Is there a device that can...

[Pooh Bear]

Then by all means, please point me to some sort of tutorial or best
practice guide, because the ones I've got never mentionned that kind of
setup in practical terms.

Just look at leakage current on a data sheet ( Icbo IIRC ) and find the tempco if you can !

Consider hfe range. Basically all the tolerances and worst case situations that might occur
simultaneously.

A b-e resistor is cheap insurance.

Graham

 

[Wouter van Ooijen]

One word. *Leakage* !

One more word: *nanoamps* !

OK, a few more words: worst-case, Beta, tempco.

The worst-case Ic of the PNP will be the maximum Beta * the worst-case
leakage current. With Beta's up to 1000 your (optimistic?) nanoamps
are amplified to microamps, which might just become a problem.

I tried to make a real-life check for the common BC547
(http://www.semiconductors.philips.com/acrobat_download/datasheets/BC546_547_4.pdf)
but I could not find a leackage figure, much less a tempco. Am I
blind?


Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
Webshop for PICs and other electronics
http://www.voti.nl/hvu
Teacher electronics and informatics

 

[Jim Thompson]

OK, a few more words: worst-case, Beta, tempco.

The worst-case Ic of the PNP will be the maximum Beta * the worst-case
leakage current. With Beta's up to 1000 your (optimistic?) nanoamps
are amplified to microamps, which might just become a problem.

I tried to make a real-life check for the common BC547
(http://www.semiconductors.philips.com/acrobat_download/datasheets/BC546_547_4.pdf)
but I could not find a leackage figure, much less a tempco. Am I
blind?


Wouter van Ooijen

You didn't notice Icbo ??

...Jim Thompson

 

[John Larkin]

OK, a few more words: worst-case, Beta, tempco.

The worst-case Ic of the PNP will be the maximum Beta * the worst-case
leakage current. With Beta's up to 1000 your (optimistic?) nanoamps
are amplified to microamps, which might just become a problem.

I tried to make a real-life check for the common BC547
(http://www.semiconductors.philips.com/acrobat_download/datasheets/BC546_547_4.pdf)
but I could not find a leackage figure, much less a tempco. Am I
blind?

I'm not qualified to evaluate your eyesight.

If you don't have data, or the gadget in question is not
cost-sensitive, it won't hurt to always add b-e resistors. But in
cost-sensitive situations where the numbers are known, one can omit
them.

Beta tends to fall off severely at nanoamp currents, and many loads
could tolerate microamps of leakage. The obligatory use of b-e
resistors is probably a holdover from germanium days, where a base
bias network was as likely to remove base current as it was to supply
it. A decent silicon ss transistor doesn't leak much.

I admit I usually use b-e resistors, but my gear isn't cost-sensitive.

John

 

[Wouter van Ooijen]

You didn't notice Icbo ??

Yes, but I do not recognise it as the leakage current. Maybe my folly,
I am basically a software guy. But it its conditions state Ie = 0 (no
emitter current) and a certain Vcb. I would have expected the leakage
to be specified for a Vcb = 0 and a certain Vec?


Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
Webshop for PICs and other electronics
http://www.voti.nl/hvu
Teacher electronics and informatics

 

[Jim Thompson]

Yes, but I do not recognise it as the leakage current. Maybe my folly,
I am basically a software guy. But it its conditions state Ie = 0 (no
emitter current) and a certain Vcb. I would have expected the leakage
to be specified for a Vcb = 0 and a certain Vec?


Wouter van Ooijen

Icbo = Icb with emitter OPEN... note the _uA_ with applied heat.

...Jim Thompson

 

[Wouter van Ooijen]

Icbo = Icb with emitter OPEN... note the _uA_ with applied heat.

OK. But I seldom use a transistor with open emitter. So how does this
Icbo figure relate to CE leakage in a realistic situation?


Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
Webshop for PICs and other electronics
http://www.voti.nl/hvu
Teacher electronics and informatics

 

[Winfield Hill]

Wouter van Ooijen (www.voti.nl wrote...

OK. But I seldom use a transistor with open emitter. So how does
this Icbo figure relate to CE leakage in a realistic situation?

beta times higher, applicable if the base is open (a more common
situation, especially without any base-emitter resistor). The
resulting Ic can be surprisingly high, at high temperatures.

