Searching \ for '[EE]: MOSFet Troubles' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=mosfet+troubles
Search entire site for: 'MOSFet Troubles'.

Exact match. Not showing close matches.
PICList Thread
'[EE]: MOSFet Troubles'
2005\03\27@035053 by Milosz Kardasinski

picon face
Is it possible for a mosfet fail and stay in conduction mode? I have a
IRF840 that I'm using in a circuit to switch an automotive coil. With
no gate voltage applied and 12 volts applied across drain-source; the
mosfet conducts approx 8 volts. When 5v is applied to the gate the fet
switches on completely.

My whether the mos is bad or did I select a part which doesn't switch
off with a negative gate voltage. Link to the spec sheet is below

http://rocky.digikey.com/WebLib/ST%20Micro/Web%20Data/IRF840.pdf

Milosz Kardasinski
~~~ EE Infant ~~~

2005\03\27@102158 by Herbert Graf

flavicon
face
On Sun, 2005-03-27 at 03:50 -0500, Milosz Kardasinski wrote:
> Is it possible for a mosfet fail and stay in conduction mode? I have a
> IRF840 that I'm using in a circuit to switch an automotive coil. With
> no gate voltage applied and 12 volts applied across drain-source; the
> mosfet conducts approx 8 volts. When 5v is applied to the gate the fet
> switches on completely.

I don't know specifically about your part, but to answer your question:
yes, that IS a mode of failure I've seen.

In my case I was building a class D output stage and everything was
fine, except I was seeing to much of the switching frequency in the
output. So I put together a simple low pass filter and added it. Only
took a second or two before two of my FETs started getting really hot.
Some investigation revealed that they were no longer turning off
properly. Some simming showed that the configuration of the filter and
my load had created a resonance at the switching frequency. Not good...

Anyways, to get back to you, do you have a flyback diode on the coil?
You absolutely require some sort of device to take the current of the
coil when you switch the FET off, otherwise the coil will do what an
inductor with current flowing through it does when you try to shut off
the current: it will generate as high a voltage as necessary to keep the
current flowing, in your case possibly damaging the FET.

TTYL


-----------------------------
Herbert's PIC Stuff:
http://repatch.dyndns.org:8383/pic_stuff/

2005\03\27@120557 by Milosz Kardasinski

picon face
The fet has a built in diode, is this diode insufficient?


On Sun, 27 Mar 2005 10:21:57 -0500, Herbert Graf
<spam_OUTmailinglist2TakeThisOuTspamfarcite.net> wrote:
>
> Anyways, to get back to you, do you have a flyback diode on the coil?

2005\03\27@123443 by Peter Johansson

flavicon
face
Milosz Kardasinski writes:

> The fet has a built in diode, is this diode insufficient?

>From what I've been able to gather, the internal protection diodes are
not sufficient for highly inductive loads.

-p.

2005\03\27@123455 by Herbert Graf

flavicon
face
On Sun, 2005-03-27 at 12:05 -0500, Milosz Kardasinski wrote:
> The fet has a built in diode, is this diode insufficient?

Possibly. Is the rating for that diode enough to handle the coil you're
using? As a start I would try adding a beefy external diode right across
the coil and swapping the FET. TTYL

-----------------------------
Herbert's PIC Stuff:
http://repatch.dyndns.org:8383/pic_stuff/

2005\03\27@162406 by Herbert Graf

flavicon
face
On Sun, 2005-03-27 at 12:34 -0500, Peter Johansson wrote:
> Milosz Kardasinski writes:
>
> > The fet has a built in diode, is this diode insufficient?
>
> From what I've been able to gather, the internal protection diodes are
> not sufficient for highly inductive loads.

Something I'd like to add is that the "diode" you see when it comes to
FETs isn't something the manufacturer ADDED to the device. The
"diode" (often referred to as the body diode) is a byproduct of how a
FET is structured (basically if you have a FET you have a reverse biased
diode as well). This is similar to how most CMOS structures have an SCR
buried in the structure, on that can be activated if certain bad things
happen (called "latchup" by many people). As such it's sometimes just an
afterthought and in many datasheets isn't even indicated as present.

