Maker Pro
Maker Pro

Diode Failure Mechanism & Mode

  • Thread starter RST Engineering \(jw\)
  • Start date
R

RST Engineering \(jw\)

Jan 1, 1970
0
Suppose I have a popcorn glass diode like a 1N4148. I will slowly advance
the temperature so as to not thermally shock the glass.

At what temperature (or range of temperatures) would you expect the diode to
fail?

What is the mechanism for the failure (thermal expansion, solder melt,
etc.)?

What is the mode of failure (short, open, gross deterioration, etc.)?


Thanks,

Jim
 
RST said:
Suppose I have a popcorn glass diode like a 1N4148. I will slowly advance
the temperature so as to not thermally shock the glass.

At what temperature (or range of temperatures) would you expect the diode to
fail?

What is the mechanism for the failure (thermal expansion, solder melt,
etc.)?

What is the mode of failure (short, open, gross deterioration, etc.)?

Depends whether you are applying a reverse voltage or not. If the diode
is back-biased, the leakage currents doubles every 10C as you raise the
temperature, and eventually the current reaches a level where the heat
dissipation within the junction is high enough for the junction
temperature to run away.

The diode then fails as a short circuit. With a suffiiciently potent
power supply, the diode can be vapourised - I did this once and all
that was left of the diode was two rounded ends on the bits of wire at
either end of where the diode used to be.

With no reverse voltage, you can probably get the diode hot enough for
the dopants that create the N- and P-type areas in the diode to diffuse
across the junction, wrecking the semi-conducting behaviour that we
rely on. Less exciting and much slower.
 
T

tlbs101

Jan 1, 1970
0
RST said:
Suppose I have a popcorn glass diode like a 1N4148. I will slowly advance
the temperature so as to not thermally shock the glass.

At what temperature (or range of temperatures) would you expect the diode to
fail?

What is the mechanism for the failure (thermal expansion, solder melt,
etc.)?

What is the mode of failure (short, open, gross deterioration, etc.)?


Thanks,

Jim

According to empirically compiled data over many years (decades?), Mil
Handbook 338 lists 3 modes of failure for diodes (such as the 1N4148)
and the likelyhood of each:

Short Circuit -- 75%
Intermittent -- 18%
Open circuit -- 6%
I suppose "other" would be the remaining 1%, or the numbers were
truncated after the decimal place.

These are failures under normal operating circumstances. At higher
temperatures, I don't know how the probabilites would change, but I
don't believe the percentages would change that much.

As for mechanisms, you would have to get a materials expert to answer
that. I have personally seen overheated diodes "go" both ways, most
recently being a zener that shorted out.

Tom
 
R

RST Engineering \(jw\)

Jan 1, 1970
0
Approximately 1 milliampere of forward current. 5 volt supply with 4.7k
resistor in series, with the diode far away from the current source.

Jim
 
S

Stanislaw Flatto

Jan 1, 1970
0
RST said:
Suppose I have a popcorn glass diode like a 1N4148. I will slowly advance
the temperature so as to not thermally shock the glass.

At what temperature (or range of temperatures) would you expect the diode to
fail?

What is the mechanism for the failure (thermal expansion, solder melt,
etc.)?

What is the mode of failure (short, open, gross deterioration, etc.)?


Thanks,

Jim
All that can be said about this IS: Whoever believes that components
fail in "certain" fashion does live in science fiction.
The best we have is some statistical numbers because you simple cannot
predict what happens in field and Mr. Murphy can always proove you wrong.

Have fun

Stanislaw
Slack user from Ulladulla.
 
J

Joerg

Jan 1, 1970
0
Hello Tom,
According to empirically compiled data over many years (decades?), Mil
Handbook 338 lists 3 modes of failure for diodes (such as the 1N4148)
and the likelyhood of each:

Short Circuit -- 75%
Intermittent -- 18%
Open circuit -- 6%
I suppose "other" would be the remaining 1%, or the numbers were
truncated after the decimal place.

More than 1% of the bad thru-hole era diodes I found had cracks in the
glass. Some of them were missing chunks of glass.
 
J

Jim Thompson

Jan 1, 1970
0
Hello Tom,


More than 1% of the bad thru-hole era diodes I found had cracks in the
glass. Some of them were missing chunks of glass.

Same as the TO-220 lead bending failures. Somewhere around here in
one of my tool boxes I have a red plastic strip that you drop
resistors (or diodes) into a slot with the lead spacing you want and
bend without stress.

...Jim Thompson
 
P

Pooh Bear

Jan 1, 1970
0
Jim said:
Same as the TO-220 lead bending failures. Somewhere around here in
one of my tool boxes I have a red plastic strip that you drop
resistors (or diodes) into a slot with the lead spacing you want and
bend without stress.

Mine was blue. No idea where it is now.

Graham
 
J

John Fields

Jan 1, 1970
0
Same as the TO-220 lead bending failures. Somewhere around here in
one of my tool boxes I have a red plastic strip that you drop
resistors (or diodes) into a slot with the lead spacing you want and
bend without stress.

