Common emitter as level switcher

[George Herold]

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\>
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?


Thanks
George H.

Oh feel free to suggest a better circuit too!

 

[Jamie]

George Herold wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?


Thanks
George H.

Oh feel free to suggest a better circuit too!
Look for "Bakers Clamp", that is a diode feed back from the

collector to the base circuit to prevent the long term cap
at the base.

Jamie

 

[krw@att.bizzzzzzzzzzzz]

On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?


Thanks
George H.

Oh feel free to suggest a better circuit too!

Tie a Schottky from the base (anode) to the collector (cathode). This will
keep the transistor out of saturation and hopefully buy you the speed you
need. Another alternative would be to use an N-FET there instead of the NPN.

 

[George Herold]

On Oct 27, 6:57 pm, George Herold <gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

         +12V--10kohm----+----Vout
                         |
                       |/
 in--5V 1MHz-1kohm-+---|   2N3904
                   |   |\
                   |     |
                   +--1k-+
                         Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!
(oops wrong title the first time.)

 

[George Herold]

On Oct 27, 7:19 pm, Jamie
<jamie_ka1lpa_not_valid_after_ka1l...@charter.net> wrote:

George Herold wrote:
First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

         +12V--10kohm----+----Vout
                         |
                       |/
 in--5V 1MHz-1kohm-+---|   2N3904
                   |   |\
                   |     |
                   +--1k-+
                         Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

Look for "Bakers Clamp", that is a diode feed back from the
collector to the base circuit to prevent the long term cap
at the base.

  Jamie- Hide quoted text -

- Show quoted text -

Excellent, Thanks Jamie.

George H.

 

[John Larkin]

On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?


Thanks
George H.

Oh feel free to suggest a better circuit too!

10K is kinda big for the collector pullup. 10K * 10 pF = 100 ns, so
even the scope probe will slow it down. Try 1K. A series inductor
might be fun, too. Try L = 0.5 * R^2 * C or thereabouts.

A small cap across the base drive 1K is a classic speedup. And make
the Thevenin voltage drive into the base more like 0,1.2 than the
existing 0,2.5, namely reduce the lower base resistor, 330 ohms maybe.

The 3904 will saturate and turn off slowly. A small mosfet, like a
2N7000, with a 1K pullup would be faster.

John

 

[krw@att.bizzzzzzzzzzzz]

On Thu, 27 Oct 2011 19:07:56 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

On Oct 27, 8:40 pm, "k...@att.bizzzzzzzzzzzz"
k...@att.bizzzzzzzzzzzz> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

Tie a Schottky from the base (anode) to the collector (cathode).  This will
keep the transistor out of saturation and hopefully buy you the speed you
need.  Another alternative would be to use an N-FET there instead of the NPN.- Hide quoted text -

- Show quoted text -

Great, I was wondering if a Fet would work better. The only charge I
have to worry about is whatever's on the gate capacitance?

And drain. The 10K pullup resistor has to charge whatever capacitance is on
the drain (+ load). You can reduce this, obviously at the cost of power.

Also, you have a 1K base resistor, which is going to give you about 3mA of
base current ((5-.7)/1K). Your collector current is only ~1mA (10V/12K). This
is going to saturate the hell out of the transistor. You might reduce the
base drive an order of magnitude (or two).

The problem is the base charge. All that charge in the base needs to flush
out through the emitter, taking Beta times as much current from the collector
with it. The Schottky keeps the transistor out of deep saturation
(technically saturation is Vbc positive) by bleeding current from the base to
the collector.

I'll try the Bakers clamp/ Schottky first since there's already an npn
'lashed' in place.

Good plan. But also try increasing the base resistor to at least 10K, maybe
50K.

using TTL circuitry seems like a nice retro touch.

That's not a bad idea either, though I'm not sure there are 1G types with 15V
or 30V outputs.

BTW, if the positive transition needs to be cleaner than the negative, turn
the whole thing upside down and change the sex of the transistor.

