Magic smoke from voltage regulator module (TPS7A4701)

[MiDi]

This module is not a cheap one, it is - sorry - was a low noise 1A linear regulator with input voltage up to 36V.

This poor guy failed during turning on with a cracking noise, a light glow and finally magic smoke escaped everyone likes to smell 

I have an idea what happened, but first I would like to know what you think 

[Zero999]

Well the inductor has a chunk missing out of it which is probably the cause of the failure, rather than effect. With less core material, the inductance will be lower and it might also saturate at a lower current and if the controller isn't designed for less inductance, then pop.

[capt bullshot]

I'd guess input overvoltage transient during turn-on.

[Hydron]

I assume that the inductor is for filtering only (thus not the cause) given you said it's a linear module.

I agree that input transient is a top possibility (though the tantalum in parallel would normally help), but the other thing I can see is the jumper pads along the side - should one of these be bridged for correct operation? Wouldn't expect a broken feedback loop to kill a linear reg but worth asking.

[Zero999]

I assume that the inductor is for filtering only (thus not the cause) given you said it's a linear module.

Oh yes, I missed that.

The board does look more like a switched mode power supply though. Not only does it have a fair sized inductor, but it lacks a heat sink and device(s) with a thermal tab/pad, which would be required for a linear regulator, especially one rated to 36V input at 1A.

[Zero999]

I've just Googled the data sheet and it's definitely linear. The thermal pad is underneath the IC, so there should be thermal vias on the PCB and if a heat sink is required, it should be on the other side. The inductor will definitely just be for filtering and the chip will not have caused the failure.
http://www.ti.com/lit/ds/symlink/tps7a4701-ep.pdf

[ogden]

This module is not a cheap one, it is - sorry - was a low noise 1A linear regulator with input voltage up to 36V.

This poor guy failed during turning on with a cracking noise, a light glow and finally magic smoke escaped everyone likes to smell 

I have an idea what happened, but first I would like to know what you think 

As you can supply 36V and ask that poor thing to supply 1A at minimum voltage 1.4V, I think this is what happened.

[glarsson]

Forgot to turn on the liquid nitrogen cooling system?

[MiDi]

Supply voltage was 30V, no load applied...

[SeanB]

Well, weasel specmanship says 36V input range, but Absolute Maximum voltage is 36V as well, so I would guess turn on transient of whatever power supply you used, or lead inductance to the module, got it above 36V by a bit, probably, because of the 30V supply, to around 45V. then the chip inside broke down and went into thermal runaway, letting out the smoke. Prevention for the next would probably involve some series resistance in the supply rail and a nice fast transient suppressor diode across the power rail to tame this.

[Zero999]

This module is not a cheap one, it is - sorry - was a low noise 1A linear regulator with input voltage up to 36V.

This poor guy failed during turning on with a cracking noise, a light glow and finally magic smoke escaped everyone likes to smell 

I have an idea what happened, but first I would like to know what you think 

As you can supply 36V and ask that poor thing to supply 1A at minimum voltage 1.4V, I think this is what happened.

Linear regulators are normally protected against this kind of abuse. They have safe operating area protection, which reduces the current limit, when the input-output differential exceeds a certain point and thermal protection which shuts off the output, when the die exceeds a certain temperature. I haven't read the TPS7A4701 data sheet in great detail, but the bullet points on the first page indicate it has all of the aforementioned protections.

Also note that the original poster hasn't said anything about what  the input, output voltages and current were. I agree, running it at its absolute maximum ratings is asking for trouble, but it's a bit premature to jump to that conclusion.

[edpalmer42]

I've just Googled the data sheet and it's definitely linear. The thermal pad is underneath the IC, so there should be thermal vias on the PCB and if a heat sink is required, it should be on the other side. The inductor will definitely just be for filtering and the chip will not have caused the failure.
http://www.ti.com/lit/ds/symlink/tps7a4701-ep.pdf

The Philips head bolt in the middle of the picture is used to attach the heatsink.  Here are some pictures of a very, very similar module.

https://www.ebay.com/itm/322870941945

No connection, I chose this one at random.  Note the thermal image.  I'm not sure I believe those numbers.  Thermal resistance of 6 °C/W with that tiny heatsink?  Maybe with 1000 CFM of air at -100 C.   

