Voltage follower with a step function (need a design for one)

[Rocky]

Hi there,
I would like to effect the following transfer function (seen below in
ASCII art) using analog components such as some opamp(s), and/or diodes
etc. I'm not too fussy about the input and output impedances since I
can buffer the design. Here is a brief description of the function:
As the input voltage increases from 0V to +4V , I would like the output
to track (i.e. follow) the input voltage only after the input voltage
has exceeded a user adjustable threshold (nominally about 2V). The
function should have no (or minimal) hysteresis -i.e. behave the same
whether Vin ramps from 0-4V or from 4-0V. It does not need a high
bandwidth, but should operate from DC to about 20Hz.


Vout
^
4V | *
| *
| *
| *
| *
2V | *
| *
| *
| *
| *
|*********
-------------------------> Vin
' ' '
0 2V 4V


regards
Rocky

 

[Ban]

Hi there,
I would like to effect the following transfer function (seen below in
ASCII art) using analog components such as some opamp(s), and/or
diodes etc. I'm not too fussy about the input and output impedances
since I can buffer the design. Here is a brief description of the
function: As the input voltage increases from 0V to +4V , I would
like the output to track (i.e. follow) the input voltage only after
the input voltage has exceeded a user adjustable threshold (nominally
about 2V). The function should have no (or minimal) hysteresis -i.e.
behave the same whether Vin ramps from 0-4V or from 4-0V. It does
not need a high bandwidth, but should operate from DC to about 20Hz.

What you have drawn will IMHO do need a high bandwidth, actually an infinite
BW. And with 20kHz this approaches pretty much the limit of a normal opamp,
especially when you want no hysteresis. To switch 20kHz in say 0.1% of its
period means 50ns and if a comparator will need already say 40ns to detect
the threshold x-ing, you have to switch 2V to gnd in 10ns, which is
possible. But then the O/P signal will need a really fast opamp buffer with
this 200V/us slew.

Vout
^
4V | *
| *
| *
| *
| *
2V | *
| *
| *
| *
| *
|*********
-------------------------> Vin
' ' '
0 2V 4V

Another thing is you mention AC signals. Do you want a symmetrical
threshold, you have not drawn anything to the left of your y-axis. What
should happen with negative values?

 

[John O'Flaherty]

Hi there,
I would like to effect the following transfer function (seen below in
ASCII art) using analog components such as some opamp(s), and/or diodes
etc. I'm not too fussy about the input and output impedances since I
can buffer the design. Here is a brief description of the function:
As the input voltage increases from 0V to +4V , I would like the output
to track (i.e. follow) the input voltage only after the input voltage
has exceeded a user adjustable threshold (nominally about 2V). The
function should have no (or minimal) hysteresis -i.e. behave the same
whether Vin ramps from 0-4V or from 4-0V. It does not need a high
bandwidth, but should operate from DC to about 20Hz.


Vout
^
4V | *
| *
| *
| *
| *
2V | *
| *
| *
| *
| *
|*********
-------------------------> Vin
' ' '
0 2V 4V

You could use a comparator with an adjustable voltage on one input, the
circuit input voltage on the other, and the comparator output going to
the control input of an analog switch, like a 4066. The analog input to
the switch is also connected to the circuit input, and the analog
switch output is connected to a resistor, say 10kohm, to ground, and to
the circuit output. You might be able to use just an FET as the analog
switch, and an opamp section as the comparator; then with a quad opamp
you could have voltage followers on input and output, and one section
left over.

 

[Fred Bloggs]

I would like to effect the following transfer function (seen below in

ASCII art) using analog components such as some opamp(s), and/or diodes
etc. I'm not too fussy about the input and output impedances since I
can buffer the design. Here is a brief description of the function:
As the input voltage increases from 0V to +4V , I would like the output
to track (i.e. follow) the input voltage only after the input voltage
has exceeded a user adjustable threshold (nominally about 2V). The
function should have no (or minimal) hysteresis -i.e. behave the same
whether Vin ramps from 0-4V or from 4-0V. It does not need a high
bandwidth, but should operate from DC to about 20Hz.


