How does the gain of a transistor go down ... ?

[Arfa Daily]

Hartke AH500 on the bench this morning. It struggled to come out of protect
from cold, but if it did make it, then it worked ok, and mostly stayed on
producing clean power of the right order, although the owner had complained
that it had cut out during a gig on more than one occasion.

As the thing powered up there was a small(ish) DC offset of a little under
negative 2 volts at the output stage midpoint, and it was this that the
protect circuit was rightly objecting to. With + / - 90 volts across the
output stage, it doesn't take too much imbalance to produce a 2 volt offset
....

As it warmed up, the amount of offset varied a bit, so I went in with a can
of freezer to see if I could see anything that was particularly sensitive.
That led me back right to the front end where one transistor in a
long-tailed pair had a significant effect when sprayed, the offset rising to
around negative 4 volts. So I pulled both out and checked them on my cheapo
component tester. Both correctly registered as NPN transistors, and the BCE
pinning agreed. Both had 645 mV B-E voltage, but one had a beta of 215, and
the other, just 35.

A new pair of matched transistors had the offset down to a few mV, and the
protect circuit was happy with that. But it got me to wondering what could
be the failure mechanism that resulted in a transistor still being a
recognisable transistor in that the tester still saw it as one, and it still
basically worked in the amp, but had a very low gain ?

Arfa

 

[N_Cook]

On 02/04/2014 11:41, Arfa Daily wrote:

Hartke AH500 on the bench this morning. It struggled to come out of
protect from cold, but if it did make it, then it worked ok, and mostly
stayed on producing clean power of the right order, although the owner
had complained that it had cut out during a gig on more than one occasion.

As the thing powered up there was a small(ish) DC offset of a little
under negative 2 volts at the output stage midpoint, and it was this
that the protect circuit was rightly objecting to. With + / - 90 volts
across the output stage, it doesn't take too much imbalance to produce a
2 volt offset ....

As it warmed up, the amount of offset varied a bit, so I went in with a
can of freezer to see if I could see anything that was particularly
sensitive. That led me back right to the front end where one transistor
in a long-tailed pair had a significant effect when sprayed, the offset
rising to around negative 4 volts. So I pulled both out and checked them
on my cheapo component tester. Both correctly registered as NPN
transistors, and the BCE pinning agreed. Both had 645 mV B-E voltage,
but one had a beta of 215, and the other, just 35.

A new pair of matched transistors had the offset down to a few mV, and
the protect circuit was happy with that. But it got me to wondering what
could be the failure mechanism that resulted in a transistor still being
a recognisable transistor in that the tester still saw it as one, and it
still basically worked in the amp, but had a very low gain ?

Arfa

What did putting the junctions on an ohmeter show?
Metal migration over time and going ohmic and thence gain drop is one
scenario

 

[Phil Allison]

"Arfa Daily has No Idea how STUPID he is,. "

A new pair of matched transistors had the offset down to a few mV, and the
protect circuit was happy with that. But it got me to wondering what could
be the failure mechanism that resulted in a transistor still being a
recognisable transistor in that the tester still saw it as one, and it
still basically worked in the amp, but had a very low gain ?

** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Think - wok too hot, not enough Soy or stale prawns in the chop suey.

Wot a fucking IDIOT question.





.... Phil

 

[Adrian Tuddenham]

Arfa Daily <arfa.daily@ntlworld.com> wrote:

...one transistor in a
long-tailed pair had a significant effect when sprayed,
[...]
Both had 645 mV B-E voltage, but one had a beta of 215, and
the other, just 35.

A new pair of matched transistors had the offset down to a few mV, and the
protect circuit was happy with that. But it got me to wondering what could
be the failure mechanism that resulted in a transistor still being a
recognisable transistor in that the tester still saw it as one, and it still
basically worked in the amp, but had a very low gain ?

Were there any diodes to protect the BE junctions against reverse
breakdown?

In a long tail pair operating from relatively high voltage supplies, it
is quite possible for more than -5vto appear across the BE junctions if
the designer hasn't built in some protection. Sometimes this can occur
during start-up, or under signal overload conditions, or if there is any
tendency to oscillate with certain loads. A 'grid stopper' resistor
will limit the breakdown current but will not prevent the junction from
being gradually destroyed by repeated reverse breakdown.



