strange defect with MMBT2222

[Jamie]

http://www.fairchildsemi.com/ds/MM/MMBT2222.pdf

We decided to condense an older design using SM components
so that we could squeeze this board into a small area.

Got 20 of these from fairchild. after initial testing
of the first one. We found some spurious noise being
generated only at fixed levels of operation.

After a little testing, we decided to put a MMBTB2222 in a
simple common emitter with resistor load on the collector circuit,
using a variable bias, we found at ~ 15 ma of Ice, there
was ~ 0.5 vpp erratic oscillation in the area of 100k..150khz depending
on where the bias was. below or above this point would then place the
transistor back in normal operation once again.
Testing with a 2N2222 operated fine, I also found another SMT type
2222 which tested fine.

We'll be looking for another semi vender with this same number, I
don't want to buy another batch of this number from fairchild just in
case I get the same defects.

Have any of you ever seen strange things like this take place?

P.S.
It works fine being used in a saturated switch circuit obviously.

 

[Wimpie]

http://www.fairchildsemi.com/ds/MM/MMBT2222.pdf

We decided to condense an older design using SM components
so that we could squeeze this board into a small area.

Got 20 of these from fairchild. after initial testing
of the first one. We found some spurious noise being
generated only at fixed levels of operation.

After a little testing, we decided to put a MMBTB2222 in a
simple common emitter with resistor load on the collector circuit,
using a variable bias, we found at ~ 15 ma of Ice, there
was ~ 0.5 vpp erratic oscillation in the area of 100k..150khz depending
on where the bias was. below or above this point would then place the
transistor back in normal operation once again.
Testing with a 2N2222 operated fine, I also found another SMT type
2222 which tested fine.

We'll be looking for another semi vender with this same number, I
don't want to buy another batch of this number from fairchild just in
case I get the same defects.

Have any of you ever seen strange things like this take place?

P.S.
It works fine being used in a saturated switch circuit obviously.

Hello Jamie,

Probably you changed more then just the transistor (also capacitors,
resistors, inductors, etc). All of these will have different
characteristics (ESR, self resonant frequency, etc).

Maybe the circuit is just at the point of oscillation or no
oscillation. Whether or not is will oscillate may depend on some
slight variation in transistor characteristics (as observed by your
experiments). I would recommend you to trace the problem, when the
product goes into mass production.

I don't know what type of circuit it is, but oscillation at 100..150
kHz is mostly not induced by little bit parasitic induction or stray
capacitance. . When changing from conventional to SMT, I would expect
oscillation at far higher frequencies (in the MHz range as this
transistor has power gain above 100 MHz). How did you determine the
oscillation? Be aware of anti-aliasing of Digital oscilloscopes.

In my opinion, it is to early to say the the Fairchild transistor is
"defect".

Transistors are more likely to oscillate when both base and collector
is inductively loaded. When the emitter (w.r.t. ground) is capacitive
loaded, it is very easy to get XXXX2222 to oscillate in the MHz
range.,

Many emitter follower circuits will oscillate when the base is
inductive loaded in combination with a capacitive emitter load (for
example the input capacitance of a digital device together with trace
capacitance). I frequently make use this type of oscillation in
oscillators into the GHz range.

Best regards,

Wim
PA3DJS
www.tetech.nl

 

[Eeyore]

http://www.fairchildsemi.com/ds/MM/MMBT2222.pdf

We decided to condense an older design using SM components
so that we could squeeze this board into a small area.

Got 20 of these from fairchild. after initial testing
of the first one. We found some spurious noise being
generated only at fixed levels of operation.

After a little testing, we decided to put a MMBTB2222 in a
simple common emitter with resistor load on the collector circuit,
using a variable bias, we found at ~ 15 ma of Ice, there
was ~ 0.5 vpp erratic oscillation in the area of 100k..150khz depending
on where the bias was. below or above this point would then place the
transistor back in normal operation once again.
Testing with a 2N2222 operated fine, I also found another SMT type
2222 which tested fine.

We'll be looking for another semi vender with this same number, I
don't want to buy another batch of this number from fairchild just in
case I get the same defects.

Have any of you ever seen strange things like this take place?

P.S.
It works fine being used in a saturated switch circuit obviously.

Scott Dorsey in rec.audio.pro has had trouble with 'Fairchild' devices recently.

The boring routine parts got subbed out to some third rate fab that produces
crap. That's almost certainly your problem too. He saw it with noise figure
IIRC.

Check out ON's devices maybe or modernise a little and use a part that doesn't
date to the 1960s.