---
Even that's no guarantee. The 100% right way to do it is to grab
the lead with a pair of longnose pliers in one hand and then to bend
the free (non-diode) end of the lead where it exits the pliers.

Same way with T0-220s; grab the lead where it necks down on the
package side of the lead and then bend the free end of it down.

Start with the middle lead. :)
 
J

Jim Thompson

Jan 1, 1970
0
Mine was blue. No idea where it is now.

Graham

Same here. I was going to post a picture, but I couldn't find it :-(

...Jim Thompson
 
R

RST Engineering \(jw\)

Jan 1, 1970
0
OOOOOOOOOkay. Let me try again. The diode is being operated forward bias
at about a milliampere. It has been handled with kid gloves from the day it
left the factory until it was carefully installed with no stress on the
leads into the circuit. I put the sucker into an oven and start slowly
bringing the temperature up.

At what temperature (or AROUND what temperature) would you expect the diode
to reach an elastic limit of some sort, either physical or electrical, and
not come back to approximately the same forward voltage at room temperature
than it had before the thermal stress test? What do you SUSPECT might be
the mode of the failure.

Jim
 
R

Robert Baer

Jan 1, 1970
0
RST said:
Suppose I have a popcorn glass diode like a 1N4148. I will slowly advance
the temperature so as to not thermally shock the glass.

At what temperature (or range of temperatures) would you expect the diode to
fail?

What is the mechanism for the failure (thermal expansion, solder melt,
etc.)?

What is the mode of failure (short, open, gross deterioration, etc.)?


Thanks,

Jim
Define "fail".
Leakage greater than a specified current at specified voltage?
Avalanche breakdown voltage change more than x percent WET room themp
value when measuerd at a specified current?
Dump the diode into an oven pre-heated to 200C and find that it is
still useable 10 seconds later when it is more than thermally stabilized.
Note i did not define "useable".
 
R

Robert Baer

Jan 1, 1970
0
RST said:
Approximately 1 milliampere of forward current. 5 volt supply with 4.7k
resistor in series, with the diode far away from the current source.

Jim
Under those conditions, 200C is just as good as 20C or -75C.
 
R

Robert Baer

Jan 1, 1970
0
Stanislaw said:
All that can be said about this IS: Whoever believes that components
fail in "certain" fashion does live in science fiction.
The best we have is some statistical numbers because you simple cannot
predict what happens in field and Mr. Murphy can always proove you wrong.

Have fun

Stanislaw
Slack user from Ulladulla.
...and so, if you want to decrease collision with Mr Murphy, then revise
the design for Justin Case...
 
R

Robert Baer

Jan 1, 1970
0
Joerg said:
Hello Tom,


More than 1% of the bad thru-hole era diodes I found had cracks in the
glass. Some of them were missing chunks of glass.
And that was due to added stress in the glass-to-metal seal; the
extrastress caused by: 1) socketed type testing (proven in MIL
reliability tests at FSC in the 80's), 2) assembly (no stress relief
when leads were bent for board insertion, and 3) no stress relief added
in the shaping of the leads (see NASA and other documents showing a "U"
type "hook" on one end).
 
R

Robert Baer

Jan 1, 1970
0
Jim said:
Same as the TO-220 lead bending failures. Somewhere around here in
one of my tool boxes I have a red plastic strip that you drop
resistors (or diodes) into a slot with the lead spacing you want and
bend without stress.

...Jim Thompson
I also have one of those, and they are very handy.
BUT.
They do *not* eliminate stress, in fact, one could show that they do
little to reduce stress at the lead-body interface, which is critical
for glass encased devices.
 
R

Robert Baer

Jan 1, 1970
0
John said:
---
Even that's no guarantee. The 100% right way to do it is to grab
the lead with a pair of longnose pliers in one hand and then to bend
the free (non-diode) end of the lead where it exits the pliers.

Same way with T0-220s; grab the lead where it necks down on the
package side of the lead and then bend the free end of it down.

Start with the middle lead. :)
Excellent start, but better if a gap is left between the part and the
pliers.
Do not remember the recommended minimum gap.
 
R

Robert Baer

Jan 1, 1970
0
Jim said:
Same here. I was going to post a picture, but I couldn't find it :-(

...Jim Thompson
The picture, the post, or the horse tied to the post?
 
D

David L. Jones

Jan 1, 1970
0
RST said:
Suppose I have a popcorn glass diode like a 1N4148. I will slowly advance
the temperature so as to not thermally shock the glass.

At what temperature (or range of temperatures) would you expect the diode to
fail?

What is the mechanism for the failure (thermal expansion, solder melt,
etc.)?

What is the mode of failure (short, open, gross deterioration, etc.)?


Thanks,

Jim

In addition to the other info already presented, I suspect it would
vary (maybe greatly) with the *brand* of diode used as well.

Dave :)
 
Top