--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

 

[William Sommerwerck]

"Arfa Daily" wrote in message news:HPR_u.304386$G21.253990@fx18.am4...

Both correctly registered as NPN transistors, and the BCE pinning agreed.
Both had 645 mV B-E voltage, but one had
a beta of 215, and the other, just 35.

A difference in the thickness of the base region, perhaps? This would change
the beta, but would have little effect on the B-E voltage.

 

[Arfa Daily]

"Phil Allison" <phil_a@tpg.com.au> wrote in message
news:bq2cirFdirhU1@mid.individual.net...

"Arfa Daily has No Idea how STUPID he is,. "


A new pair of matched transistors had the offset down to a few mV, and
the protect circuit was happy with that. But it got me to wondering what
could be the failure mechanism that resulted in a transistor still being
a recognisable transistor in that the tester still saw it as one, and it
still basically worked in the amp, but had a very low gain ?


** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Think - wok too hot, not enough Soy or stale prawns in the chop suey.

Wot a fucking IDIOT question.





... Phil

Well then, what a fucking idiot twat, as ever, you are to bother replying
....

Arfa

 

[David Platt]

In article <MPG.2da66547ae8296b9989856@news.eternal-september.org>,
Maynard A. Philbrook Jr. <jamie_ka1lpa@charter.net> wrote:

A new pair of matched transistors had the offset down to a few mV, and the
protect circuit was happy with that. But it got me to wondering what could
be the failure mechanism that resulted in a transistor still being a
recognisable transistor in that the tester still saw it as one, and it still
basically worked in the amp, but had a very low gain ?

Arfa

zenering the emitter will over time, kill Beta.

It might be worth considering the schematic, and seeing what the amp's
power-up and power-down behavior is like. Depending on how the rails
bounce around, and on what's plugged into the input, the input side of
the differential pair might end up being zenered in this way.

Adding some B-E clamp diodes might not be a bad idea.

 

[Maynard A. Philbrook Jr.]

In article <HPR_u.304386$G21.253990@fx18.am4>, arfa.daily@ntlworld.com
says...

Hartke AH500 on the bench this morning. It struggled to come out of protect
from cold, but if it did make it, then it worked ok, and mostly stayed on
producing clean power of the right order, although the owner had complained
that it had cut out during a gig on more than one occasion.

As the thing powered up there was a small(ish) DC offset of a little under
negative 2 volts at the output stage midpoint, and it was this that the
protect circuit was rightly objecting to. With + / - 90 volts across the
output stage, it doesn't take too much imbalance to produce a 2 volt offset
...

As it warmed up, the amount of offset varied a bit, so I went in with a can
of freezer to see if I could see anything that was particularly sensitive.
That led me back right to the front end where one transistor in a
long-tailed pair had a significant effect when sprayed, the offset rising to
around negative 4 volts. So I pulled both out and checked them on my cheapo
component tester. Both correctly registered as NPN transistors, and the BCE
pinning agreed. Both had 645 mV B-E voltage, but one had a beta of 215, and
the other, just 35.

A new pair of matched transistors had the offset down to a few mV, and the
protect circuit was happy with that. But it got me to wondering what could
be the failure mechanism that resulted in a transistor still being a
recognisable transistor in that the tester still saw it as one, and it still
basically worked in the amp, but had a very low gain ?

Arfa

zenering the emitter will over time, kill Beta.

Jamie

 

[Arfa Daily]

"David Platt" <dplatt@coop.radagast.org> wrote in message
news:joeu0b-i81.ln1@coop.radagast.org...

In article <MPG.2da66547ae8296b9989856@news.eternal-september.org>,
Maynard A. Philbrook Jr. <jamie_ka1lpa@charter.net> wrote:

A new pair of matched transistors had the offset down to a few mV, and
the
protect circuit was happy with that. But it got me to wondering what
could
be the failure mechanism that resulted in a transistor still being a
recognisable transistor in that the tester still saw it as one, and it
still
basically worked in the amp, but had a very low gain ?

Arfa

zenering the emitter will over time, kill Beta.

It might be worth considering the schematic, and seeing what the amp's
power-up and power-down behavior is like. Depending on how the rails
bounce around, and on what's plugged into the input, the input side of
the differential pair might end up being zenered in this way.

Adding some B-E clamp diodes might not be a bad idea.

