Topics

Failed Transistor with Low h(FE)?


 

I'm trying to fix a DM501 that I damaged by plugging it into a faulty backplane. One of the display digits is stuck on and I think that I've narrowed the cause down to one of two transistors that drives the display chip select lines (Q348 or Q350). I removed both transistors, but when I checked them in my component tester neither one looks "blown". Q348 however (a 2N3565) has a lower hFE than the replacements I have in hand.

Here are the measured hFE for the suspect and replacement parts:

old 2N3565 (Q348): hFE=287, Vf=683
old 2N2907 (Q350): hFE=227, Vf=667

new 2N3565 (two tested): hFE=(443, 343), Vf=(673, 645)
new 2N2907 (one tested): hFE=288, Vf=673

I was expected to find the failed part as a short or open. Could a failed transistor simply have a decreased gain?

-- Jeff Dutky


Ed Breya
 

Better look for different suspects to blame for the failure. Those Qs are probably just fine. To be sure, just install the new ones and see if that fixes it. I doubt it. Ed


Jared Cabot
 

Pending further testing of the transistors, I wouldn't expect those hFE numbers to be indicative of a faulty part unless they were single digits or up in the thousands, ie. orders of magnitude away from where they should be.


Ozan
 

I agree with others hFE's measured look OK if the transistor is healthy otherwise.

To rule out anything wrong with U330:
When you look at pin 17 of U330 with a scope do you see 250us high (>1V) , 750us low (<400mV)?

To check if Q348 is doing its job:
When you look at base of Q350 do you see 250us of ~ 4.3V and 750us of ~ 5V?

Ozan


Ozan
 

Correction: Off time should be 1ms, not 750us according to SM.

On Thu, Feb 18, 2021 at 10:40 AM, Ozan wrote:


I agree with others hFE's measured look OK if the transistor is healthy
otherwise.

To rule out anything wrong with U330:
When you look at pin 17 of U330 with a scope do you see 250us high (>1V) ,
1ms (was 750us) low (<400mV)?

To check if Q348 is doing its job:
When you look at base of Q350 do you see 250us of ~ 4.3V and 1ms (was 750us) of ~ 5V?

Ozan



 

Ozan,

I did not have an extender until a few days ago, and had not checked any electrical signals in operation. I will do that now.

It also occurred to me that I can swap the pins going to the display board and see if the malfunction follows the pin swap (this would rule out the display element).

-- Jeff Dutky


Ozan
 

I did not have an extender until a few days ago, and had not checked any
electrical signals in operation. I will do that now.
Got it, it is one of those units that need an extender to probe inside.


It also occurred to me that I can swap the pins going to the display board and
see if the malfunction follows the pin swap (this would rule out the display
element).
When you removed Q350 if the problematic digit turned off, it says only path is through Q350. As an interesting experiment you can try putting the decimal point on the problematic segment by changing scale (i.e. close S10-36) and check if decimal point shows up in all segments. If it does pin 17 is stuck high.

Ozan


 

Hi Jeff,
Transistors do not have a "specific" hFE. Look at the datasheets for the 2N3565 and the 2N2907 transistors. For that matter, you can look at the datasheet for ANY transistors. The datasheets always show a Min(imum), Nom(inal), and Max(imum), or at least two of these three values, for the transistor's hFE.

If you look a little closer at a transistor datasheet you will notice that those Min / Non / Max values are specified at a SPECIFIC collector current. The reason manufacturers specify the collector current they measured the hFE at is because the hFE can vary widely as the collector current is changed. They may also specify the temperature because hFE varies as the temperature of the transistor changes. Unless otherwise noted the temperature is usually room temperature (25C).

Circuit Design Engineers know all of this so they never design a circuit using the manufacturer's maximum hFE figures because there is always the chance they will turn out to be 1/5th of that value under actual operating conditions. Instead they use many different forms of feedback to insure the circuit is not dependent on a transistor's hFE, but on other more stable circuit components.

When you see an hFE over 100 in most cases it no longer matters whether it is 200, or 400, or 600. An hFE over 100 is excellent and no designer would design this circuit assuming all the transistors would have an hFE of at 400. The ONLY EXCEPTION is when transistors have to be hand selected to meet some specific criteria that a circuit requires.

