Why do "short proof" votlage regulator chips short out?

[larrymoencurly]

I was using an LM317 to charge up a big capacitor. I connected it
after I turned on the power, and the LM317 blew out instantly. I
realize that it was stupid to turn on the power first, but why didn't
the LM317's built-in short-circuit and thermal protection work?

The LM317 was hooked up in linear mode, and I had bypass capacitors on
the input and output, just as National Semiconductor recommends (.1 uF
ceramic in parallel with 10 uF low-ESR Sanyo OS-Con, plus 2,000 uF
filter on the input side). The regulator was mounted on a fairly
large heatsink, roughly 3" x 3" x .75", and I think that I followed
proper layout recommendations.

 

[William R. Walsh]

Hi!

Any number of reasons, including, but not limited to:

* Minorly defective part being pushed to the edge
* Some kind of problem occured that was so big that the protection circuit
could do nothing to stop the destruction of the device and circuit.

Some internal protection circuits just don't seem to work well for some
reason...and maybe that was the case here. Perhaps you should an external
protection circuit.

William

 

[Jovit P]

Hi!

Did you forget to add protection diodes? The bypass capacitors are for
improved performance and the large heatsink provides adequate power
dissipation for long term reliability. These do not protect the LM317
against reverse current flow due to transient capacitor discharge. The
capacitor you've connected probably has higher potential than the
regulator's programmed output voltage thereby driving current into the
regulator's output pin/junction.

Jovit

larrymoencurly wrote:

I was using an LM317 to charge up a big capacitor. I connected it
after I turned on the power, and the LM317 blew out instantly. I
realize that it was stupid to turn on the power first, but why didn't
the LM317's built-in short-circuit and thermal protection work?

The LM317 was hooked up in linear mode, and I had bypass capacitors on
the input and output, just as National Semiconductor recommends (.1 uF
ceramic in parallel with 10 uF low-ESR Sanyo OS-Con, plus 2,000 uF
filter on the input side). The regulator was mounted on a fairly
large heatsink, roughly 3" x 3" x .75", and I think that I followed
proper layout recommendations.

 

[mike]

larrymoencurly wrote:

I was using an LM317 to charge up a big capacitor. I connected it
after I turned on the power, and the LM317 blew out instantly. I
realize that it was stupid to turn on the power first, but why didn't
the LM317's built-in short-circuit and thermal protection work?

The LM317 was hooked up in linear mode, and I had bypass capacitors on
the input and output, just as National Semiconductor recommends (.1 uF
ceramic in parallel with 10 uF low-ESR Sanyo OS-Con, plus 2,000 uF
filter on the input side). The regulator was mounted on a fairly
large heatsink, roughly 3" x 3" x .75", and I think that I followed
proper layout recommendations.

The obviously missing part of your story is the input voltage.
mike

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
Yaesu FTV901R Transverter, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[Tweetldee]

"mike" <spamme0@juno.com> wrote in message news:407B7C8B.205@juno.com...

larrymoencurly wrote:
I was using an LM317 to charge up a big capacitor. I connected it
after I turned on the power, and the LM317 blew out instantly. I
realize that it was stupid to turn on the power first, but why didn't
the LM317's built-in short-circuit and thermal protection work?

The LM317 was hooked up in linear mode, and I had bypass capacitors on
the input and output, just as National Semiconductor recommends (.1 uF
ceramic in parallel with 10 uF low-ESR Sanyo OS-Con, plus 2,000 uF
filter on the input side). The regulator was mounted on a fairly
large heatsink, roughly 3" x 3" x .75", and I think that I followed
proper layout recommendations.

The obviously missing part of your story is the input voltage.
mike

And the size of the "big capacitor"...


--
Dave M

Good judgement comes from experience, and a lot of that comes from bad
judgement.

 

[larrymoencurly]

Jovit P <pamarang_@_telus.net> wrote in message news:<407B7B64.7D7380EC@_telus.net>...

Did you forget to add protection diodes?