Vout
^
4V | *
| *
| *
| *
| *
2V | *
| *
| *
| *
| *
|*********
-------------------------> Vin
' ' '
0 2V 4V

You could "effect" it like so- but there are details in stability to
attend to, and the output amp clamps low at 0V:
View in a fixed-width font such as Courier.

 

[Tony]

Rocky, how about this:

just clamp the input to the op amp to zero until the input is above 2
volts, then let it track the input. For example, run the input through
a couple of series resistors, and put a clamping transistor driven by a
voltage comparator at the node between the two resistors. The clamping
transistor is driven to saturation by a second op amp when the input is
less than two volts, and driven off when 2 Volts is exeeded.

 

[John Fields]

Hi there,
I would like to effect the following transfer function (seen below in
ASCII art) using analog components such as some opamp(s), and/or diodes
etc. I'm not too fussy about the input and output impedances since I
can buffer the design. Here is a brief description of the function:
As the input voltage increases from 0V to +4V , I would like the output
to track (i.e. follow) the input voltage only after the input voltage
has exceeded a user adjustable threshold (nominally about 2V). The
function should have no (or minimal) hysteresis -i.e. behave the same
whether Vin ramps from 0-4V or from 4-0V. It does not need a high
bandwidth, but should operate from DC to about 20Hz.


Vout
^
4V | *
| *
| *
| *
| *
2V | *
| *
| *
| *
| *
|*********
-------------------------> Vin
' ' '
0 2V 4V

---
View in Courier:

+V>----------+--------------------------+
| |
Vin--+-------|------[10k]--+--[10k]----|+\
| | | | >------+-->Vout
| [10k] | +--|-/LM392 |
| | | | | OPAMP |
[7500] | +--[4M7]---+ +---|--------+
| | | | |
+-------|--+--|+\ | |
| | | >----+ |
[12.1k] +-----|-/LM392 |
| | COMPARATOR |
| [LM385-1.2] |
| | |
GND>-+-------+--------------------------+----------->GND

 

[John Fields]

What you have drawn will IMHO do need a high bandwidth, actually an infinite
BW. And with 20kHz

 

[John Fields]

+V>----------+--------------------------+
| |
Vin--+-------|------[10k]--+--[10k]----|+\
| | | | >------+-->Vout
| [10k] | +--|-/LM392 |
| | | | | OPAMP |
[7500] | +--[4M7]---+ +---|--------+
| | | | |
+-------|--+--|+\ | |
| | | >----+ |
[12.1k] +-----|-/LM392 |
| | COMPARATOR |
| [LM385-1.2] |
| | |
GND>-+-------+--------------------------+----------->GND

---
I think this is better:

+V>----------+---------------------+
| |
Vin--+-------|------[10k]--+------|+\
| | | | >------+-->Vout
| [10k] | +--|-/LM392 |
| | | | | OPAMP |
[7500] | +--[4M7]---+ +---|--------+
| | | | |
+-------|--+--|+\ | |
| | | >----+ |
[12.1k] +-----|-/LM392 |
| | COMPARATOR |
| [LM385-1.2] |
| | |
GND>-+-------+---------------------+----------->GND

 

[Jim Thompson]

Hi there,
I would like to effect the following transfer function (seen below in
ASCII art) using analog components such as some opamp(s), and/or diodes
etc. I'm not too fussy about the input and output impedances since I
can buffer the design. Here is a brief description of the function:
As the input voltage increases from 0V to +4V , I would like the output
to track (i.e. follow) the input voltage only after the input voltage
has exceeded a user adjustable threshold (nominally about 2V). The
function should have no (or minimal) hysteresis -i.e. behave the same
whether Vin ramps from 0-4V or from 4-0V. It does not need a high
bandwidth, but should operate from DC to about 20Hz.


