Popping components, oh no!

[ITman496]

So I'm working on a little battery management board for a chunky battery for a drone I'm making.  It has a soft latching switch so the system doesn't stay on all the time, only when plugged into a data cable or the button is pressed.  It works by powering the micro up and then the micro grabs onto the pin and holds it high until it decides to cut power.  Here is the section of my schematic.



And this is what the actual board looks like.  It's my first smd board I've ever made and I think it came out okay.



So when I tested it at 6.5V, the minimum to boot the dc/dc converter up, all was well.  But when I ramped it up to 25.2V, the max the 6 cell lithium pack should ever make, it was good until I hit the power button.  POP.  The p-fet is now shorted.  The rest of the circuit still works.  I was freaked out by a smudge on the micro looking like it had popped but as it turns out it was just flux residue.

The fet has a 30V rating.  A little low, and I'll replace it with a 60V rating one if this design is okay.  But the rating was not exceeded.

Is there some critical design error here?  I've been following most of the schematics for soft latching systems I've found on google and they seemed to match what I have here.

The fet is this:

http://www.newark.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=15003&langId=-1&urlRequestType=Base&partNumber=63J7613&storeId=10194

the transistor under it is this:

http://www.newark.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=15003&langId=-1&urlRequestType=Base&partNumber=75R0831&storeId=10194

Now, I think the fet is shorted because I measured the resistance between source and drain and got zip.  That's how it goes for p-fets, right?  I'll admit this is the first project I've ever done with mosfets.

I just hope that this is something dumb that I can bodge back to working condition and not have to get another pcb made..

[Neverther]

First page on datasheet

Absolute Maximum Ratings:
Vgs Gate-to-Source voltage +-20v

This?

[theatrus]

You are dropping the gate voltage to near-zero, yet the part has an absolute maximum Vgs of 20V. This is being exceeded, which is likely destroying the FET

[c4757p]

Yup. You are destroying the very thin oxide (O in MOSFET!) layer between gate and channel, turning the whole thing into a big shorting block.

[mij59]

Exceeding the max. gate source voltage is a really bad idea, put a zener diode e.g. 15 V across the gate and source, add a resistor in the collector of Q3.

In general inputs are drawn on the left of the schematic, so put the cell6 on the left of the schematic, and the voltage regulator on the right. 

[ITman496]

Whoops!  See, like I said, I've never used fets before.  I saw 30V max something on this and figured it was good.  Heh.

As for the backwards inputs, I forget why I did that, I think it was because I couldn't get it to have a nice layout without mirroring the part.  I'll fix it for future reference.

Does anyone know of a nice sot23 p-fet that has a 30V gate to source voltage? Even the much higher rated ones seem to only like 20..  Is that a common max?

Thanks again! I knew it was some bone headed decision I made.  I just want to see if I can swap a component out and not have to change the pcb.

[ITman496]

If I cant find a fet with a higher max VGS (which I can't seem to)

This is what I want to do, correct?  I should be able to bodge this on.

[mij59]

If I cant find a fet with a higher max VGS (which I can't seem to)

This is what I want to do, correct?  I should be able to bodge this on.

Yes looks much better, should work.

[T3sl4co1l]

He's got it the right way around, just drawn backwards (right to left).

Transistor ratings you generally want 1.5 to 2 times the peak voltage seen in the circuit.  For a 26V battery, you want 40V or higher (Vds, and Vgs(max) protected somehow: the resistor and zener method is quite typical for this).  It also wouldn't hurt to check how much energy must be delivered to charge the bypass capacitors in the circuit.  You don't indicate any capacitors around the regulator, which seems unlikely.

Tim

[ITman496]

It works!  Thanks everyone!

I left out the zener diode (because I didn't have one on hand) and instead put a 100k resistor after the gate to divide the voltage in half.  It's on a battery anyway.  There will be very little in the way of spiking and the battery would have to hit 40V for the Vgs to be 20v.  You have bigger issues when that occurs.  (rating of DC/DC supply exceeded, rating of pretty much everything in the drone exceeded, rating of caps and fet exceeded)

I have seen similar circuits like this on the internet, so I assume its also acceptable.  Is there a reason for the zener method vs other ones?  Or is the zener clamping method for situations where there is significant noise/ripple?

I had issues when I first put this back together actually, where it kept turning on! And then at the same time, wouldn't turn on.  Somehow when the p-fet oxide layer blew up, it must have connected gate to source.  (Which I suppose is a thing) and so my little diode was blown to bits, and the npn transistor was also shorted.  So it kept turning it on, and not responding to the switch.  Now that I've replaced pretty much this entire section, all is well.  Hooked up to the scope and power supply set to 28V (way higher then the battery should be capable of) the gate of the fet only sees 14V and there are no captured spikes or dips in that higher or lower.  Everything else seems okay.  And now a short is impossible as the gate is isolated on each side by 100k resistors.

This is the new schematic, and the bodge. Works great!

[mij59]

A zener diode ( 10..15V) is best way to protect the mosfet, e.g. against esd.
Also the gate source voltage is constant over a wider input voltage range.

[ITman496]

I hadn't thought about that actually..  I'll work a zener into the circuit and all future circuits involving this kind of fet use in the future. Thank you!  This has been an educational failure.

[jlmoon]

I hadn't thought about that actually..  I'll work a zener into the circuit and all future circuits involving this kind of fet use in the future. Thank you!  This has been an educational failure.

Actually, I think you might have become more "educated" using those FETS!   

[langwadt]

I hadn't thought about that actually..  I'll work a zener into the circuit and all future circuits involving this kind of fet use in the future. Thank you!  This has been an educational failure.

I'd consider using one of the numerous $1 mp1584 buck converter boards available on ebay etc.

much cheaper than that tracopower module and has an enable pin

[ITman496]

Good point, I'll definitely consider it for a future design.  It does seem to be slightly annoying to find for usa though after quickly glancing around?  But I'll definitely look for those little modules when my little ribbon of traco supplies runs out because I honestly totally forgot about them..