0
\$\begingroup\$

I'm trying to design an isolated LED driver using this IC. The problem is that when I turn on the driver the MOSFET heats up too quickly and it fails if it stays on for a few seconds. I have attached a moderately sized heat sink suitable for To-220 package to the MOSFET. I have tried changing the gate resistors and even shorted them because I thought the gate was not turned on fully but the result is still the same. I don't have a temperature sensor but the heat sink heats up to extremely hot to touch in just a couple of seconds (~5 seconds). Obviously the MOSFET shouldn't dissipate this much energy when I'm running a 15W LED. What could be the reason for this? enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ I can't see anything obvious wrong with your circuit, and the FET should have plenty of Gate drive. What waveforms do you get on the FET Gate/Source/Drain, and BP3319 FB (pin 2)? What are the values of the unmarked resistors (RL, RU, RS1...)? \$\endgroup\$ – Bruce Abbott Jun 17 '17 at 19:17
  • \$\begingroup\$ Those resistors are for overvoltage protection and current sense. RU and RL make a voltage divider to sense overvoltage condition while RS1-5 are low value current sense resistors. I don't think their values are important here because the circuit doesn't go in overvoltage protection mode and the current is set to an appropriate value (0.5 ohm for 300mA) \$\endgroup\$ – Rutwij M Jun 17 '17 at 19:30
  • 1
    \$\begingroup\$ Resistors values are needed to evaluate the voltage waveforms. \$\endgroup\$ – Bruce Abbott Jun 17 '17 at 19:52
  • \$\begingroup\$ The value of RL is 20K and the value of RU is 150K. Only two of the current sense resistors are populated, each of them has value of 1 Ohm. \$\endgroup\$ – Rutwij M Jun 17 '17 at 20:08
3
\$\begingroup\$

Designing switched power supplies is not easy. (Not for beginners)

Well, a power transister becomes hot out of two possible reasons:

The current that flows when it switches through multiplied with the voltage that remains between Drain and Source. You can connect an oscilloscope and measure the remaining voltage between drain and source when it is switched through and calulate the current by measuring the the voltage at the resistors RS1 - RS5. Form that you calulate the power that the transistor is consuming. If this power is 1 Watt you already feel the transistor is warm. If it exceeds 3 Watt it will become so hot that you cannot touch it anymore with your finger if there is no cooler.

The other reason for extreme heat (and that is more probable) is that the signal at the gate is not sufficient square. If the voltage rises and falls too slowly you will have a lot of loss in the transistor.

So to answer your question you must provide the oscilloscope signals measured between gate and source and between drain and source and at RS1 - RS5.

If the signal at the gate is OK I suppose that your transformator is not well designed.

Generally the information you give is very basic. What voltage is DCPOS ? What are you connecting at the output? Does the transistor also become so hot when the output is without load?

\$\endgroup\$
  • \$\begingroup\$ DCPOS is the rectified mains positive coming from the bridge rectifier. It bootstraps the IC by providing voltage through RVCC1-2 and once the IC starts driving the transformer the auxiliary winding provides required power to IC through D4 and R8. The IC is said have operating frequency between 20-100kHz and maximum duty cycle of 42%. Let's say the transformer is poorly designed, could it be the case that the IC is working outside of these limits and that affects the MOSFET? I don't have an oscilloscope at hand right now but I will try to measure it once I get my hands on one. \$\endgroup\$ – Rutwij M Jun 17 '17 at 21:13
  • \$\begingroup\$ Also neither the transformer nor the output rectifying diode gets noticeably hot in the short amount of time that I could be able to run the driver before the MOSFET gets too hot. \$\endgroup\$ – Rutwij M Jun 17 '17 at 21:23
  • \$\begingroup\$ For this type of electronic design an oscilloscope is completely indispensable. Well, the transformer stores magnetic energy while the Mosfet is swithing through and when the transistor opens, this energy is converted into electric energy at the output. If the frequency is too slow for the transformer, the transistor tries to charge more energy into the transformer, although this is already saturated. In this case the energy is lost and everything becomes hot. So how do you know that the transformer is designed correctly for this circuit and for this frequency? \$\endgroup\$ – Elmue Jun 19 '17 at 15:06
  • \$\begingroup\$ And do you use C1 to smoothen the AC power from the power line? Should't this be at least 100µF or more? And: I can't see any sense in the condensator CY2. It sems that there are several design errors in your circuit. \$\endgroup\$ – Elmue Jun 19 '17 at 17:56
  • \$\begingroup\$ And at the end: You want to drive LED's switched in serie with a high voltage and a constant current? Is this a good idea? If only one LED dies, they will all go off. Did you think that? I would take a computer power supply instead which gives me 12V and 15A and connect each 3 LED's in serie with a resistor to 12V. If one LED fails, you have only 3 LED's turned off instead of all. Like this: commons.wikimedia.org/wiki/File:LED_strip_circuit_diagram.svg \$\endgroup\$ – Elmue Jun 19 '17 at 18:07
0
\$\begingroup\$

Gate resistors are very high. And the 100k across G-S is very high, too.

That resistors ( R5 || [R3+R4] ) and the input capacitance of the MOSFET form a low-pass filter having a cutoff frequency of f=1/(2 pi 340R 1.2nF) = 390kHz. This can seem high enough, but it can be low enough to smooth the sharp edges of the gate drive signal which can cause the MOSFET to overheat due to insufficient driving.

Remove R3(10R) gate resistor, put a 0R. R4 can be between 1R and 4R7. R5 can be in 1k-10k range.

Another possible reason is the ringing across D-S caused by switched primary inductor. There is a snubber formed by D3-R6-R7-C5 but they may not be sufficient. Note that the snubbers in a Flyback should be designed carefully.

\$\endgroup\$
0
\$\begingroup\$

If i may weigh in, there are some circuits controlling massive amounts of power and hardly getting warm at all. The trick is either on or off, and not too much time in the transistion period. Some devices benifit from dampers across the device (diodes, caps,) but often its a case of adequate drive: on or off, nothing between.

\$\endgroup\$
  • \$\begingroup\$ you might want to add to your answer, especially your assessment of how fast the circuit would be switching on/off. \$\endgroup\$ – RJR Sep 7 '18 at 5:55
-1
\$\begingroup\$

From your design it is seen that you have connected gate and source with resistor R5 remove this resistor and check if the MOSFET is getting hot or not

\$\endgroup\$
  • \$\begingroup\$ I'm not sure R5 is needed but at 100K compared to 330R and 10R it's not going to have any significant effect on the operation of the MOSFET. \$\endgroup\$ – Finbarr Oct 26 '17 at 16:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.