 

[George Herold]

On Oct 27, 8:40 pm, "k...@att.bizzzzzzzzzzzz"
<k...@att.bizzzzzzzzzzzz> wrote:

On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

Tie a Schottky from the base (anode) to the collector (cathode).  This will
keep the transistor out of saturation and hopefully buy you the speed you
need.  Another alternative would be to use an N-FET there instead of the NPN.- Hide quoted text -

- Show quoted text -

Great, I was wondering if a Fet would work better. The only charge I
have to worry about is whatever's on the gate capacitance?

I'll try the Bakers clamp/ Schottky first since there's already an npn
'lashed' in place.
using TTL circuitry seems like a nice retro touch.

George H.

 

[Jon Kirwan]

On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?


Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current. That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.) You might want to run it closer to a
beta of 20 or 30, not 0.25. Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz. So 10% turnoff isn't good for you. You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave. Have you tried a 'speed up' capacitor across your
base drive resistor? I've used them with some success. And
they go back a LONG TIME. I think before Baker clamps.

I'm not that much of a designer. But this is basics. So
that's me. Anyway, here is a shot at it.

Take your above circuit. Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta. (It is a 2N3904, after all. It can do well.) So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT. That would get near 80k
ohm, but drop it to 56k or thereabouts. At 1mA Ic, you need
about Vbe=0.65V (or less, really.) So 0.65/5*56k is around
7k pull down. But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too. This means the pull down is
about the same -- 13k-ish. I'd use 15k. Then put a speed up
cap across the 56k. I think most BJTs have small pF to worry
about. So make it a 12-15pF. Something about that size.

Then see how it goes. You won't be saturating the hell out
of it, anyway. But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon

 

[George Herold]

On Oct 27, 10:45 pm, John Larkin
<jjlar...@highNOTlandTHIStechnologyPART.com> wrote:

On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

10K is kinda big for the collector pullup. 10K * 10 pF = 100 ns, so
even the scope probe will slow it down. Try 1K.

Great that's much nicer! The base to collector zener got rid of the
100ns delay, but it was till taking another 100nS to get up to 12
Volts. (looking with a ~16pF 'scope probe)

A series inductor

might be fun, too. Try L = 0.5 * R^2 * C or thereabouts.

Ouuu.. that's getting a little too complicated. (..tune out a bit of
the C with the L.) I just wanted to add a some voltage to the clock.
And now it's a growing rat's nest on the side of the prototype.
http://imageshack.us/photo/my-images/828/dscf0043s.jpg/

A small cap across the base drive 1K is a classic speedup.

Yeah, I got that. The Bakers clamp 'name' from Jamie was golden.
There was a reference in the first editon of AoE (which lives at
home.) and the 'speed up' cap was right next to the Bakers clamp. At
work today I coudn't find this in the second edition.


And make

the Thevenin voltage drive into the base more like 0,1.2 than the
existing 0,2.5, namely reduce the lower base resistor, 330 ohms maybe.

The 3904 will saturate and turn off slowly. A small mosfet, like a
2N7000, with a 1K pullup would be faster.

Thanks George H.

John- Hide quoted text -

- Show quoted text -

 

[George Herold]

On Oct 27, 11:41 pm, "k...@att.bizzzzzzzzzzzz"
<k...@att.bizzzzzzzzzzzz> wrote:

On Thu, 27 Oct 2011 19:07:56 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:
On Oct 27, 8:40 pm, "k...@att.bizzzzzzzzzzzz"
k...@att.bizzzzzzzzzzzz> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold

gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

Tie a Schottky from the base (anode) to the collector (cathode). This will
keep the transistor out of saturation and hopefully buy you the speed you
need. Another alternative would be to use an N-FET there instead of the NPN.- Hide quoted text -

- Show quoted text -

Great, I was wondering if a Fet would work better.  The only charge I
have to worry about is whatever's on the gate capacitance?

And drain.  The 10K pullup resistor has to charge whatever capacitance is on
the drain (+ load).  You can reduce this, obviously at the cost of power.  