The fact that IR included it in the datasheet makes me think that they
have taken some care into designing the structure of the diode so that
it can substitute for the external diode normally present (for example
when using BJTs with inductive loads) in SOME cases. I don't believe
that IR has ensured that the body diode is structured in such a way that
it should be able to substitute for ALL cases (especially highly
inductive loads).

Generally, unless I have a VERY good reason (cost, space, etc.) I ALWAYS
include an external diode, it just makes me more comfortable that things
will behave the way I expect them to.

TTYL


-----------------------------
Herbert's PIC Stuff:
http://repatch.dyndns.org:8383/pic_stuff/

2005\03\27@184443 by Nigel Duckworth

picon face
I'm a little out of my depth here but isn't the FET
body diode in the wrong place to suppress the
back EMF from the coil?

I thought current needed to circulate within the coil
at switch-off which is the reason for the reverse
biased diode commonly seen across relay coils etc?

And assuming the "automotive coil" referred to
by the OP is an ignition circuit then is
suppressing the back EMF something
you want to be doing anyway?



--{Original Message removed}

2005\03\27@185423 by Dave VanHorn

flavicon
face
At 06:42 PM 3/27/2005, Nigel Duckworth wrote:
>I'm a little out of my depth here but isn't the FET
>body diode in the wrong place to suppress the
>back EMF from the coil?
>

If it's open-drain, and the diode was a zener, then it would be in the
right place.

I sometimes use zeners across the transistor, if putting them at the coil
isn't practical, and I can't get to the coil's V+.


2005\03\27@220658 by Russell McMahon

face
flavicon
face
> I'm a little out of my depth here

But you are still completel;y correct :-).


> but isn't the FET
> body diode in the wrong place to suppress the
> back EMF from the coil?

Yes. Wrong polarity.

> I thought current needed to circulate within the coil
> at switch-off which is the reason for the reverse
> biased diode commonly seen across relay coils etc?

Yes, and ...

> And assuming the "automotive coil" referred to
> by the OP is an ignition circuit then is
> suppressing the back EMF something
> you want to be doing anyway?

No - not if the plug fires correctly.
But, YES if it fails to fire.

In an ignition circuit you are relying on the "flyback" EMF to
generate the spark.
The primary will rise to a high voltage, and the secondary to a
proportionally higher voltage.

IF the spark plug fires then it will clamp the primary to a lower
voltage based on plug voltage and transformer turns ratio (and other
2nd order factors).  If the plug does NOT fire then primary bvoltage
may rise VERY high. In an automotove circuit you MUST design for
somethng to dissipate the energy when the plug mis-fires. This may be
an auxilliary sparkgap, a zener (as Dave mentions) or the FET can be a
type designed to take repetitive avalanche energy. The energy is
usually severe and few FETs would stand it indefinitely.  The FET MUSt
be specifically avalanche rated if used this way - ones that aren't
will conduct unevenly across the die and fry parts of thermselves
which leads to total failure.

The failed FET has probably been destroyed by avalanche energy. What
precautions are taken in the circuit to prevent this?



       RM




>
>
>
> --{Original Message removed}

2005\03\27@224028 by Herbert Graf

flavicon
face
On Mon, 2005-03-28 at 00:42 +0100, Nigel Duckworth wrote:
> I'm a little out of my depth here but isn't the FET
> body diode in the wrong place to suppress the
> back EMF from the coil?

That depends on the circuit config, which I don't think the op
specifically mentioned.

> I thought current needed to circulate within the coil
> at switch-off which is the reason for the reverse
> biased diode commonly seen across relay coils etc?

I internally was thinking of an H-Bridge where the body diodes ARE in
the "correct" place, I'd still put in real diodes though...

> And assuming the "automotive coil" referred to
> by the OP is an ignition circuit then is
> suppressing the back EMF something
> you want to be doing anyway?