The Fairchild 2N3565 datasheet specifies hFE as 120 min to 600 max. I will bet the circuit they are in will work just fine with any transistor with an hFE over 50.

Dennis Tillman W7pF

-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of Jeff Dutky
Sent: Thursday, February 18, 2021 7:57 AM
To: TekScopes@groups.io
Subject: [TekScopes] Failed Transistor with Low h(FE)?

I'm trying to fix a DM501 that I damaged by plugging it into a faulty backplane. One of the display digits is stuck on and I think that I've narrowed the cause down to one of two transistors that drives the display chip select lines (Q348 or Q350). I removed both transistors, but when I checked them in my component tester neither one looks "blown". Q348 however (a 2N3565) has a lower hFE than the replacements I have in hand.

Here are the measured hFE for the suspect and replacement parts:

old 2N3565 (Q348): hFE=287, Vf=683
old 2N2907 (Q350): hFE=227, Vf=667

new 2N3565 (two tested): hFE=(443, 343), Vf=(673, 645) new 2N2907 (one tested): hFE=288, Vf=673

I was expected to find the failed part as a short or open. Could a failed transistor simply have a decreased gain?

-- Jeff Dutky







--
Dennis Tillman W7pF
TekScopes Moderator


Chuck Harris <cfharris@...>
 

On hfe and other transistor parameters:

What Dennis says is 100% true. But there is a little
something that you must remember for early manufacture,
and early uses of transistors: They were very, very
expensive! I can remember paying $20 each for some of
the early silicon transistors in the 1960's.... Gas was
$0.20/gallon back then, to give you some perspective...

So, in the early days, companies that made transistors
hand selected the transistors coming off the line, and
binned them into different ranges for hfe, ft, or
leakage... all depending on what customers might want.

We used to see transistors of a given type with different
color paint dots on them, and later we saw "families"
of JEDEC numbered transistors, like 2N3903/2N3904, or,
2N3905/2N3906...

And, because they were expensive, designers tried their
level best to use as few transistors as they could get
away with in a given design... especially in consumer
equipment. So, they would often use the transistors a lot
closer to their hfe than one might wish, and use the
transistors that selected less good in less important
stages.

Companies like tektronix didn't ascribe so much to this
philosophy, but still, they graded bins of common transistors
and gave them different part numbers based on different
parameters.

When all was said and done, and the best, the OK, and the
good enough transistors were all selected out and sold
into industry, there was PolyPaks and RadioShack.

So yes, it is common to use a part with an hfe range of
say 50 to 150 in a stage with feedback set to a gain of 10,
but it used to be quite common to use such a transistor
really close to the wall, with a gain of 100, and expect
to do some swapping when it didn't quite work out.

When you see a transistor, in a common emitter stage,
without an emitter resistor to ground, you can be pretty sure
that it is running full tilt... with as much hfe as it can
give.

-Chuck Harris

Dennis Tillman W7pF wrote:

Hi Jeff,
Transistors do not have a "specific" hFE. Look at the datasheets for the 2N3565 and the 2N2907 transistors. For that matter, you can look at the datasheet for ANY transistors. The datasheets always show a Min(imum), Nom(inal), and Max(imum), or at least two of these three values, for the transistor's hFE.

If you look a little closer at a transistor datasheet you will notice that those Min / Non / Max values are specified at a SPECIFIC collector current. The reason manufacturers specify the collector current they measured the hFE at is because the hFE can vary widely as the collector current is changed. They may also specify the temperature because hFE varies as the temperature of the transistor changes. Unless otherwise noted the temperature is usually room temperature (25C).

Circuit Design Engineers know all of this so they never design a circuit using the manufacturer's maximum hFE figures because there is always the chance they will turn out to be 1/5th of that value under actual operating conditions. Instead they use many different forms of feedback to insure the circuit is not dependent on a transistor's hFE, but on other more stable circuit components.

When you see an hFE over 100 in most cases it no longer matters whether it is 200, or 400, or 600. An hFE over 100 is excellent and no designer would design this circuit assuming all the transistors would have an hFE of at 400. The ONLY EXCEPTION is when transistors have to be hand selected to meet some specific criteria that a circuit requires.

