LMG3422R050: LMG3422R050 Very strong heating

Ruslan Kovtun

Part Number: LMG3422R050

Tool/software:

Hello, I have a problem with very strong heating of transistors when supplying PWM with a frequency of 2MHz, although the current consumption approximately corresponds to the graph:

Heating occurs when the PWM signal and power are applied to the transistors (VDD - GND), even without the main high-voltage power supply (Drain - Source).

The scheme looks like this:

Power Schematic:

When supplying power (5V), the consumption current is 60mA. As soon as I apply the PWM signal, the current increases to 615mA, that is, 615mA/2 = 307mA per 1 transistor. But this current for 5 V, for 12 V ~ 5/12 * 307 = 128mA, it seems that everything is fine with consumption, but there is strong heating.

Is this behavior normal and do we need a large radiator for these transistors?

over 1 year ago

0 Madhur Wagle over 1 year ago

TI__Intellectual 1220 points

Hi Ruslan,

Few questions -

1. Is the heat equally distributed between both LMG3422 devices or is there any hotspot on top device/inductor? How much is the temperature rise from ambient?
2. Please check if the ISO_H_12V or ISO_L_12V net voltage dips after you apply the PWM.
3. Cross check the deadtime between HI/LI pulses exactly at IN pins.

Best Regards,
Madhur

0 Ruslan Kovtun over 1 year ago in reply to Madhur Wagle

Prodigy 30 points

1) When the output voltage (Drain - Source) is turned off, only the transistors heat up, here is a screenshot of the TEMP signal:

From these screenshots, it can be seen that 2 transistors heat up equally to ~85 degrees. As soon as I apply the main voltage (20 volts 2 poles) the temperature quickly rises to more than 150 degrees until the temperature error

At the same time, the inductor and snubber are also heated.

2) Supply voltages ISO_H_12V and ISO_L_12V are fine, here are the screenshots:

ISO_H_12V:

ISO_L_12V:

3) Here is the input signal:

After the isolator (IN):

SW output signal (+-3.0 V on inputs):

Maximum current on power line less than 1A

0 John Gomez over 1 year ago in reply to Ruslan Kovtun

TI__Expert 6625 points

Hi Ruslan,

We are looking into it, Madhur will get back to you within the next business day.

Thanks,

John

0 Madhur Wagle over 1 year ago in reply to John Gomez

TI__Intellectual 1220 points

Hi Ruslan,

We are trying to mimic this in our lab. I will let you know once we have tested it.

One more question- Did you observer FET/ Inductor current waveforms when Vin was applied?

Best Regards,
Madhur

0 Madhur Wagle over 1 year ago in reply to Madhur Wagle

TI__Intellectual 1220 points

Hi Ruslan,

We ran the same test on our daughter card and observed 45deg of temperature on the device.
Can you share the design files to m-wagle@ti.com for a review?

Best Regards,
Madhur

0 Zachary Soviero over 1 year ago in reply to Madhur Wagle

TI__Expert 4520 points

Hello,

As Madhur mentioned, I have run a test in our lab to recreate this issue you are seeing. Here are my results:

From these results, the TEMP pin is reporting 40.4 deg C and I am measuring around 45C. Given the margin of error for both TEMP pin and thermal camera, these results are expected and well matched.

Thanks,
Zach

0 Ruslan Kovtun over 1 year ago in reply to Zachary Soviero

Prodigy 30 points

Hello,

Sorry for the long answer, thanks for your tests. When we supply +-20 Volts at the input, the circuit consumes 400mA without connected load. I added a copper plate (10mm) to the back of our board but that solved the problem without supplying the Vin. When we supply Vin +-20V the transistors still overheat.