I have one between input and output and between output and ground for
reverse voltage protection, but I don't have one between output and
the adjustment pin because I don't have a capacitor across the
adjustment-ground resistor.

The bypass capacitors are for improved performance and
the large heatsink provides adequate power dissipation
for long term reliability. These do not protect the LM317
against reverse current flow due to transient capacitor discharge.
The capacitor you've connected probably has higher potential than
the regulator's programmed output voltage thereby driving current
into the regulator's output pin/junction.

Do the bypass capacitors on the output also help prevent excessive
voltage from inductance in the wiring, and are a .1uF ceramic and 10uF
electrolytic enough?

 

[larrymoencurly]

mike <spamme0@juno.com> wrote in message news:<407B7C8B.205@juno.com>...

I was using an LM317 to charge up a big capacitor. I
connected it after I turned on the power, and the LM317
blew out instantly. why didn't the LM317's built-in
short-circuit and thermal protection work?

I had bypass capacitors on the input and output, just as
National Semiconductor recommends (.1 uF ceramic in parallel
with 10 uF low-ESR Sanyo OS-Con, plus 2,000 uF filter on
the input side).

The obviously missing part of your story is the input voltage.

I'm using a transformer rated at 18VAC, and the maximum I once
measured at the LM317 input with no load was 31VDC. The capacitor I
was charging up is rated 50V, 3000uF.

 

[Jovit P]

I have one between input and output and between output and ground for
reverse voltage protection, but I don't have one between output and
the adjustment pin because I don't have a capacitor across the
adjustment-ground resistor.

Did you check the diode connected between the input and output for
short?

Do the bypass capacitors on the output also help prevent excessive
voltage from inductance in the wiring, and are a .1uF ceramic and 10uF
electrolytic enough?

The decoupling capacitor on the regulator output is for stability of the
regulator's feedback loop and improve the regulator's output impedance.
Im sure it's not intended to protect the regulator output.

 

[mike]

larrymoencurly wrote:

mike <spamme0@juno.com> wrote in message news:<407B7C8B.205@juno.com>...


I was using an LM317 to charge up a big capacitor. I
connected it after I turned on the power, and the LM317
blew out instantly. why didn't the LM317's built-in
short-circuit and thermal protection work?



I had bypass capacitors on the input and output, just as
National Semiconductor recommends (.1 uF ceramic in parallel
with 10 uF low-ESR Sanyo OS-Con, plus 2,000 uF filter on
the input side).


The obviously missing part of your story is the input voltage.

u
I'm using a transformer rated at 18VAC, and the maximum I once
measured at the LM317 input with no load was 31VDC. The capacitor I
was charging up is rated 50V, 3000uF.

31V is near the max that you can put on an LM317. Combination of high
voltage and high current can melt the junction before the current limit
can respond. Thermal protection is useless for transients. Output
transistor can pop long before the whole die heats up. The output cap
limits the rate of rise of the current to something the current limit
can keep up with. You didn't have one to slow down that transient.

I don't recall your output voltage, but you really don't want a lot
of volts on the chip if you can help it. Check out the derating curves
in the spec...then derate it some more...like twice more.

I wouldn't do what you did, but I would have expected it to
survive...usually. Maybe you're just lucky. Try it again.
mike

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
Yaesu FTV901R Transverter, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[Asimov]

"mike" bravely wrote to "All" (14 Apr 04 00:13:38)
--- on the heady topic of "Re: Why do "short proof" votlage regulator chips
short out?"

mi> From: mike <spamme0@juno.com>

mi> larrymoencurly wrote:

mike <spamme0@juno.com> wrote in message news:<407B7C8B.205@juno.com>...

I was using an LM317 to charge up a big capacitor. I
connected it after I turned on the power, and the LM317
blew out instantly. why didn't the LM317's built-in
short-circuit and thermal protection work?



I had bypass capacitors on the input and output, just as
National Semiconductor recommends (.1 uF ceramic in parallel
with 10 uF low-ESR Sanyo OS-Con, plus 2,000 uF filter on
the input side).