The Fairchild 2N3565 datasheet specifies hFE as 120 min to 600 max. I will bet the circuit they are in will work just fine with any transistor with an hFE over 50.

Dennis Tillman W7pF

-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of Jeff Dutky
Sent: Thursday, February 18, 2021 7:57 AM
To: TekScopes@groups.io
Subject: [TekScopes] Failed Transistor with Low h(FE)?

I'm trying to fix a DM501 that I damaged by plugging it into a faulty backplane. One of the display digits is stuck on and I think that I've narrowed the cause down to one of two transistors that drives the display chip select lines (Q348 or Q350). I removed both transistors, but when I checked them in my component tester neither one looks "blown". Q348 however (a 2N3565) has a lower hFE than the replacements I have in hand.

Here are the measured hFE for the suspect and replacement parts:

old 2N3565 (Q348): hFE=287, Vf=683
old 2N2907 (Q350): hFE=227, Vf=667

new 2N3565 (two tested): hFE=(443, 343), Vf=(673, 645) new 2N2907 (one tested): hFE=288, Vf=673

I was expected to find the failed part as a short or open. Could a failed transistor simply have a decreased gain?

-- Jeff Dutky







 

Dennis, Chuck, et-al,

thank you for the input and elucidation, it has been quite helpful. Unfortunately it appears that my display board, not the transistors that drive the select lines, that is the damaged component, and the damage affects all three low order digits.

-- Jeff Dutky


Roy Thistle
 

On Thu, Feb 18, 2021 at 10:08 PM, Chuck Harris wrote:


tektronix didn't ascribe so much to this
philosophy, but still, they graded bins of common transistors
and gave them different part numbers based on different
parameters.
From the: Replaceable Electrical Parts-DM501
Q152 151-0192-00 TRANSISTOR: SILICON:NPN,SEL FROM MPS6521 80009 151-0192-00,

This says for Q153, "SEL FROM MPS6521," ... which means to me ... select the "right" MPS6521, from a pile of MPS6521.
The MPS6521 had a maximum Hfe of 600 (according to the Fairchild Discrete databk) ... so did/could that mean that Tektronix sorted the MPS6521 they got from Fairchild? ... for, the best gain? ( It's 80009, or a Tek part.)

I don't see a 2N2907 in the DM501 parts list.
But there are some 2N2907A ... which indeed has a significantly higher Hfe than the 2N2907 (according to the Motorola small signal databk) ... and so since they came from the same Motorola process? ... was it Motorola binning parts, in this case?


Chuck Harris <cfharris@...>
 

When tektronix said a part was selected, they didn't usually
share with us why.

I would expect that the listing meant the technician was to
try out transistors from the bin until one worked... for whatever
reason. That is a short sighted self defeating process, as the
bin will become more and more concentrated with parts that won't
work...

Back in the 1960's, when most of the small signal transistors
were originally designed, the processes were not as tightly
controlled as they are today. It was not unusual to get a
whole run of parts that didn't quite make the grade.

They sorted those different grade parts out, and gave them
different numbers, like the 2N3903/3904, example I gave in my
earlier post... though I think those are different voltage
capabilities.

Usually, when a part got an "A" version, that meant a process
change occurred. Often a reduction in process size, or maybe
an enhancement in some aspect of the design.

Selection of parts is a particular bugaboo of mine.

I generally buy resistors in bulk, thousands of a value at a
time, usually 5%. Occasionally, I need a 1% value and have
tried to select such a part our of my 5% parts. What I generally
find is that out of 1000's of 5% resistors, not a single one
was within 1% of its nominal value. But, all were within 5%.

Pretty unlikely unless someone at the factory beat me to all
of the 1% parts.

-Chuck Harris

Roy Thistle wrote:

On Thu, Feb 18, 2021 at 10:08 PM, Chuck Harris wrote:


tektronix didn't ascribe so much to this
philosophy, but still, they graded bins of common transistors
and gave them different part numbers based on different
parameters.