Vout
^
4V | *
| *
| *
| *
| *
2V | *
| *
| *
| *
| *
|*********
-------------------------> Vin
' ' '
0 2V 4V


regards
Rocky

See "DeadZone.pdf", a two-sided equivalent to your problem, on the
S.E.D/Schematics page of my website.

But there are many ways to implement such circuit functions... surf on
"piece wise linear" or "breakpoint analysis"... commonly taught in
undergraduate EE classes.

...Jim Thompson

 

[Fred Bloggs]

Vout
^
4V | *
| *
| *
| *
| *
2V | *
| *
| *
| *
| *
|*********
-------------------------> Vin
' ' '
0 2V 4V

This should be trouble free using a single-supply type OA:
View in a fixed-width font such as Courier.

 

[Rocky]

Thank you everyone for you contributions to my question. In summary
I'll probably use either John Fields' or Tony's circuit because they
are single chip solutions. They are both similar in their approach,
just that John used a comparator in place of Tony's transistor to pull
down the input when 0<Vin<2V.
Jim Thompson's suggestion was kind of similar, although it uses a 4066
analog switch to gate the signal.
I haven't yet wrapped my head around how Fred Blogg's circuit works,
but given the higher active component count I probably wouldn't use it
unless it has specific advantages that I'm not aware of.

regards
Rocky

 

[Rocky]

On Sun, 16 Apr 2006 10:20:13 -0500, John Fields


---
I think this is better:

+V>----------+---------------------+
| |
Vin--+-------|------[10k]--+------|+\
| | | | >------+-->Vout
| [10k] | +--|-/LM392 |
| | | | | OPAMP |
[7500] | +--[4M7]---+ +---|--------+
| | | | |
+-------|--+--|+\ | |
| | | >----+ |
[12.1k] +-----|-/LM392 |
| | COMPARATOR |
| [LM385-1.2] |
| | |
GND>-+-------+---------------------+----------->GND

Hi John,
I was just about to implement your suggested circuit when I came across
something I don't understand. When Vin >2V, does the output from the
LM392 comparator swing to the comparator's positive supply rail? I
thought that's how a comparator worked - basically an opamp that allows
rail to rail output swing.
If however the comparator's output impedance went high (like a
transistor when switched off) when its positive input voltage >neg.
input voltage then that would make sense, but because I can't see how
it would, then surely we need the opamp to drive a transistor as Tony
suggested above.

While I think about it, is there any reason why I need to use
comparator IC instead of say an LM158/LM358 opamp ? I justify my
reasoning later on but I ask this question because I firstly don't have
any comparators lying around in my spare parts box to prototype the
circuit, and secondly I'd like to keep the I.C. count low if possible
(I know the LM392 is a good choice with comparator+ompamp in the one
pkg, but I'd need to mail order the bit through Farnell or Radio Spares
which would unnecessarily bump up the cost for me.)
Anyway, the datasheet for an LM358 opamp says:
"In the linear mode the input common-mode voltage range includes ground
and the output voltage can also swing to ground, even though operated
from only a single power supply voltage.". Additionally, because I
only need to run at about 20Hz, even if I do drive the output into the
+ve supply rail (thus saturating it), I don't mind if it takes up to
10ms to recover!

regards
Rocky

 

[John Fields]

On Sun, 16 Apr 2006 10:20:13 -0500, John Fields


---
I think this is better:

+V>----------+---------------------+
| |
Vin--+-------|------[10k]--+------|+\
| | | | >------+-->Vout
| [10k] | +--|-/LM392 |
| | | | | OPAMP |
[7500] | +--[4M7]---+ +---|--------+
| | | | |
+-------|--+--|+\ | |
| | | >----+ |
[12.1k] +-----|-/LM392 |
| | COMPARATOR |
| [LM385-1.2] |
| | |
GND>-+-------+---------------------+----------->GND

Hi John,
I was just about to implement your suggested circuit when I came across
something I don't understand. When Vin >2V, does the output from the
LM392 comparator swing to the comparator's positive supply rail?

