LM3492: PWM switch on / activation damages the LED Driver

Hello Marek,

Did you measure the output of the function generator? In your example 3V output would be into a 50 ohm load, if there is no 50 ohm load the output could be 6V if that's how your function generator works. Have you measured the output of the function generator when running your experiment? You could also add a series resistor from the function generator output to the PWMs such as 50K ohm or 20K ohm depending on what voltage you are driving the pin at. You can also add a zener clamp to prevent the voltage from nearing the absmax of 6V on the PWM pin. This would be after the series resistor. Also make sure that the grounds can be tied together without current paths. Is your power source isolated? Many function generators are not isolated outputs.

Regards,

Hello Irwin

Thank you for your answer.

Finally I measured a voltage of 0V and 3V at the output clamps of the function generator and took a 1k resistor.

First I set a frequency of 250mHz and 50% Duty Squarewave to see if the LEDs are switching on.
Result: Only one chain of 7 LEDs were on, the other 7 were  after a very short flashing time off. Again to control if one chain is damaged I connected the PWM_IN over Rcomm to VCC, switched on the circuit an the LEDs were on, both chains.

Then I tried a frequency of 1kHz and 50% Duty.

Result: Everything worked, although with a noise, but it worked. I tried to set Duties from 20% to 80%, the LEDs all dimmed fine. When I switched off the output of the function generator, LEDs turnded off, and switching on the output of the FG LEDs turned on, worked also.

To see if I can dim down to 10% Duty I had to switch over to Pulse function on the same channel. I set correctly the output voltage, Duty, frequency. Switched on the circuit and then the output of the FG.
Result: LM3492 damaged (or at least I think so). Input Voltage is around 5.5 V and the circuit draws the full current limit of 450mA, no humming.

After a few minutes switched on again with FG off to control the Input Voltage again. It was higher and I saw that at the coil was a small Spark with a humming in the switching frequency of the device,  (I think the spark is because not proper soldering ...)

The longer I ran the circuit, the lower was the humming (with a spark at the coil) and the higer the Input Voltage. Now the Input Voltage is 24V and the circuit draws 7mA, which comes from the polarity protection circuitry. And the coil or Pads are burnt.

I guess when before the coil was damaged the circuit had a short, and now the coil (or a Pad) is damaged and there is no short anymore, then the short came over one of the SW1,SW2,RT,VOUT or FB Pins?

I don't understand.

Shall I take another device?

Hello Marek,

I'm not sure what's going on in your case. Sometimes you have to be careful with function generators when switching waveforms, depending on the function generator, I've seen some have large voltage excursions when changing output function. Seems like this should just work. Not sure if you have opto-couplers but maybe isolation on the PWM in your test set-up might be an issue?

Sounds like something is getting stressed and then it fails, it shouldn't do that.

Regards,

Hello Irwin

I watched the changing of the output function on our oscilloscope and haven't seen anything peaking out.

Also I damaged the device once when I put the PWM on Rcomm manually (which is the default of the design report) ... maybe one shouldn't do this? I don't know... just to tell that you don't need a FG;)

My colleague thinks to take a nMOS instead of an opto-coupler, because it's cheaper. What do you think, try?

Also I have seen that I put the Feedback Resistors too far away from the LM.

Greeting

Hello Marek,

An opto-coupler will isolate the signal nMOS won't. When you have a bunch of test equipment connected the ground loops can cause issues. I have had issues with function generators connected to non-isolated circuits. If the signals are in the correct range it should have these issues.

Is this a PCB or is this a test circuit?

Regards,

Hello Irwin

It is a PCB.

I will design it new in the next days. I watched the TI Videos about switching converters, so I hope I can improve some things.

Maybe also put some additional Pads to measure or change and add parts.

Greeting, Marek

addendum:
The PCB is only Top layered.
Or use a Gate Driver for the dimming Input, such as LM5112.

addendum:
The PCB is only Top layered.
Or use a Gate Driver for the dimming Input, such as LM5112.

Hello Marek,

It should just work without that. If it is a transient issue an opto-isolator would be a better solution. In reality there should be no issue but with test equipment and multiple grounds it could be. You could also add something that generates the PWM on the board to see it that works.

I'm going to close this thread, open a new one if you have other questions or issues.

Regards,