Ed

[Zero999]

I'm going to go with reverse polarity killing it, as it's not specified to survive below -0.4V. This is an all too common user error. One which I've done myself too many times, even though I know better. 

[MiDi]

There is a 2.2uH inductor (~60mOhm ESR) and a 40uF cap (~10mOhm ESR) at the input - a nice resonant circuit with quite low damping factor...
The wiring from power supply adds inductance too, not sure how relevant this is 

I guess the input of the ic went to more than 40V peak when connecting the 30V power supply - as SeanB suggested.

Now let's try to measure

[maginnovision]

Alot of the tps parts have very close max and abs max. Gotta be careful. I add tvs on every supply I use with their parts.

[MiDi]

I killed it another time with desoldered ic 
Just after applying the power there was a flash on the pcb and the rails are now shorted.

Pictures and results are attached - for the sake of completeness.

Hm, the main damage seems to be occured between Vin IC and EN.
Makes sense as the EN pin is rated at max 0.4V above Vin IC and the inductor is located between this Pins, when shorted to Vin.

[MiDi]

After replacing shorted cap, a new measuring attempt was taken.
First shot got a bit of contact bouncing, so it is not shown here.
The cap went down from initially 47uF to 40uF after first treatment, after second it remained at 40uF.

The results are not as expected: Vin IC peak is below max. spec of 36V and both measurements are close to each other with a bit over 34Vpeak.
At 30V with this setup no problem, but its close to max specs, so higher Vin or more lead inductance would come up to over rated spec.

What remains as a cause is the Enable Pin which is driven far above limited specs of Vin IC + 0.4V, on startup it gets 30V above Vin IC!
En is equal to Vin as in this setup they are shorted and therefore chances are high to break it independent of the applied voltage 

As this device is ESD rated, it could be enough to place 10k resistor instead of short between En and Vin (ESD diode at En clamps to allowed voltage and the current is limited to well under 10mA).
There is a 100k resistor between En and GND Vin IC, have to check the value.
Hopefully it has a high resistance value and works together with 10k...
So no need to connect En Pin to anything for normal operation and dangerous to do so - my mistake.
Edit 16.9.: corrected wrong statements

Not sure if the inductance does help more or does more harm...
There are some other issues with the design of this module, so it leaves a tasteless aftertaste - as we say in german 
But it was a good practice where to look for the traps in designing in this kind of linear regulators 

[maginnovision]

Try removing the inductor and shorting the pads to see if it still goes? Or cut the en trace and solder a wire from inductor out to en. Best to find what it takes to make it work than waste time just producing another revision, I'd think.

[jackenhack]

 I'm using that IC on my headphone amplifier and it works great. Don't understand why they need an inductor? I have had an eBay bought negative version TPS7A33 go short circuit due to insufficient solder on the heat pad. Otherwise the IC is bulletproof.


Sent from my iPad using Tapatalk

[MiDi]

As often: the problem sits between the ears... 
The En Pin is connected to Vin IC with 100k on this module, so no need to connect it externally for normal use case (and dangerous to do so) - corrected my last post.

I'm using that IC on my headphone amplifier and it works great. Don't understand why they need an inductor? I have had an eBay bought negative version TPS7A33 go short circuit due to insufficient solder on the heat pad. Otherwise the IC is bulletproof.

Think the inductor has multiple intentions: kind of inrush current limiting, decoupling from power supply and rejection of higher frequencies.

IMO it is far from bulletproof...

[SiliconWizard]

You said the inductor is between EN and Vin? That's pretty odd. Wouldn't that be a consequence of a misconfiguration of some jumper pins on the board?

Anyway, yes it was probably added to lower input noise as the PSRR of this regulator, given at 78 dB, is good but not that exceptional either - have seen linear regs with higher PSRR.
Not sure it really works well as a noise reduction measure. For inrush current maybe.

[MiDi]

You said the inductor is between EN and Vin? That's pretty odd. Wouldn't that be a consequence of a misconfiguration of some jumper pins on the board?

Vin - inductor - cap/Vin IC
En - 100k - Vin IC (pull-up)

If an external jumper is put from Vin to En than the inductor is between En and Vin IC.
Should not have done that, this caused the magic smoke 

If active shutdown is required for this module, the En Pin should only be driven by an open collector circuit to GND.