The obviously missing part of your story is the input voltage.

I'm using a transformer rated at 18VAC, and the maximum I once
measured at the LM317 input with no load was 31VDC. The capacitor I
was charging up is rated 50V, 3000uF.

mi> 31V is near the max that you can put on an LM317. Combination of high
mi> voltage and high current can melt the junction before the current
mi> limit can respond. Thermal protection is useless for transients.
mi> Output transistor can pop long before the whole die heats up. The
mi> output cap limits the rate of rise of the current to something the
mi> current limit can keep up with. You didn't have one to slow down that
mi> transient.
mi> I don't recall your output voltage, but you really don't want a lot
mi> of volts on the chip if you can help it. Check out the derating
mi> curves in the spec...then derate it some more...like twice more.

mi> I wouldn't do what you did, but I would have expected it to
mi> survive...usually. Maybe you're just lucky. Try it again.
mi> mike

The specs show 40 volts maximum input/output differential so the 31v
isn't so bad. However, looking at the internal diagram of the LM317
the one possibility is that the Adjust Pin 1 received power before
Vout and being reverse biased this may have zapped a junction to +Vcc.
Then without output feedback input the regulator would act as if it
was shorted. It would seem as if the output pass transistor was fried.
A fluke no doubt...

Moral: don't plug IC's into live powered circuitry!

Asimov
******

.... Engineers do it to maximum stress loads.

 

[larrymoencurly]

mike <spamme0@juno.com> wrote in message news:<407CE4A2.4070005@juno.com>...

larrymoencurly wrote:

I'm using a transformer rated at 18VAC, and the maximum I once
measured at the LM317 input with no load was 31VDC.

31V is near the max that you can put on an LM317. Combination
of high voltage and high current can melt the junction before
the current limit can respond. Thermal protection is useless
for transients. Output transistor can pop long before the
whole die heats up. The output cap limits the rate of rise
of the current to something the current limit can keep up
with. You didn't have one to slow down that transient.

I had 10uF in parallel with 0.1uF on the output. Was that too little?

I don't recall your output voltage, but you really don't want
a lot of volts on the chip if you can help it. Check out
the derating curves in the spec...then derate it some more...
like twice more.

I wouldn't do what you did, but I would have expected it to
survive...usually. Maybe you're just lucky. Try it again.

My output voltage was at the maximum, about 27VDC, and I thought that
feeding 31VDC into the LM317 was OK because National warns against
exceeding 40V across it. The LM317 definitely blew because the
voltage adjustment pot would no longer lower the voltage. I tried
another LM317, same T-220 package, and it didn't blow when the same
capacitor (also 0V) was connected across it, but it's a different
brand, National, instead of ST like the first one.

 

[mike]

larrymoencurly wrote:

mike <spamme0@juno.com> wrote in message news:<407CE4A2.4070005@juno.com>...

larrymoencurly wrote:


I'm using a transformer rated at 18VAC, and the maximum I once
measured at the LM317 input with no load was 31VDC.


31V is near the max that you can put on an LM317. Combination
of high voltage and high current can melt the junction before
the current limit can respond. Thermal protection is useless
for transients. Output transistor can pop long before the
whole die heats up. The output cap limits the rate of rise
of the current to something the current limit can keep up
with. You didn't have one to slow down that transient.


I had 10uF in parallel with 0.1uF on the output. Was that too little?

Your regulator thought so...

I don't recall your output voltage, but you really don't want
a lot of volts on the chip if you can help it. Check out
the derating curves in the spec...then derate it some more...
like twice more.

I wouldn't do what you did, but I would have expected it to
survive...usually. Maybe you're just lucky. Try it again.


My output voltage was at the maximum, about 27VDC, and I thought that
feeding 31VDC into the LM317 was OK because National warns against
exceeding 40V across it.