OK. All good stuff. There appears to be neither clamp diodes nor current
limiter Rs in the bases. That said, it's quite an old design, and this amp
is quite an old example that's been working ok to date, and is working ok
again now, so I'm reluctant to start modifying it by adding R into the
circuit, but reverse diodes across the B-E junctions might be worth doing.

Cheers all for the insights and suggestions.

Arfa

 

[Phil Allison]

"Arfa Daily" >

"Phil Allison"

"Arfa Daily has No Idea how STUPID he is,. "


A new pair of matched transistors had the offset down to a few mV, and
the protect circuit was happy with that. But it got me to wondering what
could be the failure mechanism that resulted in a transistor still being
a recognisable transistor in that the tester still saw it as one, and it
still basically worked in the amp, but had a very low gain ?


** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Think - wok too hot, not enough Soy or stale prawns in the chop suey.

Wot a fucking IDIOT question.



Well then, what a fucking idiot twat, as ever, you are to bother replying

** 100% WRONG - you fucking pommy twat.

Pointing out the dumb errors and sheer stupidity of others is a very
important, socially responsible task.



.... Phil

 

[Arfa Daily]

"Phil Allison" <phil_a@tpg.com.au> wrote in message
news:bq3qqeFn9qcU1@mid.individual.net...

"Arfa Daily"
"Phil Allison"

"Arfa Daily has No Idea how STUPID he is,. "


A new pair of matched transistors had the offset down to a few mV, and
the protect circuit was happy with that. But it got me to wondering
what could be the failure mechanism that resulted in a transistor still
being a recognisable transistor in that the tester still saw it as one,
and it still basically worked in the amp, but had a very low gain ?


** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Think - wok too hot, not enough Soy or stale prawns in the chop suey.

Wot a fucking IDIOT question.



Well then, what a fucking idiot twat, as ever, you are to bother replying


** 100% WRONG - you fucking pommy twat.

Pointing out the dumb errors and sheer stupidity of others is a very
important, socially responsible task.



... Phil

No it isn't. You're just a picky wobble-eyed psycho who always feels that he
has to demonstrate his self importance and self-believed superiority. You
probably wouldn't suffer from these delusions if you just kept popping your
meds according to your psychiatrists instructions. Now if you've got nothing
more interesting to say, fuck off back into the desert, and get back to your
other hobby of shagging kangaroos ...

Arfa

 

[Phil Allison]

"Arfa Daily has No Idea how STUPID he is,. "

** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Think - wok too hot, not enough Soy or stale prawns in the chop suey.

Wot a fucking IDIOT question.



Well then, what a fucking idiot twat, as ever, you are to bother
replying


** 100% WRONG - you fucking pommy twat.

Pointing out the dumb errors and sheer stupidity of others is a very
important, socially responsible task.


No it isn't.

** Fuck off and Die - you sad, pathetic pommy cunt.



..... Phil

 

[Arfa Daily]

"Phil Allison" <phil_a@tpg.com.au> wrote in message
news:bq6s53Fccu1U1@mid.individual.net...

"Arfa Daily has No Idea how STUPID he is,. "


** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Think - wok too hot, not enough Soy or stale prawns in the chop
suey.

Wot a fucking IDIOT question.



Well then, what a fucking idiot twat, as ever, you are to bother
replying


** 100% WRONG - you fucking pommy twat.

Pointing out the dumb errors and sheer stupidity of others is a very
important, socially responsible task.


No it isn't.

** Fuck off and Die - you sad, pathetic pommy cunt.



.... Phil

Back at ya, Philip !

Arfa

 

[Mark Zacharias]

"Arfa Daily" <arfa.daily@ntlworld.com> wrote in message
news:v_U_u.62800$8p4.45430@fx25.am4...

"Phil Allison" <phil_a@tpg.com.au> wrote in message
news:bq2cirFdirhU1@mid.individual.net...

"Arfa Daily has No Idea how STUPID he is,. "


A new pair of matched transistors had the offset down to a few mV, and
the protect circuit was happy with that. But it got me to wondering what
could be the failure mechanism that resulted in a transistor still being
a recognisable transistor in that the tester still saw it as one, and it
still basically worked in the amp, but had a very low gain ?


** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Don't know why, but see it fairly often. Seems like a preamp pair can
function with mismatched hFe up top a point but when the gain difference is
too great, DC offsets, sometimes intermittent, occur.

