Buck converter burns if I plug an already opened power supply... vice versa it works!

Hi,

I very strange thing happens that I hope somebody else had the same issue before.

I've designed a buck converter circuit using L7981A exactly as it is described in its datasheet. I normally works fine.

If I first plug the connector, then open the power supply, it works perfectly fine.
However, if I first open the power supply, then plug the connector, it burns (i.e. the input terminal of the buck IC becomes short).

Circuit draws around 100mA. My schematic is as follows:



Thanks in advance.

A little clarification, please: When you say "plug the connector" I assume you are connecting the 3.3 and 5 volt outputs to the load? And what do you mean "the input terminal...becomes short"? What do you mean "it burns"? Does the device actually catch fire? Does the device fail?

The big question is: what is your load? Is it highly capacitive?

barry said:

A little clarification, please: When you say "plug the connector" I assume you are connecting the 3.3 and 5 volt outputs to the load? And what do you mean "the input terminal...becomes short"? What do you mean "it burns"? Does the device actually catch fire? Does the device fail?

The big question is: what is your load? Is it highly capacitive?

Click to expand...

Okay.. What I mean by plugging the connector is plugging the input connector PJ-102AH (a very common adaptor plug).

Device does not catch fire because I've limited the power supply current to 300mA, so it basically hits the limit and cuts the voltage, but I'm sure I would see some smoke if I did not limit the current. Input terminal becoming short is actually same as failing or burning.

My load is 100mA as I mentioned (this is what I see in my power supply screen), but I'm not sure about the capacitance of the load but it must be few hundred microfarads, because there are bypass capacitors all around the circuit. This is just the power part of my design.

I would like to note that I'm feeding the circuit with 19V.

When you say "open the power supply" do you mean 'turn the power switch on'?

When you say "it burns" what you really mean is 'it draws excessive input current'. Is that right?

My guess is that it has to do with your large capacitive load. When you plug the input connector, you're getting some 'bounce' (your switch will do that, too) and this bounce is causing a number of large current spikes from your power supply. It would be helpful if you could actually monitor your current input with an oscilloscope.

The problem is solved when I added a 100uF aluminum electrolytic capacitor to the input. But I don't know the science behind it

When you say "open the power supply" do you mean 'turn the power switch on'?

Click to expand...

Yes, turning the power supply on.

When you say "it burns" what you really mean is 'it draws excessive input current'. Is that right?

Click to expand...

No, I mean it really burns and fails, consequenty it draws excessive input current.

Well, I still think your problem has to do with your capacitive load. But doesn't the device have current limiting? Can you try this with no load attached to see what happens?

Check your cct values.
Is inductor sized for peak current?
Is FB pin accidentally grounded?
Is frequency setting res correct, if give too low a frequency than that could be bad like you say.

treez said:

Check your cct values.
Is inductor sized for peak current?
Is FB pin accidentally grounded?
Is frequency setting res correct, if give too low a frequency than that could be bad like you say.

Click to expand...

Thanks for your interest.

I did my design based on the example in the datasheet, did not do deep calculations, but I'm sure the FB pin is not accidentally grounded.



Knowing the problem occurs due to the in-rush current leading overshoot and consequently passing the buck's internal MOSFET's avalance breakpoint; I've decided to use TVS diodes both before and after the buck converter.

However, I'm not sure how to decide which kind of TVS is suitable for such purpose, while there are "TVS diodes", "TVS thyristors", "TVS varistors" etc.

There are also many parameters like "Voltage - Reverse Standoff", "Voltage - Breakdown", "Voltage - Clamping", "Current - Peak Pulse (10/1000µs)" etc.

Any suggestions?

ogulcan123 said:

Thanks for your interest.

I did my design based on the example in the datasheet, did not do deep calculations, but I'm sure the FB pin is not accidentally grounded.

Knowing the problem occurs due to the in-rush current leading overshoot and consequently passing the buck's internal MOSFET's avalance breakpoint; I've decided to use TVS diodes both before and after the buck converter.

However, I'm not sure how to decide which kind of TVS is suitable for such purpose, while there are "TVS diodes", "TVS thyristors", "TVS varistors" etc.

There are also many parameters like "Voltage - Reverse Standoff", "Voltage - Breakdown", "Voltage - Clamping", "Current - Peak Pulse (10/1000µs)" etc.

Any suggestions?

Click to expand...

Using TVS on the input will accomplish nothing other than giving the TVS vendor some money. If your problem is excessive inrush current the TVS is not going to limit it. Are you sure your problem is due to a voltage overshoot and not the excessive current? Did you actually observe an overshoot?

Also, are you using the soft-start feature? That might help a lot.

barry said:

Using TVS on the input will accomplish nothing other than giving the TVS vendor some money. If your problem is excessive inrush current the TVS is not going to limit it. Are you sure your problem is due to a voltage overshoot and not the excessive current? Did you actually observe an overshoot?

Also, are you using the soft-start feature? That might help a lot.

Click to expand...

I actually did not observe an overshoot at the input and confirmed the soft-start feature works perfectly throught the oscilloscope. However, I'm informed by the vendor's field applicaiton engineer that it might be caused due to the overshoot arised from the inrush current; but in fact, according to him, this inrush current is happening at the output due to the output capacitance. However, this seemed very awkward to me as the transient effect is happening at the input side, where there is no chance of the current passing to output without going through the buck converter.

So, why am I also looking for a TVS at the input? To make it more failsafe, as we will upgrade the design to be compatible with the automotive industry, where we will use AEC-Q100 qualified chip.

I would like to note that I am disappointed that such self sustained failsafe chips are burning although we used the recommended design with the materials even with better values.

You're putting in TVS to protect against an overshoot that you've already established doesn't exist? I REALLY think you need to understand z what the real problem is rather than just throwing arbitrary components into you're design. If you don't really understand the problem you are destined to have it come back to haunt you.