This is a popular misconception. Just cause the maximum sez 40V,
doesn't mean you can do anything useful at that voltage. Make sure
the spec you're reading is for the actual part you have. I pulled up
a random spec when I first read your post and it said 37V.
Semiconductors get extremely fragile near their breakdown voltage.
Read up on secondary breakdown and current crowding. The junction
can melt before the current limit can take over. At higher voltages,
maximum power dissipation is often NOT the limiting factor.

TheLM317 is particularly vulnerable because you can float the thing
at an arbitrarily high voltage as long as you don't exceed 40 volts
across it. But you have to make sure it doesn't blow up on shorted
output or during turn-on.

Another thing to worry about is where you got it. Direct from the
manufacturer, it's likely to meet it's specification...BUT the critical
parameter in your application may not even be on the spec. If you got
it from Radio Shack or a generic replacement or at a swapmeet or any
other place, you may
have a part that was rejected somewhere along the line.

Bean counters like to buy from the lowest bidder. Different vendor
parts vary over time.
mike


The LM317 definitely blew because the

voltage adjustment pot would no longer lower the voltage. I tried
another LM317, same T-220 package, and it didn't blow when the same
capacitor (also 0V) was connected across it, but it's a different
brand, National, instead of ST like the first one.

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
Yaesu FTV901R Transverter, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[tempus fugit]

I think you are misreading the spec. A 317 cannot exceed 37v difference
between the input and output voltages. There is obviously going to be an
upper limit, but 40v ain't it. I am using a 317 with about 70v on the input
to put out 48v without any trouble at all.


"mike" <spamme0@juno.com> wrote in message news:407E52E6.7010204@juno.com...

larrymoencurly wrote:
mike <spamme0@juno.com> wrote in message
news:<407CE4A2.4070005@juno.com>...

larrymoencurly wrote:


I'm using a transformer rated at 18VAC, and the maximum I once
measured at the LM317 input with no load was 31VDC.


31V is near the max that you can put on an LM317. Combination
of high voltage and high current can melt the junction before
the current limit can respond. Thermal protection is useless
for transients. Output transistor can pop long before the
whole die heats up. The output cap limits the rate of rise
of the current to something the current limit can keep up
with. You didn't have one to slow down that transient.


I had 10uF in parallel with 0.1uF on the output. Was that too little?

Your regulator thought so...



I don't recall your output voltage, but you really don't want
a lot of volts on the chip if you can help it. Check out
the derating curves in the spec...then derate it some more...
like twice more.

I wouldn't do what you did, but I would have expected it to
survive...usually. Maybe you're just lucky. Try it again.


My output voltage was at the maximum, about 27VDC, and I thought that
feeding 31VDC into the LM317 was OK because National warns against
exceeding 40V across it.

This is a popular misconception. Just cause the maximum sez 40V,
doesn't mean you can do anything useful at that voltage. Make sure
the spec you're reading is for the actual part you have. I pulled up
a random spec when I first read your post and it said 37V.
Semiconductors get extremely fragile near their breakdown voltage.
Read up on secondary breakdown and current crowding. The junction
can melt before the current limit can take over. At higher voltages,
maximum power dissipation is often NOT the limiting factor.

TheLM317 is particularly vulnerable because you can float the thing
at an arbitrarily high voltage as long as you don't exceed 40 volts
across it. But you have to make sure it doesn't blow up on shorted
output or during turn-on.

Another thing to worry about is where you got it. Direct from the
manufacturer, it's likely to meet it's specification...BUT the critical
parameter in your application may not even be on the spec. If you got
it from Radio Shack or a generic replacement or at a swapmeet or any
other place, you may
have a part that was rejected somewhere along the line.

Bean counters like to buy from the lowest bidder. Different vendor
parts vary over time.
mike


The LM317 definitely blew because the
voltage adjustment pot would no longer lower the voltage. I tried
another LM317, same T-220 package, and it didn't blow when the same
capacitor (also 0V) was connected across it, but it's a different
brand, National, instead of ST like the first one.



--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
Yaesu FTV901R Transverter, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/