If this is a distributerless ignition we are talking about then a hard
clamp of one diode drop would not be what you want, but even then you
MUST ensure the primary doesn't go above what the FET can handle. This
can occur if the plug doesn't fire (i.e. trying to crank the engine with
the plug wire disconnected, something that is sometimes done to diagnose
certain issues), in which case you need to clamp the back EMF to
something that will not damage the driver.

TTYL

-----------------------------
Herbert's PIC Stuff:
http://repatch.dyndns.org:8383/pic_stuff/

2005\03\28@000836 by Dave VanHorn

flavicon
face

>
>IF the spark plug fires then it will clamp the primary to a lower voltage
>based on plug voltage and transformer turns ratio (and other 2nd order
>factors).  If the plug does NOT fire then primary bvoltage may rise VERY
>high. In an automotove circuit you MUST design for somethng to dissipate
>the energy when the plug mis-fires.

Or when the mechanic pulls that plug wire to see which cylinder is
misfiring. Wouldn't THAT be a tragic design? :)


>  This may be an auxilliary sparkgap, a zener (as Dave mentions) or the
> FET can be a type designed to take repetitive avalanche energy. The
> energy is usually severe and few FETs would stand it indefinitely.  The
> FET MUSt be specifically avalanche rated if used this way - ones that
> aren't will conduct unevenly across the die and fry parts of thermselves
> which leads to total failure.

I've seen SMPS designs that also limit the turn-off speed a bit, so that
the collapse is slower, but I'm not sold on the technique.

I do know that the destructive spikes can be so fast that you can't see
them with a 100 MHz scope and proper probe placement.


2005\03\28@002225 by Russell McMahon
face
flavicon
face
> The fact that IR included it in the datasheet makes me think that
> they
> have taken some care into designing the structure of the diode so
> that
> it can substitute for the external diode normally present (for
> example
> when using BJTs with inductive loads) in SOME cases. I don't believe
> that IR has ensured that the body diode is structured in such a way
> that
> it should be able to substitute for ALL cases (especially highly
> inductive loads).

If the body diode is characterised in the data sheet it has probably
been to some extent designed as well, even though it's an unavoidable
attribute of a MOSFET. When they have avalanche energy ratings it
means (at least that) the manufacturer has taken special care to
spread the current evenly over all the sub transistors and to ensure
that it doesn't get hot spotting under avalanche breakdown. Exceeding
rated avalanche energy is a recipe for disaster. I'd guess the noted
"part on" behaviour mentioned earlier in this thread MAY have been
caused by some sub transistors in a die being permanently on and
others off.


       RM

2005\03\28@003805 by Denny Esterline

picon face
> >
> >IF the spark plug fires then it will clamp the primary to a lower voltage
> >based on plug voltage and transformer turns ratio (and other 2nd order
> >factors).  If the plug does NOT fire then primary bvoltage may rise VERY
> >high. In an automotove circuit you MUST design for somethng to dissipate
> >the energy when the plug mis-fires.
>
> Or when the mechanic pulls that plug wire to see which cylinder is
> misfiring. Wouldn't THAT be a tragic design? :)
>
>

Actually I've seen a failure that turned out to be very similar to that.

It was a '86 Buick skyhawk (IIRC). No spark, test light on the coil primary
blinked when cranking- so I replaced the coil. Ran fine, came back two days
later with the same symptoms. I replaced the coil and it ran fine again. I
assumed I had gotten a bad coil and I sent the car back out. Came back a
week later. Long story short, after the fourth coil from a different
supplier I traced it back to a wire on the pickup that was rubbed through.
Net effect was that the coil was firing when the distributor was pointing
between cylinders and the voltage went high enough to cause flashover inside
the coil.

Seems like I always get the weird failures....

-Denny

2005\03\28@012458 by Dave VanHorn

flavicon
face

>
>
>Seems like I always get the weird failures....


I just had my truck repaired again..
Took it into a ford dealer a year ago (time flies)  when I was traveling in
Ga, and sick as a dog.
It suddenly started missing very badly, about 1000 miles from home.