Also, you have a 1K base resistor, which is going to give you about 3mA of
base current ((5-.7)/1K).  Your collector current is only ~1mA (10V/12K). This
is going to saturate the hell out of the transistor.  You might reduce the
base drive an order of magnitude (or two).

The problem is the base charge.  All that charge in the base needs to flush
out through the emitter, taking Beta times as much current from the collector
with it.  The Schottky keeps the transistor out of deep saturation
(technically saturation is Vbc positive) by bleeding current from the base to
the collector.  

I'll try the Bakers clamp/ Schottky first since there's already an npn
'lashed' in place.

Good plan.  But also try increasing the base resistor to at least 10K, maybe
50K.

using TTL circuitry seems like a nice retro touch.

That's not a bad idea either, though I'm not sure there are 1G types with 15V
or 30V outputs.

BTW, if the positive transition needs to be cleaner than the negative, turn
the whole thing upside down and change the sex of the transistor.
OK, I can see that,

Thanks.
George H.

- Hide quoted text -

- Show quoted text -

George H.

 

[George Herold]

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:

On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current.  That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.)  You might want to run it closer to a
beta of 20 or 30, not 0.25.  Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz.  So 10% turnoff isn't good for you.  You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave.  Have you tried a 'speed up' capacitor across your
base drive resistor?  I've used them with some success.  And
they go back a LONG TIME.  I think before Baker clamps.

I'm not that much of a designer.  But this is basics.  So
that's me.  Anyway, here is a shot at it.

Take your above circuit.  Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta.  (It is a 2N3904, after all.  It can do well.)  So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT.  That would get near 80k
ohm, but drop it to 56k or thereabouts.  At 1mA Ic, you need
about Vbe=0.65V (or less, really.)  So 0.65/5*56k is around
7k pull down.  But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too.  This means the pull down is
about the same -- 13k-ish.  I'd use 15k.  Then put a speed up
cap across the 56k.  I think most BJTs have small pF to worry
about.  So make it a 12-15pF.  Something about that size.

Then see how it goes.  You won't be saturating the hell out
of it, anyway.  But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

 

[George Herold]

On Oct 28, 9:44 am, George Herold <gher...@teachspin.com> wrote:

On Oct 27, 10:45 pm, John Larkin





jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold

gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

10K is kinda big for the collector pullup. 10K * 10 pF = 100 ns, so
even the scope probe will slow it down. Try 1K.

Great that's much nicer!  The base to collector zener got rid of the

--oops that's a base to collector Schottky diode----

100ns delay, but it was till taking another 100nS to get up to 12
Volts.  (looking with a ~16pF 'scope probe)

A series inductor

might be fun, too. Try L = 0.5 * R^2 * C or thereabouts.

Ouuu.. that's getting a little too complicated. (..tune out a bit of
the C with the L.)  I just wanted to add a some voltage to the clock.
And now it's a growing rat's nest on the side of the prototype.http://imageshack.us/photo/my-images/828/dscf0043s.jpg/



A small cap across the base drive 1K is a classic speedup.

Yeah, I got that.  The Bakers clamp 'name' from Jamie was golden.
There was a reference in the first editon of AoE (which lives at
home.) and the 'speed up' cap was right next to the Bakers clamp.  At
work today I coudn't find this in the second edition.

And make

the Thevenin voltage drive into the base more like 0,1.2 than the
existing 0,2.5, namely reduce the lower base resistor, 330 ohms maybe.

The 3904 will saturate and turn off slowly. A small mosfet, like a
2N7000, with a 1K pullup would be faster.

Thanks George H.





John- Hide quoted text -

- Show quoted text -- Hide quoted text -

- Show quoted text -- Hide quoted text -

- Show quoted text -

 

[Jon Kirwan]

On Fri, 28 Oct 2011 07:01:34 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current.  That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.)  You might want to run it closer to a
beta of 20 or 30, not 0.25.  Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz.  So 10% turnoff isn't good for you.  You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave.  Have you tried a 'speed up' capacitor across your
base drive resistor?  I've used them with some success.  And
they go back a LONG TIME.  I think before Baker clamps.