From the: Replaceable Electrical Parts-DM501
Q152 151-0192-00 TRANSISTOR: SILICON:NPN,SEL FROM MPS6521 80009 151-0192-00,

This says for Q153, "SEL FROM MPS6521," ... which means to me ... select the "right" MPS6521, from a pile of MPS6521.
The MPS6521 had a maximum Hfe of 600 (according to the Fairchild Discrete databk) ... so did/could that mean that Tektronix sorted the MPS6521 they got from Fairchild? ... for, the best gain? ( It's 80009, or a Tek part.)

I don't see a 2N2907 in the DM501 parts list.
But there are some 2N2907A ... which indeed has a significantly higher Hfe than the 2N2907 (according to the Motorola small signal databk) ... and so since they came from the same Motorola process? ... was it Motorola binning parts, in this case?






Ed Breya
 

Chuck wrote: " What I generally
find is that out of 1000's of 5% resistors, not a single one
was within 1% of its nominal value. But, all were within 5%."

These two-lobed distributions are common. The ones that fell in the middle range near 1% were likely separated out and sold as a better grade part, leaving the rest of the distribution (which should be normal) with a hole in the middle.

Ed


Siggi
 

On Fri, Feb 19, 2021 at 2:26 PM Roy Thistle <roy.thistle@mail.utoronto.ca>
wrote:

This says for Q153, "SEL FROM MPS6521," ... which means to me ... select
the "right" MPS6521, from a pile of MPS6521.
The MPS6521 had a maximum Hfe of 600 (according to the Fairchild Discrete
databk) ... so did/could that mean that Tektronix sorted the MPS6521 they
got from Fairchild? ... for, the best gain? ( It's 80009, or a Tek part.)
If you look on page 6-4 of the Common Parts Catalog, I believe it gives you
the selection criteria. I think hFE_MIN = 300 is possibly the most salient
selection criteria for this part.
Note that there are also 151-0192-01 etc parts in the catalog. Perhaps
those were selected for the same criteria?


Roy Thistle
 

On Thu, Feb 18, 2021 at 07:56 AM, Jeff Dutky wrote:


I was expected to find the failed part as a short or open. Could a failed
transistor simply have a decreased gain?
If the design is pushing the transistor into saturation... to make sure switching occurs under different operating conditions... then since under saturation, the beta drops to a fraction. Then yes... since a minimum beta is usually assumed, for reliable switch design. And if the beta of the "failed" transistor is less than (or around) that assumed minimum, there could be problems.
It's interesting to note... that research shows silicon epitaxial transistors do start to diminish in performance, with age and with operation. The thin silicon layer between the base and emitter, tends to degrade with time or operation.


shalopt
 

Yes and that is how you get the 10% values, two piles none will be with in the 5%, and a long while back the
+- 20%. It has always been this way. I have found if I buy a batch they will all be on either the + or - side but not both, seldom both sides. That old bell curve in play.


Chuck Harris <cfharris@...>
 

I have often wondered if many circuit designers
take these "holes" caused by factory robbery into
account when they specify their parts... I never
did, but perhaps the production guys would have
liked me better if I had?

You think you are using a 1K nominal part, +/-5%,
but in reality, your part is really:
[1.05K|0.95K] +/-2.5% of 1K

You thought you bought a one humped dromedary camel,
but Amil the camel dealer came rolling up to your
camp with a two humped bactrain camel, and none of
your saddles will fit.

It puts a completely different spin on a Monte Carlo
analysis on part tolerances.

No wonder HP only used 1% or better parts.

-Chuck Harris


Ed Breya via groups.io wrote:

Chuck wrote: " What I generally
find is that out of 1000's of 5% resistors, not a single one
was within 1% of its nominal value. But, all were within 5%."

These two-lobed distributions are common. The ones that fell in the middle range near 1% were likely separated out and sold as a better grade part, leaving the rest of the distribution (which should be normal) with a hole in the middle.

Ed







Ed Breya
 

Chuck wrote: "No wonder HP only used 1% or better parts"

Haha - I wouldn't be too surprised if there's a hole in the middle of those too, where the 0.1 % tolerance grade parts are "pre-removed" for your convenience. It depends on how far you need to, and can, grade parts from a given process, and on the economic value of doing so.

Ed