He replaced three cracked spark plugs for $300

He didn't grease them, and he didn't re-loom the wires. He also didn't
tighten them very well.
One of them unscrewed itself in operation recently, and they were all
arcing under the boots enough to ruin the plug wires.
Some of the wires were shorting to ground where they had worn through, and
others were just damaged.

Unfortunately, changing the plug wires on this beast is quite a job, and
I'm pretty busy these days, so I had to pay to get it repaired again. ($200)

What with buying two houses, moving my office, cranking up my business,
taking on a major rush project, commuting to the east coast for alternate
weeks, becoming a grandfather again, and loosing my grandmother, I've been
rather busy lately.

There was a time when I considered it foolish to pay someone to change my oil.
Then again, I also used to play games..


Embedded development is sort of a game, but the graphics suck.



2005\03\28@032720 by Nigel Duckworth

picon face
Thanks to all who replied for this timely information.

I'm currently researching suitable switching devices
and techniques for an electronic ignition system for
my old motorcycle, PIC driven of course :)

There are some interesting IGBTs available specifically
designed for ignition applications (Fairchild & IR).

Regards,

Nigel Duckworth



--{Original Message removed}

2005\03\28@080050 by olin_piclist

face picon face
Herbert Graf wrote:
> If this is a distributerless ignition we are talking about then a hard
> clamp of one diode drop would not be what you want,

I haven't really been following this that closely, but a diode to catch
kickback from the primary of an ignition coil would cause a diode drop plus
12V, not just a diode drop.  Again, maybe a different topology was discussed
than what I'm thinking about.

An ignition coil is essentially a pulse transformer with a very high output
ratio.  You want to let the primary go as high as you can tolerate to force
most of the energy out the secondary.  It seems to me that a simple diode to
+12V in series with a resistor would be adequate.  The maximum possible
current in the primary is easily known, so the resistor can be sized so that
the primary voltage is just below the tolerable limit at maximum current.
The voltage will be less in all other cases.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

2005\03\28@083031 by Russell McMahon

face
flavicon
face
> An ignition coil is essentially a pulse transformer with a very high
> output
> ratio.  You want to let the primary go as high as you can tolerate
> to force
> most of the energy out the secondary.  It seems to me that a simple
> diode to
> +12V in series with a resistor would be adequate.  The maximum
> possible
> current in the primary is easily known, so the resistor can be sized
> so that
> the primary voltage is just below the tolerable limit at maximum
> current.
> The voltage will be less in all other cases.

Yes. Using a zener has the advantage of taking no energy until
misfiring occurs. It then however transfers the dissipation problem
from FET to zener. The resistor has the advantage of better ruggedness
per $.

   RM



2005\03\28@122358 by Milosz Kardasinski

picon face
> IF the spark plug fires then it will clamp the primary to a lower
> voltage based on plug voltage and transformer turns ratio (and other
> 2nd order factors).  If the plug does NOT fire then primary bvoltage
> may rise VERY high. In an automotove circuit you MUST design for
> somethng to dissipate the energy when the plug mis-fires. This may be
> an auxilliary sparkgap, a zener (as Dave mentions) or the FET can be a
> type designed to take repetitive avalanche energy. The energy is
> usually severe and few FETs would stand it indefinitely.  The FET MUSt
> be specifically avalanche rated if used this way - ones that aren't
> will conduct unevenly across the die and fry parts of thermselves
> which leads to total failure.
>
> The failed FET has probably been destroyed by avalanche energy. What
> precautions are taken in the circuit to prevent this?

You bring up some excellent points, I never considered effect of the
coil not firing. This is probably the reason I have a dead FET.

Righ now I don't have anything in the circuit to handle avalanche
energy. I simply whipped up the circuit to test an idea.

Although I'm using an ignition coil, this little experiment has
nothing to do with automotive ignition. I'm trying to whip up a device
that will aid in the striking of a plasma. The ignition coil is a
cheap and easy method to get HV arcs across a gap.