Just the other day had two relay drive transistors in a Marantz 2270 with an
hFe of 2. Don't know for sure how or why they failed, but I did notice black
tarnish on the leads. Perhaps migrated into the semi material?

Mark Z.

 

[Mark Zacharias]

"Arfa Daily" <arfa.daily@ntlworld.com> wrote in message
news:1co%u.566$9t2.561@fx01.am4...

"Phil Allison" <phil_a@tpg.com.au> wrote in message
news:bq3qqeFn9qcU1@mid.individual.net...

"Arfa Daily"
"Phil Allison"

"Arfa Daily has No Idea how STUPID he is,. "


A new pair of matched transistors had the offset down to a few mV, and
the protect circuit was happy with that. But it got me to wondering
what could be the failure mechanism that resulted in a transistor
still being a recognisable transistor in that the tester still saw it
as one, and it still basically worked in the amp, but had a very low
gain ?


** Obviously some kind of manufacturing defect in the particular
transistor - thousands of possibilities.

Think - wok too hot, not enough Soy or stale prawns in the chop suey.

Wot a fucking IDIOT question.



Well then, what a fucking idiot twat, as ever, you are to bother
replying


** 100% WRONG - you fucking pommy twat.

Pointing out the dumb errors and sheer stupidity of others is a very
important, socially responsible task.



... Phil


No it isn't. You're just a picky wobble-eyed psycho who always feels that
he has to demonstrate his self importance and self-believed superiority.
You probably wouldn't suffer from these delusions if you just kept popping
your meds according to your psychiatrists instructions. Now if you've got
nothing more interesting to say, fuck off back into the desert, and get
back to your other hobby of shagging kangaroos ...

Arfa

Don't know why, but see it fairly often. Seems like a preamp pair can
function with mismatched hFe up top a point but when the gain difference is
too great, DC offsets, sometimes intermittent, occur.

Just the other day had two relay drive transistors in a Marantz 2270 with an
hFe of 2. Don't know for sure how or why they failed, but I did notice black
tarnish on the leads. Perhaps migrated into the semi material?

Mark Z.

 

[Arfa Daily]

Don't know why, but see it fairly often. Seems like a preamp pair can
function with mismatched hFe up top a point but when the gain difference
is
too great, DC offsets, sometimes intermittent, occur.

Just the other day had two relay drive transistors in a Marantz 2270 with
an
hFe of 2. Don't know for sure how or why they failed, but I did notice
black
tarnish on the leads. Perhaps migrated into the semi material?

Mark Z.

Hmmm. Interesting ...

Arfa

 

[Dimitrij Klingbeil]

"Mark Zacharias" <mark_zacharias@labolgcbs.net> wrote in message
news:533fe7eb$0$2381$862e30e2@ngroups.net...

Just the other day had two relay drive transistors in a Marantz 2270 with
an
hFe of 2. Don't know for sure how or why they failed, but I did notice
black
tarnish on the leads. Perhaps migrated into the semi material?

Black tarnish on the leads... Good catch, that's probably one more of those
things to watch for (like yellow glue turning conductive and similar
horrors).

Yesterday I've had a set of 5 identical transistors, previously unused but
of
unknown age (sitting in a drawer from before my time) all fail with CB break
down around 10V (supposed to be around 50V) when I put them into a prototype
and powered the thing up. The transistors (a drawer full of them) have no
traces of being used previously (solder remains after being desoldered, bent
leads, or anything else), they look totally unsuspicious, except from a dark
tarnish on the leads. Yet they don't work. Leaky like hell, even with the
base resistively pulled down to the emitter, and up from ca. 10V they just
go
ohmic (no avalance or such thing, more like extreme leakage). The device's
whole function went haywire and took me some hours to figure out why.

So beware of transistors in TO-92 (or otherwise) with black tarnished leads.

Regards
Dimitrij

 

[whit3rd]

On Wednesday, April 2, 2014 3:41:42 AM UTC-7, Arfa Daily wrote:

As the thing powered up there was a small(ish) DC offset of a little under
negative 2 volts at the output stage midpoint... so I went in with a can
of freezer to see if I could see anything that was particularly sensitive.

That led me back right to the front end where one transistor in a
long-tailed pair had a significant effect when sprayed...
Both had 645 mV B-E voltage, but one had a beta of 215, and
the other, just 35.