I'm not that much of a designer.  But this is basics.  So
that's me.  Anyway, here is a shot at it.

Take your above circuit.  Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta.  (It is a 2N3904, after all.  It can do well.)  So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT.  That would get near 80k
ohm, but drop it to 56k or thereabouts.  At 1mA Ic, you need
about Vbe=0.65V (or less, really.)  So 0.65/5*56k is around
7k pull down.  But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too.  This means the pull down is
about the same -- 13k-ish.  I'd use 15k.  Then put a speed up
cap across the 56k.  I think most BJTs have small pF to worry
about.  So make it a 12-15pF.  Something about that size.

Then see how it goes.  You won't be saturating the hell out
of it, anyway.  But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

Try 100pF, then.

Jon

 

[John Larkin]

On Fri, 28 Oct 2011 07:01:34 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current.  That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.)  You might want to run it closer to a
beta of 20 or 30, not 0.25.  Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz.  So 10% turnoff isn't good for you.  You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave.  Have you tried a 'speed up' capacitor across your
base drive resistor?  I've used them with some success.  And
they go back a LONG TIME.  I think before Baker clamps.

I'm not that much of a designer.  But this is basics.  So
that's me.  Anyway, here is a shot at it.

Take your above circuit.  Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta.  (It is a 2N3904, after all.  It can do well.)  So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT.  That would get near 80k
ohm, but drop it to 56k or thereabouts.  At 1mA Ic, you need
about Vbe=0.65V (or less, really.)  So 0.65/5*56k is around
7k pull down.  But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too.  This means the pull down is
about the same -- 13k-ish.  I'd use 15k.  Then put a speed up
cap across the 56k.  I think most BJTs have small pF to worry
about.  So make it a 12-15pF.  Something about that size.

Then see how it goes.  You won't be saturating the hell out
of it, anyway.  But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

Use a mosfet! 2N3904s belong in museums next to the 6SN7s.

John

 

[krw@att.bizzzzzzzzzzzz]

On Sat, 29 Oct 2011 10:04:56 -0700, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Fri, 28 Oct 2011 07:01:34 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current.  That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.)  You might want to run it closer to a
beta of 20 or 30, not 0.25.  Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz.  So 10% turnoff isn't good for you.  You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave.  Have you tried a 'speed up' capacitor across your
base drive resistor?  I've used them with some success.  And
they go back a LONG TIME.  I think before Baker clamps.

I'm not that much of a designer.  But this is basics.  So
that's me.  Anyway, here is a shot at it.

Take your above circuit.  Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta.  (It is a 2N3904, after all.  It can do well.)  So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT.  That would get near 80k
ohm, but drop it to 56k or thereabouts.  At 1mA Ic, you need
about Vbe=0.65V (or less, really.)  So 0.65/5*56k is around
7k pull down.  But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too.  This means the pull down is
about the same -- 13k-ish.  I'd use 15k.  Then put a speed up
cap across the 56k.  I think most BJTs have small pF to worry
about.  So make it a 12-15pF.  Something about that size.

Then see how it goes.  You won't be saturating the hell out
of it, anyway.  But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

Use a mosfet! 2N3904s belong in museums next to the 6SN7s.

At 5-10x the cost, you just lost 1000 Joerg points.

 

[Michael A. Terrell]

"krw@att.bizzzzzzzzzzzz" wrote:

On Sat, 29 Oct 2011 10:04:56 -0700, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Fri, 28 Oct 2011 07:01:34 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current. That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.) You might want to run it closer to a
beta of 20 or 30, not 0.25. Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz. So 10% turnoff isn't good for you. You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave. Have you tried a 'speed up' capacitor across your
base drive resistor? I've used them with some success. And
they go back a LONG TIME. I think before Baker clamps.

I'm not that much of a designer. But this is basics. So
that's me. Anyway, here is a shot at it.