Graham

 

[Jamie]

Hello Jamie,

Probably you changed more then just the transistor (also capacitors,
resistors, inductors, etc). All of these will have different
characteristics (ESR, self resonant frequency, etc).

Maybe the circuit is just at the point of oscillation or no
oscillation. Whether or not is will oscillate may depend on some
slight variation in transistor characteristics (as observed by your
experiments). I would recommend you to trace the problem, when the
product goes into mass production.

In the circuit, it flat lines at what I assume is it's 15 ma
Slew/Skew point of the ramp signal on the output. In circuit you don't
see the 100k..150khz signal, you just see roll over instead of a linear
ramp. At least for that time period. I contribute this to the parasitic
resistors and caps I put in that is removing the actual oscillation and
just producing this effect. This lead me to think there was something
wrong in the transistor.

I don't know what type of circuit it is, but oscillation at 100..150
kHz is mostly not induced by little bit parasitic induction or stray
capacitance. . When changing from conventional to SMT, I would expect
oscillation at far higher frequencies (in the MHz range as this
transistor has power gain above 100 MHz). How did you determine the
oscillation? Be aware of anti-aliasing of Digital oscilloscopes.

It's used as part of a bipolar output to drive a magnetic coil for
a beam steering system to adjust the focus point of an irradiation unit.

I set up a simple common emitter test (DC), using a load resistor
on the collector from a 15 volt source. Just biased the BASE via
a 10k resistor from a variable trimmer reference, Place a by pass
cap (.1) from the base to common. at about 15 ma's, You can see this
OSC at the collector. In the real circuit the collector is connected to
the + rail and the emitter is driving the next stage via an R.
I used 3 different scopes to verify this, I found my 350 Mhz Tek give
me the clearest reading . All 3 scopes show the problem at the same
place. The other two were cheap digital LCD's.

I don't get oscillation at the final output because of rest of the
circuit, I just get ramp level shift at ~ 15 ma's, and it's because the
transistor is some how OSC at that point. It Almost act's like a tunnel
diode effect.

In my opinion, it is to early to say the the Fairchild transistor is
"defect".

Maybe, the other vender's worked fine, as well as the old stand by
plastic type 2222's

Transistors are more likely to oscillate when both base and collector
is inductively loaded. When the emitter (w.r.t. ground) is capacitive
loaded, it is very easy to get XXXX2222 to oscillate in the MHz
range.,
Many emitter follower circuits will oscillate when the base is
inductive loaded in combination with a capacitive emitter load (for
example the input capacitance of a digital device together with trace
capacitance). I frequently make use this type of oscillation in
oscillators into the GHz range.

Good point how ever, Since I strictly used a DC test circuit with no
inductive components, I don't see this happening.
The test circuit was a simple DC test jig.

Best regards,

Wim
PA3DJS
www.tetech.nl

Thanks

 

[Michael A. Terrell]

http://www.fairchildsemi.com/ds/MM/MMBT2222.pdf

We decided to condense an older design using SM components
so that we could squeeze this board into a small area.

Got 20 of these from fairchild. after initial testing
of the first one. We found some spurious noise being
generated only at fixed levels of operation.

After a little testing, we decided to put a MMBTB2222 in a
simple common emitter with resistor load on the collector circuit,
using a variable bias, we found at ~ 15 ma of Ice, there
was ~ 0.5 vpp erratic oscillation in the area of 100k..150khz depending
on where the bias was. below or above this point would then place the
transistor back in normal operation once again.
Testing with a 2N2222 operated fine, I also found another SMT type
2222 which tested fine.

We'll be looking for another semi vender with this same number, I
don't want to buy another batch of this number from fairchild just in
case I get the same defects.

Have any of you ever seen strange things like this take place?

P.S.
It works fine being used in a saturated switch circuit obviously.

MMBT2222 is On Semiconductor, formerly Motorola. The '2222' makes it
the surface mount version of the 2N2222A TO-92 NPN silicon transistor.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[Winfield]

MMBT2222 is On Semiconductor, formerly Motorola. The '2222' makes it
the surface mount version of the 2N2222A TO-92 NPN silicon transistor.

Actually, the 2n2222 was a metal-can TO-18 part, right?
Not that it matters. Fab lines and therefore some BJT
die designs have dramatically changed with time. If you
check Fairchild's lineup you'll see many '2222 variants,
no doubt positioned to grab sales from whatever p/n the
buyer's BOM happens to have. Fairchild may also make
'2222 parts at more than one fab, since they purchased
or aligned with several other manufacturers.

 

[Eeyore]

MMBT2222 is On Semiconductor, formerly Motorola.

And also sold by 'Fairchild'.
http://www.fairchildsemi.com/pf/MM/MMBT2222.html

The '2222' makes it
the surface mount version of the 2N2222A TO-92 NPN silicon transistor.