2005\03\29@103121 by Support - KF4HAZ

flavicon
face
Better still, use a 6 volt coil with an automotive "coil resistor" in series with coil+
add a "condenser" (small high voltage cap) from coil- to Gnd. and a 2.5Amp 1Kv diode from coil- to 12V.
Connect Drain lead to coil- with source grounded.
I ran this setup on a VW for many years, gateing the mosfet with a pnp pull-up with the base grounded by the points (through a resistor), not only did the mosfet hold up, but the points no longer burned-up so frequently. I did have to bump the timing up a degree or 2? to compensate for the switching times of the transistor and mosfet.

KF4HAZ - Lonnie

----- From: "Russell McMahon" <apptech@

{Quote hidden}

2005\03\29@103522 by Support - KF4HAZ

flavicon
face
Forgot to mention connecting a line from the "start" position of the ignition switch to the coil+ to provide more spark during cranking (when the battery was being pulled lower than normal)

Better still, use a 6 volt coil with an automotive "coil resistor" in series with coil+
add a "condenser" (small high voltage cap) from coil- to Gnd. and a 2.5Amp 1Kv diode from coil- to 12V.
Connect Drain lead to coil- with source grounded.
I ran this setup on a VW for many years, gateing the mosfet with a pnp pull-up with the base grounded by the points (through a resistor), not only did the mosfet hold up, but the points no longer burned-up so frequently. I did have to bump the timing up a degree or 2? to compensate for the switching times of the transistor and mosfet.

KF4HAZ - Lonnie

----- From: "Russell McMahon" <apptech@

{Quote hidden}

2005\03\30@055227 by Milosz Kardasinski

picon face
> Better still, use a 6 volt coil with an automotive "coil resistor" in series with coil+

What will a 6V coil give me that a 12V can't?

> add a "condenser" (small high voltage cap) from coil- to Gnd. and a 2.5Amp 1Kv diode from coil- to 12V.

If I remember correctly the purpose of the condeners is to stop buring
and pitting of the points. I'm not using to the coil to fire a spark
plug so I can't think of a reason I would need it. The diode I can see
being needed as Russell McMahon pointed out a flyback issue that most
likely killed my FET.

> Connect Drain lead to coil- with source grounded.
> I ran this setup on a VW for many years, gateing the mosfet with a pnp pull-up with the base grounded by the points (through a resistor), not only did the mosfet hold up, but the points no longer burned-up so frequently. I did have to bump the timing up a degree or 2? to compensate for the switching times of the transistor and mosfet.

I'm not using this to trigger spark in a car...I'm using this to
trigger fire the cutting arc of a plasma cutter, but thanks for the
suggestion.

Milosz K.

2005\03\30@064314 by Michael Rigby-Jones

picon face


{Quote hidden}

The 'condenser' actually speeds up the collapse of the magnetic field in
the coil by preventing (or at least reducing) any arc forming across the
points that could sustain the primary current.  A contact breaker system
without the condenser produces a very weak spark.

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\03\30@101926 by Support - KF4HAZ

flavicon
face

----- From: "Michael Rigby-Jones" <Michael.Rigby-Jones@
> >> Better still, use a 6 volt coil with an automotive "coil
> >resistor" in
> >> series with coil+
> >
> >What will a 6V coil give me that a 12V can't?
A 6 volt coil, with the appropriate resistor in series will give as hot of a spark as a 12v coil with the added bonus being that when the diode quenches the primary voltage, it 1 doesn't quench the secondary as much, and 2 produces less of a spike on the supply.
I have scoped side by side comparisons of the 2 configurations and seen the improvement with the 6 volt coil.
Of coarse the 3rd benefit (bypassing the resistor during cranking) is of no use in your application.
{Quote hidden}

> --

2005\03\30@104131 by Alan B. Pearce

face picon face
>Of coarse the 3rd benefit (bypassing the resistor during
>cranking) is of no use in your application.

except for the usual reason it has historically been done - starting under
real cold conditions. But then an electronic system may still have a beefy
enough spark that shorting the resistor is still unnecessary - this was
supposed to be one of the plusses for the early CDI units.

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...