Take your above circuit. Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta. (It is a 2N3904, after all. It can do well.) So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT. That would get near 80k
ohm, but drop it to 56k or thereabouts. At 1mA Ic, you need
about Vbe=0.65V (or less, really.) So 0.65/5*56k is around
7k pull down. But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too. This means the pull down is
about the same -- 13k-ish. I'd use 15k. Then put a speed up
cap across the 56k. I think most BJTs have small pF to worry
about. So make it a 12-15pF. Something about that size.

Then see how it goes. You won't be saturating the hell out
of it, anyway. But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

Use a mosfet! 2N3904s belong in museums next to the 6SN7s.

At 5-10x the cost, you just lost 1000 Joerg points.

That's only .01% of a cent.


--
You can't have a sense of humor, if you have no sense.

 

[krw@att.bizzzzzzzzzzzz]

On Sat, 29 Oct 2011 13:19:12 -0400, "Michael A. Terrell"
<mike.terrell@earthlink.net> wrote:

"krw@att.bizzzzzzzzzzzz" wrote:

On Sat, 29 Oct 2011 10:04:56 -0700, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Fri, 28 Oct 2011 07:01:34 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

+12V--10kohm----+----Vout
|
|/
in--5V 1MHz-1kohm-+---| 2N3904
| |\
| |
+--1k-+
Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped. The base to ground resistor was added to try and
turn the base off quicker. It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current. That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.) You might want to run it closer to a
beta of 20 or 30, not 0.25. Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz. So 10% turnoff isn't good for you. You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave. Have you tried a 'speed up' capacitor across your
base drive resistor? I've used them with some success. And
they go back a LONG TIME. I think before Baker clamps.

I'm not that much of a designer. But this is basics. So
that's me. Anyway, here is a shot at it.

Take your above circuit. Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta. (It is a 2N3904, after all. It can do well.) So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT. That would get near 80k
ohm, but drop it to 56k or thereabouts. At 1mA Ic, you need
about Vbe=0.65V (or less, really.) So 0.65/5*56k is around
7k pull down. But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too. This means the pull down is
about the same -- 13k-ish. I'd use 15k. Then put a speed up
cap across the 56k. I think most BJTs have small pF to worry
about. So make it a 12-15pF. Something about that size.

Then see how it goes. You won't be saturating the hell out
of it, anyway. But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

Use a mosfet! 2N3904s belong in museums next to the 6SN7s.

At 5-10x the cost, you just lost 1000 Joerg points.


That's only .01% of a cent.

Ok, make that 1000K Joerg points.

 

[John Larkin]

On Sat, 29 Oct 2011 12:10:09 -0500, "krw@att.bizzzzzzzzzzzz"
<krw@att.bizzzzzzzzzzzz> wrote:

On Sat, 29 Oct 2011 10:04:56 -0700, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Fri, 28 Oct 2011 07:01:34 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current.  That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.)  You might want to run it closer to a
beta of 20 or 30, not 0.25.  Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz.  So 10% turnoff isn't good for you.  You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave.  Have you tried a 'speed up' capacitor across your
base drive resistor?  I've used them with some success.  And
they go back a LONG TIME.  I think before Baker clamps.

I'm not that much of a designer.  But this is basics.  So
that's me.  Anyway, here is a shot at it.

Take your above circuit.  Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta.  (It is a 2N3904, after all.  It can do well.)  So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT.  That would get near 80k
ohm, but drop it to 56k or thereabouts.  At 1mA Ic, you need
about Vbe=0.65V (or less, really.)  So 0.65/5*56k is around
7k pull down.  But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too.  This means the pull down is
about the same -- 13k-ish.  I'd use 15k.  Then put a speed up
cap across the 56k.  I think most BJTs have small pF to worry
about.  So make it a 12-15pF.  Something about that size.

Then see how it goes.  You won't be saturating the hell out
of it, anyway.  But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

Use a mosfet! 2N3904s belong in museums next to the 6SN7s.

At 5-10x the cost, you just lost 1000 Joerg points.