And you think we didn't know that ?

Graham

 

[Eeyore]

Actually, the 2n2222 was a metal-can TO-18 part, right?

And the 2N2222A.

Not that it matters.

Aside from the huge cost difference. Hermetic metal can devices aren't cheap.

Fab lines and therefore some BJT
die designs have dramatically changed with time. If you
check Fairchild's lineup you'll see many '2222 variants,
no doubt positioned to grab sales from whatever p/n the
buyer's BOM happens to have. Fairchild may also make
'2222 parts at more than one fab, since they purchased
or aligned with several other manufacturers.

The TO-92 version was the PN2222. All have very low current gain by modern
standards which is one reason I'm forever amazed that they still feature in
cookbooks.

Graham

 

[Tim Williams]

The TO-92 version was the PN2222. All have very low current gain by modern
standards which is one reason I'm forever amazed that they still feature in
cookbooks.

Wazzat? To the best of my recollection, there's no difference between the
2222 and the 2N4401. Both big dies good for <500mA saturated switching,
hFE ~100, fT ~ 200MHz, etc.

Tim

 

[Eeyore]

Wazzat? To the best of my recollection, there's no difference between the
2222 and the 2N4401. Both big dies good for <500mA saturated switching,
hFE ~100, fT ~ 200MHz, etc.

Indeed fine for switching but not the best amplifier transistor which is where
you find it used typically.

Graham

 

[Michael A. Terrell]

Actually, the 2n2222 was a metal-can TO-18 part, right?

Yes, the original version was metal. It's an old part.

Not that it matters. Fab lines and therefore some BJT
die designs have dramatically changed with time. If you
check Fairchild's lineup you'll see many '2222 variants,
no doubt positioned to grab sales from whatever p/n the
buyer's BOM happens to have. Fairchild may also make
'2222 parts at more than one fab, since they purchased
or aligned with several other manufacturers.

The 'A' was the TO-92 version, wasn't it? I don't have my older
reference books handy, today.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[qrk]

In the circuit, it flat lines at what I assume is it's 15 ma
Slew/Skew point of the ramp signal on the output. In circuit you don't
see the 100k..150khz signal, you just see roll over instead of a linear
ramp. At least for that time period. I contribute this to the parasitic
resistors and caps I put in that is removing the actual oscillation and
just producing this effect. This lead me to think there was something
wrong in the transistor.

It's used as part of a bipolar output to drive a magnetic coil for
a beam steering system to adjust the focus point of an irradiation unit.

I set up a simple common emitter test (DC), using a load resistor
on the collector from a 15 volt source. Just biased the BASE via
a 10k resistor from a variable trimmer reference, Place a by pass
cap (.1) from the base to common. at about 15 ma's, You can see this
OSC at the collector. In the real circuit the collector is connected to
the + rail and the emitter is driving the next stage via an R.
I used 3 different scopes to verify this, I found my 350 Mhz Tek give
me the clearest reading . All 3 scopes show the problem at the same
place. The other two were cheap digital LCD's.

I don't get oscillation at the final output because of rest of the
circuit, I just get ramp level shift at ~ 15 ma's, and it's because the
transistor is some how OSC at that point. It Almost act's like a tunnel
diode effect.


Maybe, the other vender's worked fine, as well as the old stand by
plastic type 2222's
Good point how ever, Since I strictly used a DC test circuit with no
inductive components, I don't see this happening.
The test circuit was a simple DC test jig.

Thanks

As to your comment at the end, your strictly DC test circuit has
"stray" inductive and capacitive components which come from
interconnect wires and the device. Can you post a schematic of this
stage showing the driving stage and what this is driving? A picture of
your test set up can be informative. Have you tried sticking a 47 ohm
resistor in series with the base? Is your power supply bypassed near
the circuit? As Wim stated, if you're using a digital scope, be sure
its not aliasing. Digital scopes are notorious for giving wrong
information when you are dealing with unknown signals.

Mark

 

[Eeyore]

Yes, the original version was metal. It's an old part.

NO.

The 2N2222 is a TO-18 part. The whole point of a JEDEC spec is that it's a
*standard*. It doesn't change willy nilly.

The 'A' was the TO-92 version, wasn't it? I don't have my older
reference books handy, today.

NO.

The *PN*2222 is TO-92.

Graham

 

[Eeyore]

The 'A' was the TO-92 version, wasn't it? I don't have my older
reference books handy, today.

You do apparently have a computer and an internet connection though.

Ever heard of this thing called google ? Or you could even go to a semiconductor
manufacturer's website !