 

[Ban]

OK. But I seldom use a transistor with open emitter. So how does this
Icbo figure relate to CE leakage in a realistic situation?

when you draw a simple model of a transistor it shows two diodes with
kathodes or anodes together (PNP/NPN), just like a double diode. now you can
leave one leg open without problems of understanding. The collector/base
diode is biased inversly and the leakage current is measured against
temperature and bias voltage. When you connect the emitter to gnd this
current will flow into the other diode and forward bias it.
As Win writes, now this current will be multiplied with hfe and cause a
collector current that might be substancial. This happens when the base is
isolated by hiZ.

 

[Wouter van Ooijen]

Thanks Wilfried and Ban.

So now for the real calculation: assume a BC547 and BC557. BC547 Icbo
is 15nA at 30C, 5uA (!) at 150C. Bmax is 800. BC557 Bmax also 800. So
at 30C the resulting leakage-caused Ic of the BC557 (PNP) would be
15nA * 800 * 800 = 9.6 mA. At 150C the BC557 would probably be in
saturation.

Now some of you will probably rise up and shout that you have never
seen this in real life. That is possible. If you can live with the
'statistical certianty' that you will never combine two of these
transistors that happen to have the worst case leakage and the (for
this case) worst case Beta, OK, your choice.


Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
Webshop for PICs and other electronics
http://www.voti.nl/hvu
Teacher electronics and informatics

 

[Ian Stirling]

Thanks Wilfried and Ban.

So now for the real calculation: assume a BC547 and BC557. BC547 Icbo
is 15nA at 30C, 5uA (!) at 150C. Bmax is 800. BC557 Bmax also 800. So
at 30C the resulting leakage-caused Ic of the BC557 (PNP) would be
15nA * 800 * 800 = 9.6 mA. At 150C the BC557 would probably be in
saturation.

Now some of you will probably rise up and shout that you have never
seen this in real life. That is possible. If you can live with the
'statistical certianty' that you will never combine two of these
transistors that happen to have the worst case leakage and the (for
this case) worst case Beta, OK, your choice.

Kind of depends.

If it's a fish tank level monitor, immersed in the tank, performance
at 150C may be slightly less than relevant.
Many consumer goods will have largely melted at 150C.

Indeed, design to worst cases, but it's probably worth actually thinking
about what the worst case might possibly be, rather than assuming 150C,
and a million rads for everything.

 

[Wouter van Ooijen]

So now for the real calculation: assume a BC547 and BC557. BC547 Icbo

Kind of depends.

If it's a fish tank level monitor, immersed in the tank, performance
at 150C may be slightly less than relevant.
Many consumer goods will have largely melted at 150C.

Indeed, design to worst cases, but it's probably worth actually thinking
about what the worst case might possibly be, rather than assuming 150C,
and a million rads for everything.

The above calculation which results in 9.6mA worst case is for 30C.


Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
Webshop for PICs and other electronics
http://www.voti.nl/hvu
Teacher electronics and informatics

 

[Frank Bemelman]

The above calculation which results in 9.6mA worst case is for 30C.

With the bc547 base floating? What if there is a small base resistor to
ground, on the first NPN, bc547?

 

[Ban]

With the bc547 base floating? What if there is a small base resistor
to ground, on the first NPN, bc547?

Frank, that's where the thread is about, read Wins replies a bit back.
Your resistor will eliminate the current by factor 800, but still the
leakage from the c/b-diode of the 2nd transistor will cause some collector
current. But now also a part will be leaking through the e/b-junction of the
1st transistor, so it will be even lower. It will need a second resistor to
gnd from the base of T2 to eliminate this effect completely. Many
darlingtons have it inbuilt..

 

[Wouter van Ooijen]

With the bc547 base floating? What if there is a small base resistor to

ground, on the first NPN, bc547?

The whole discussion was about not using such resistors.

I guess using a BE resistor on the NPN would reduce the Ic of the NPN
to the figure in the datasheet, so the Ic of the PNP would be reduced
to 12uA at 30C, 5 mA at 150C. That might be acceptable in some (most?)
situations. But do might want to check the Icbo / temperature
relation.


Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
Webshop for PICs and other electronics
http://www.voti.nl/hvu
Teacher electronics and informatics

 

[Frank Bemelman]

Frank, that's where the thread is about, read Wins replies a bit back.
Your resistor will eliminate the current by factor 800, but still the
leakage from the c/b-diode of the 2nd transistor will cause some collector
current. But now also a part will be leaking through the e/b-junction of the
1st transistor, so it will be even lower. It will need a second resistor to
gnd from the base of T2 to eliminate this effect completely. Many
darlingtons have it inbuilt..