The predictable aging of a transistor involves the base getting thinner (i.e. beta
rises and collector sustain voltage drops). This wouldn't imbalance a matched
pair, so the more likely culprit is chemical contamination (i.e. maybe something
as simple as a dustmote trapped in the epoxy). Probably some slight
contamination that eventually (under voltage bias) got to the sensitive
base region of the transistor and killed the carrier lifetime.

A professor once related that an assembly line for transistors suddenly
evidenced very high failure rates. Eventually the problem was traced to
a wafer wash with detergent and water. They were using Joy dishwashing
detergent, and the manufacturer had just shifted to marketing new, lemon-fresh
Joy! The manufacturing line shifted over to PhotoFlo for detergent.

 

On Wednesday, April 2, 2014 6:41:42 AM UTC-4, Arfa Daily wrote:

Hartke AH500 on the bench this morning. It struggled to come out of protect

from cold, but if it did make it, then it worked ok, and mostly stayed on

producing clean power of the right order, although the owner had complained

that it had cut out during a gig on more than one occasion.



As the thing powered up there was a small(ish) DC offset of a little under

negative 2 volts at the output stage midpoint, and it was this that the

protect circuit was rightly objecting to. With + / - 90 volts across the

output stage, it doesn't take too much imbalance to produce a 2 volt offset

...



As it warmed up, the amount of offset varied a bit, so I went in with a can

of freezer to see if I could see anything that was particularly sensitive..

That led me back right to the front end where one transistor in a

long-tailed pair had a significant effect when sprayed, the offset rising to

around negative 4 volts. So I pulled both out and checked them on my cheapo

component tester. Both correctly registered as NPN transistors, and the BCE

pinning agreed. Both had 645 mV B-E voltage, but one had a beta of 215, and

the other, just 35.



A new pair of matched transistors had the offset down to a few mV, and the

protect circuit was happy with that. But it got me to wondering what could

be the failure mechanism that resulted in a transistor still being a

recognisable transistor in that the tester still saw it as one, and it still

basically worked in the amp, but had a very low gain ?



Arfa

Arfa
As Mr. Cook had asked earlier, did you happen to look at these two that you pulled out forward and reverse with an analog ohm meter such as a 260 on the R X100 or 1K range? I would be very interested to hear of the results of that little experiment if you do. BTW, these aren't germanium are they? Lenny

 

[amdx]

On 4/2/2014 7:36 PM, Arfa Daily wrote:

"David Platt" <dplatt@coop.radagast.org> wrote in message
news:joeu0b-i81.ln1@coop.radagast.org...
In article <MPG.2da66547ae8296b9989856@news.eternal-september.org>,
Maynard A. Philbrook Jr. <jamie_ka1lpa@charter.net> wrote:

A new pair of matched transistors had the offset down to a few mV,
and the
protect circuit was happy with that. But it got me to wondering what
could
be the failure mechanism that resulted in a transistor still being a
recognisable transistor in that the tester still saw it as one, and
it still
basically worked in the amp, but had a very low gain ?

Arfa

zenering the emitter will over time, kill Beta.

It might be worth considering the schematic, and seeing what the amp's
power-up and power-down behavior is like. Depending on how the rails
bounce around, and on what's plugged into the input, the input side of
the differential pair might end up being zenered in this way.

Adding some B-E clamp diodes might not be a bad idea.


OK. All good stuff. There appears to be neither clamp diodes nor current
limiter Rs in the bases. That said, it's quite an old design, and this
amp is quite an old example that's been working ok to date, and is
working ok again now, so I'm reluctant to start modifying it by adding R
into the circuit, but reverse diodes across the B-E junctions might be
worth doing.

Cheers all for the insights and suggestions.

Arfa

I don't know the makeup of your transistor (or any), but I ran across
this picture showing electromigration, a following post says " So the
symptom was gain loss due to loss of emitter area? That's subtle; I was
looking for a short. I don't know what I was thinking. Great pictures,
by the way."
Page down about 4/5 to see the gold and orange picture with arrows
pointing to the migration.

http://www.antiqueradios.com/forums/viewtopic.php?f=8&t=196236

btw: this guy is doing some amazing reconstruction of unobtainium HP parts.
Here's one thread, http://www.antiqueradios.com/forums/vie ... p?t=190955

Mikek