Graham

 

[Jamie]

MMBT2222 is On Semiconductor, formerly Motorola. The '2222' makes it
the surface mount version of the 2N2222A TO-92 NPN silicon transistor.

We got a reply back from our supplier today, Fairchild verified a
defect that I found. Guess I wasn't the only one. Our supplier is going
to make good on it.. They're getting new stock. where? I don't know and
don't care. Long as they're good.

Mean while, i'll used the leftovers I found in my junk drawer made
by ON. They work fine, as did all the other non-smt type I had around
here.

 

[Robert Baer]

Winfield wrote:




And the 2N2222A.





Aside from the huge cost difference. Hermetic metal can devices aren't cheap.





The TO-92 version was the PN2222. All have very low current gain by modern
standards which is one reason I'm forever amazed that they still feature in
cookbooks.

Graham

Hermetic metal can devices seem to be useable to at least 200C...

 

[Eeyore]

Hermetic metal can devices seem to be useable to at least 200C...

Generally so.

I can't see much real advantage of that in the case of the 2N2222 though. Back when
it was introduced, hermetic cans were used because the plastics/epoxies of the day
weren't yet up to the task for semiconductor encapsulation.

Graham

 

[Tom Bruhns]

http://www.fairchildsemi.com/ds/MM/MMBT2222.pdf

We decided to condense an older design using SM components
so that we could squeeze this board into a small area.

Got 20 of these from fairchild. after initial testing
of the first one. We found some spurious noise being
generated only at fixed levels of operation.

After a little testing, we decided to put a MMBTB2222 in a
simple common emitter with resistor load on the collector circuit,
using a variable bias, we found at ~ 15 ma of Ice, there
was ~ 0.5 vpp erratic oscillation in the area of 100k..150khz depending
on where the bias was. below or above this point would then place the
transistor back in normal operation once again.
Testing with a 2N2222 operated fine, I also found another SMT type
2222 which tested fine.

We'll be looking for another semi vender with this same number, I
don't want to buy another batch of this number from fairchild just in
case I get the same defects.

Have any of you ever seen strange things like this take place?

P.S.
It works fine being used in a saturated switch circuit obviously.

I haven't seen this possibility mentioned in others' postings--may
have just missed it:

When I see "erratic behavior" mentioned (or observe it myself) I
immediately think of oscillations at frequencies beyond the capability
of the 'scope being used to look at the circuit. A '2222 could easily
be oscillating in the couple-hundred MHz region or somewhat higher,
and if it is, you can get all sorts of apparently strange things going
on in an otherwise simple circuit.

Someone else mentioned trying some other brands, I believe. That
seems like a good idea, too, at least for comparison.

We use lots of MMBT2222's, and I haven't seen any notes about
strangenesses in them; not sure if Fairchild is an approved vendor for
us for that particular part though.

Cheers,
Tom

 

[Michael A. Terrell]

When I see "erratic behavior" mentioned (or observe it myself) I
immediately think of oscillations at frequencies beyond the capability
of the 'scope being used to look at the circuit. A '2222 could easily
be oscillating in the couple-hundred MHz region or somewhat higher,
and if it is, you can get all sorts of apparently strange things going
on in an otherwise simple circuit.

Someone else mentioned trying some other brands, I believe. That
seems like a good idea, too, at least for comparison.

We use lots of MMBT2222's, and I haven't seen any notes about
strangenesses in them; not sure if Fairchild is an approved vendor for
us for that particular part though.

Cheers,
Tom

Motorola/On was our only approved source.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[Winfield]

I haven't seen this possibility mentioned in others' postings--may
have just missed it:

When I see "erratic behavior" mentioned (or observe it myself)
I immediately think of oscillations at frequencies beyond the
capability of the 'scope being used to look at the circuit.
A '2222 could easily be oscillating in the couple-hundred MHz
region or somewhat higher, and if it is, you can get all sorts of
apparently strange things going on in an otherwise simple circuit.

Someone else mentioned trying some other brands, I believe.
That seems like a good idea, too, at least for comparison.

We use lots of MMBT2222's, and I haven't seen any notes about
strangenesses in them; not sure if Fairchild is an approved
vendor for us for that particular part though.

Cheers,
Tom

Yes, I agree, high-freq RF oscillation. I'd lay the blame
to the simple wiring of Jamie's test fixture, which should
have one or two judiciously-placed ferrite beads. He gets
a transistor with slightly higher fT and lower capacitance,
which is normally considered a better transistor, and calls
it defective. I see he tells us Fairchild is replacing the
defective shipment. It's good when a supplier jumps to say
the customer is right, but sometimes they may be wrong.