---
No, it goes open collector, unclamping the + input to the opamp.
---

I
thought that's how a comparator worked - basically an opamp that allows
rail to rail output swing.

---
It's more like an opamp with all the compensation removed to make it
fast, and an open-collector transistor output.
---

If however the comparator's output impedance went high (like a
transistor when switched off) when its positive input voltage >neg.
input voltage then that would make sense, but because I can't see how
it would, then surely we need the opamp to drive a transistor as Tony
suggested above.

---
It's precisely because it _does_ work that way that we don't need
the transistor; there's one already in the comparator.
---

While I think about it, is there any reason why I need to use
comparator IC instead of say an LM158/LM358 opamp ?

---
It makes life easier, but if you can't use a proper comparator you
could do this:

+V>----------+-------------------------------------+
| |
Vin--+-------|------[10k]-----------------+-------|+\LM358
| | | | >--+-->Vout
| [10k] | +--|-/ |
| | | | | |
[7K5] | +--[1M]--+ | +---|----+
| | | | | |
| +--[1k]--+--|+\ | C |
| | | >--+--[1k]--B Q1 |
+-------|-----------|-/LM358 E |
| | | |
[12k1] |K | |
| [LM385-1.2] | |
| | | |
GND>-+-------+----------------------------+--------+------->GND

Choose Q1 to give you the lowest Vce(sat) you can get or, if you
choose a MOSFET, the lowest Rds(on).

 

[Rocky]

Hi John,
thanks again for your timely and valuable information, I am very
appreciative.
regards
Rocky.

 

[Ken Smith]

+V>----------+-------------------------------------+
| |
Vin--+-------|------[10k]-----------------+-------|+\LM358
| | | | >--+-->Vout
| [10k] | +--|-/ |
| | | | | |
[7K5] | +--[1M]--+ | +---|----+
| | | | | |
| +--[1k]--+--|+\ | C |
| | | >--+--[1k]--B Q1 |
+-------|-----------|-/LM358 E |
| | | |
[12k1] |K | |
| [LM385-1.2] | |
| | | |
GND>-+-------+----------------------------+--------+------->GND

Choose Q1 to give you the lowest Vce(sat) you can get or, if you
choose a MOSFET, the lowest Rds(on).

If the max signal voltage is under about 7V, swap Q1's E and C legs.

 

[Rocky]

The circuit John drew shows a transistor in common emitter
configuration, (which must be an NPN tranny assuming we want the
conventional current to flow from C to E).
So you are suggesting a common collector config. for Vin<7V (which must
be PNP tranny assuming we want the conventional current to flow from E
to C). Why? What will that achieve? I thought that emitter follower
designs were really only used for impedance matching and buffering
applications whereas common emitters are used for voltage gain
applications.
http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/npncc.html#c2

regards
Rocky

 

[John Fields]

The circuit John drew shows a transistor in common emitter
configuration, (which must be an NPN tranny assuming we want the
conventional current to flow from C to E).
So you are suggesting a common collector config. for Vin<7V (which must
be PNP tranny assuming we want the conventional current to flow from E
to C).

---
No, it'll stay an NPN.
---

Why?

---
What reversing the emitter and collector does is provide a dramatic
reduction in saturation voltage across the transistor.


Try it.

If you don't have a copy of LTSPiCE, go to:

http://www.linear.com/company/software.jsp

download a copy, and run this with the transistor hooked up first
one way and then the other:

Version 4
SHEET 1 880 680
WIRE -208 320 -208 112
WIRE -208 464 -208 400
WIRE -64 320 -64 240
WIRE -64 464 -208 464
WIRE -64 464 -64 400
WIRE -32 112 -208 112
WIRE -32 240 -64 240
WIRE 80 240 48 240
WIRE 144 112 48 112
WIRE 144 192 144 112
WIRE 144 464 -64 464
WIRE 144 464 144 288
WIRE 144 496 144 464
FLAG 144 496 0
SYMBOL npn 80 192 R0
SYMATTR InstName Q1
SYMATTR Value 2N3904
SYMBOL voltage -64 304 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 2
SYMBOL res 64 224 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R1
SYMATTR Value 1k
SYMBOL voltage -208 304 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V2
SYMATTR Value 4
SYMBOL res 64 96 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R2
SYMATTR Value 10k
TEXT -242 520 Left 0 !.tran .1

 

[Jonathan Kirwan]

<snip>
What reversing the emitter and collector does is provide a dramatic
reduction in saturation voltage across the transistor.