Yes, but I was under the impression it was about leaving out the resistor
between e-b of the top PNP transistor. Since the drive for the lower
NPN has to come from somewhere, a push-pull logic gate perhaps, or something
that has some resistance to ground anyway, when in off condition.

Wouter calculated what may happen with the base of the lower NPN floating,
which is perhaps not something that will happen in practice.

Not all that important anyway.

 

[Frank Bemelman]

The whole discussion was about not using such resistors.

I assumed leaving out only the resistor between e-b of the
upper PNP. You would still need one resistor between collector
of the lower NPN and the base of the upper PNP, to limit the
base current for the PNP.

I guess using a BE resistor on the NPN would reduce the Ic of the NPN
to the figure in the datasheet, so the Ic of the PNP would be reduced
to 12uA at 30C, 5 mA at 150C. That might be acceptable in some (most?)
situations. But do might want to check the Icbo / temperature
relation.

That's what I meant. In practice the circuit that drives the first NPN
will have some resistance to ground. It can be a push-pull output of
a logic gate or something. If the driver is another emitter of a NPN
transistor, you get 800 * 800 * 800. You have to stop somewhere

So, I figured the thread was only about leaving out the resistor between
b-e of the upper PNP. Which does not seem too bad, if it is simple
low speed switching.

 

[Winfield Hill]

Frank Bemelman wrote...

Wouter van Ooijen wrote...

I assumed leaving out only the resistor between e-b of the
upper PNP. You would still need one resistor between collector
of the lower NPN and the base of the upper PNP, to limit the
base current for the PNP.


That's what I meant. In practice the circuit that drives the first NPN
will have some resistance to ground. It can be a push-pull output of
a logic gate or something. If the driver is another emitter of a NPN
transistor, you get 800 * 800 * 800. You have to stop somewhere

So, I figured the thread was only about leaving out the resistor between
b-e of the upper PNP. Which does not seem too bad, if it is simple
low speed switching.

There are other ways to save parts, without compromising reliability.
In case you missed my post, in one of the other forks in this thread...

Here's a classic level-shifting pass-element switch, four resistors.

.. hv in ---+--- E C ------- out
.. | B pnp switched
.. R2 |
.. | |
.. '------+
.. |
.. R4
.. |
.. C
.. logic -- R3 -+- B
.. control | E npn
.. R1 |
.. | gnd
.. gnd

Current-switched level shifting, by contrast, saves two resistors
and also has a few other advantages (there is a requirement that
the switched voltage be higher than the logic-control voltage).

.. hv in ---+-- E C ------- out
.. | B Q2 switched
.. R2 2.7k | pnp
.. | |
.. '-----+
.. |
.. C
.. logic ----- B Q1
.. control E npn
.. 5.0V |
.. R1 4.3k
.. |
.. gnd

The current provided by R1 is the desired base-drive current, plus
the base discharge current Vbe/R2. The base-drive current level is
provided independently of any changes in the switched power voltage.
The circuit is suitable for switching high voltages, for example
if Q1 is a mpsA42, up to 300V can be switched. If Q2 is a MOSFET,
chose R2/R1 to provide about 12V of gate voltage for the ON state.

.. hv in ---+-- S D -------- out
.. | G Q2 switched
.. R2 12k | MOSFET
.. | | p-channel
.. '-----+
.. |
.. C scale resistor values
.. logic ----- B Q1 according to desired
.. control E npn MOSFET switching speed
.. 5.0V |
.. R1 4.3k
.. |
.. gnd

Note, no Q1 base-emitter resistor is needed for the OFF condition,
because the logic-control line is strongly driven to ground, which
bypasses any leakage currents, and also rapidly turns off Q1.

 

[Kevin]

One thing I like to do in these circuits is add a resistor between the
collector and the base/gate of the switch transistor to protect against
cascading catastrophic failure if the bottom device fails - or more
likely I slip with the scope probe!

kevin

 

[Rich The Newsgroup Wacko]

Hey, I like that one. You just saved me 500 resistors per year
Now I have to find a way to buy the Guru a cup of coffee and a
donut...

Buy a copy of his book. ;-D