Feb 21 '18 at 16:18
• Its rated operating temp range is -60C to 150C. Your temp of 64C is well within that range. It should not hurt it to operate at that temp. – Norm Feb 21 '18 at 18:05
• I forgot to mention @Norm that the temperature was measured at the surface of the transistor case were is written TIP36C. The temperature was measured using an thermocouple. Your answer is still valid in this case ? – beard999 Feb 21 '18 at 19:12
• If you can contact the metal tab that would be a better measurement point, but the front side of the case should be just as good if the tab is not accessible. According to the data sheet at 64C junction temp, you're still well within the power dissipation envelope. – Norm Feb 21 '18 at 20:08
• Thanks for the reply @Norm . If the case temperature measured on the spot were is written TIP36C is 64C, how much is the junction temperature ? – beard999 Feb 22 '18 at 12:50

It is probably normal for your inefficient linear supply.

Temp rise = Rth × (Vin - Vout) × Iout

e.g. if heat sink is 2°C/W and Vin-Vout is (39V-1V) @ 1A that results in 38W × 2°C/W = 76°C rise.

Using a pulsed inductor on the input will store energy to reduce conduction losses. Or you can improve your heatsink thermal resistance ( e.g. CPU cooler)

Search for PWM buck reg & LM317 solutions.

• Thanks for the reply @Tony Stewart. EE since '75 If I will use the power supply (with the mentioned transistor temperature of 64C) the transistor will be damaged ? – beard999 Feb 21 '18 at 16:02
• 85'C under all conditions max. but will survive 125'C, better if 70'C max. your 76'C rise + 25'C room=101'C . If you put in a light bulb in series on input to dump heat, that would be ok. 2x 12V@1A = two 12W 12V bulbs or one if using higher voltage or none if using max Vout – Tony Stewart Sunnyskyguy EE75 Feb 21 '18 at 16:25
• Thanks for the reply @Tony Stewart. EE since 75 I measured using a termocouple, which indicates abou 23C room temperature. And the temperature of 64C was already with the room temperature in it. – beard999 Feb 21 '18 at 16:31

You've made a mistake somewhere because Q1 shouldn't turn on at all in that circuit you've drawn. $V_{BE}$ will always be zero.

This is the more conventional way to use a PNP pass transistor with an LM317:

• Thanks for the reply. @Finbarr If I will use the power supply (with the mentioned transistor temperature of 64C) the transistor will be damaged ? – beard999 Feb 21 '18 at 16:19
• No simple answer to that. The datasheet tells you that the maximum operating junction temperature is 150°C, and that the case has a maximum thermal resistance of 1°C/W. So if it's dissipating 40W, the junction will be 40°C higher than the case, i.e. 124°C and be within the maximum. But as your circuit is clearly wrong it's difficult to say what the actual dissipation is. – Finbarr Feb 21 '18 at 16:33