Try it.

I see the reduction.

Reversed: Vce 2.47mV
Ie 0.395mA
Ib 1.280mA
Ic 1.675mA

Forward: Vce 7.73mV
Ic 0.400mA
Ib 1.270mA
Ie 1.670mA

However, that is VERY DEEP saturation case with a beta of about 0.3.
If I keep the transistor saturated, but reduce the 10k resistor to 1k
thus bringing the collector current up, but still seriously saturated
with only a beta of 2 or 3:

Reversed: Vce 40.77mV
Ie 3.960mA
Ib 1.275mA
Ic 5.235mA

Forward: Vce 20.07mV
Ic 3.980mA
Ib 1.260mA
Ie 5.240mA

So when your base drive significantly dominates everything else, you
can squeeze a few more millivolts out of the Vce, reversed.

Is that about it?

Jon

 

[John Fields]

I see the reduction.

Reversed: Vce 2.47mV
Ie 0.395mA
Ib 1.280mA
Ic 1.675mA

Forward: Vce 7.73mV
Ic 0.400mA
Ib 1.270mA
Ie 1.670mA

However, that is VERY DEEP saturation case with a beta of about 0.3.
If I keep the transistor saturated, but reduce the 10k resistor to 1k
thus bringing the collector current up, but still seriously saturated
with only a beta of 2 or 3:

Reversed: Vce 40.77mV
Ie 3.960mA
Ib 1.275mA
Ic 5.235mA

Forward: Vce 20.07mV
Ic 3.980mA
Ib 1.260mA
Ie 5.240mA

So when your base drive significantly dominates everything else, you
can squeeze a few more millivolts out of the Vce, reversed.

Is that about it?

---
I guess so, except that it seems that by doing it backwards and
keeping the base drive high we can achieve emitter-to-collector
voltage saturation levels of less than 10mV with low collector
currents.

There's gotta be an application for which that's a perfect fit,
don't you think?

And how about the tempco?

 

[Rocky]

---
What reversing the emitter and collector does is provide a dramatic
reduction in saturation voltage across the transistor.


Try it.

If you don't have a copy of LTSPiCE, go to:

http://www.linear.com/company/software.jsp

download a copy, and run this with the transistor hooked up first
one way and then the other:

Version 4
SHEET 1 880 680
WIRE -208 320 -208 112
WIRE -208 464 -208 400
WIRE -64 320 -64 240
WIRE -64 464 -208 464
WIRE -64 464 -64 400
WIRE -32 112 -208 112
WIRE -32 240 -64 240
WIRE 80 240 48 240
WIRE 144 112 48 112
WIRE 144 192 144 112
WIRE 144 464 -64 464
WIRE 144 464 144 288
WIRE 144 496 144 464
FLAG 144 496 0
SYMBOL npn 80 192 R0
SYMATTR InstName Q1
SYMATTR Value 2N3904
SYMBOL voltage -64 304 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 2
SYMBOL res 64 224 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R1
SYMATTR Value 1k
SYMBOL voltage -208 304 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V2
SYMATTR Value 4
SYMBOL res 64 96 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R2
SYMATTR Value 10k
TEXT -242 520 Left 0 !.tran .1

I downloaded LTSpice and opened up the above text as a circuit file
(*.cir), when I ran it, I got the error message
"Circuit: Version 4

Fatal Error: Multiple instances of "Symattr"

Do you know why?
regards
Rocky