We pay $ 0.0208 for 2N7002s. And it doesn't need a schottly clamp
diode plus two or three passive components in the base drive path.

John

 

[Jon Kirwan]

On Sat, 29 Oct 2011 11:32:46 -0700, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Sat, 29 Oct 2011 12:10:09 -0500, "krw@att.bizzzzzzzzzzzz"
krw@att.bizzzzzzzzzzzz> wrote:

On Sat, 29 Oct 2011 10:04:56 -0700, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Fri, 28 Oct 2011 07:01:34 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Oct 28, 3:44 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Thu, 27 Oct 2011 15:57:39 -0700 (PDT), George Herold





gher...@teachspin.com> wrote:

First, I'm a transistor idiot.

I wanted to change a 5V 1 MHz clock signal to 12V.

I put this together,

        +12V--10kohm----+----Vout
                        |
                      |/
in--5V 1MHz-1kohm-+---|   2N3904
                  |   |\
                  |     |
                  +--1k-+
                        Gnd

Worked, but it took forever (~100 nS) to turn off.
The base was driven to near 800 mV. When the square wave clock
shut off the base just drifted down slowly for 100ns and then
dropped.  The base to ground resistor was added to try and
turn the base off quicker.  It barely helped.

If I don't drive it as hard will it turn off faster?
Make the input to base resistor ~5k
to give 0.8V at the base with 5V in?

Thanks
George H.

Oh feel free to suggest a better circuit too!

First thing that crosses my mind is the roughly 4mA of base
drive and the roughly 1mA of collector current.  That is hard
saturation and will place a lot of electrons in the collector
drift region (I think.)  You might want to run it closer to a
beta of 20 or 30, not 0.25.  Also, the other thing I note is
that you are getting 100ns turn off (no surprise) and running
at 1MHz.  So 10% turnoff isn't good for you.  You want less?
From your numbers, the two 1k resistors (moving to off state)
are in parallel at provide 500 ohms to ground for that charge
to leave.  Have you tried a 'speed up' capacitor across your
base drive resistor?  I've used them with some success.  And
they go back a LONG TIME.  I think before Baker clamps.

I'm not that much of a designer.  But this is basics.  So
that's me.  Anyway, here is a shot at it.

Take your above circuit.  Get rid of the two 1k resistors and
replace them with something that delivers on the order of a
20 beta.  (It is a 2N3904, after all.  It can do well.)  So
1/20th of 1mA or 20,000 times the 4V or so differential you
will have, less the Vbe of the BJT.  That would get near 80k
ohm, but drop it to 56k or thereabouts.  At 1mA Ic, you need
about Vbe=0.65V (or less, really.)  So 0.65/5*56k is around
7k pull down.  But that is R_tot and at 0.65V and 1/20mA
base, this is about R_base=13k for the BJT so the 7k is after
taking that into account, too.  This means the pull down is
about the same -- 13k-ish.  I'd use 15k.  Then put a speed up
cap across the 56k.  I think most BJTs have small pF to worry
about.  So make it a 12-15pF.  Something about that size.

Then see how it goes.  You won't be saturating the hell out
of it, anyway.  But I've no idea what you are driving,
either.

Just a hobbyist thought.

Jon- Hide quoted text -

- Show quoted text -

I'll give it a quick try. (Reducing the base current and voltage...
KRW suggested the same thing.)
I've reduced the collector resistor to 1k so my collector current is a
bit more now.. but I'll scale things accordingly.

George H.

Use a mosfet! 2N3904s belong in museums next to the 6SN7s.

At 5-10x the cost, you just lost 1000 Joerg points.

We pay $ 0.0208 for 2N7002s. And it doesn't need a schottly clamp
diode plus two or three passive components in the base drive path.

Since you bring it up, I never pay more than 0.6 cents for a
2N2222 or 2N3904/6 BJT. And I'm a damned hobbyist.

But the main thing is that George already said, "I'll try the
Bakers clamp/ Schottky first since there's already an npn
'lashed' in place." And who knows? He might get by with a
speed up, even.

Jon