Andrew Holme
Guest
Wed Jan 11, 2012 11:13 pm
I messed-up.
I made a pair of clamps to limit the min and max output voltage of a
*precision* programmable current source/sink.
I used a pair of diodes and the SO8-packaged dual LM6142: non-inverting
inputs to upper/lower limit references; inverting inputs to the node being
clamped; diodes between outputs and inverting inputs.
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision* is
wrecked.
Does anybody know a dual op-amp with the standard SO8 pinout that could be a
drop-in replacement?
I don't need stellar performance. A few MHz GBW and even uA input bias
could salvage the day.
TIA
Spehro Pefhany
Guest
Wed Jan 11, 2012 11:18 pm
On Wed, 11 Jan 2012 22:13:35 -0000, "Andrew Holme" <ah_at_nospam.com>
wrote:
Quote:
I messed-up.
I made a pair of clamps to limit the min and max output voltage of a
*precision* programmable current source/sink.
I used a pair of diodes and the SO8-packaged dual LM6142: non-inverting
inputs to upper/lower limit references; inverting inputs to the node being
clamped; diodes between outputs and inverting inputs.
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision* is
wrecked.
Does anybody know a dual op-amp with the standard SO8 pinout that could be a
drop-in replacement?
I don't need stellar performance. A few MHz GBW and even uA input bias
could salvage the day.
TIA
What supply voltage?
Andrew Holme
Guest
Wed Jan 11, 2012 11:28 pm
"Spehro Pefhany" wrote in message
news:8j2sg7h4fp68e32sl8uupgq0jifn9711dd_at_4ax.com...
On Wed, 11 Jan 2012 22:13:35 -0000, "Andrew Holme" <ah_at_nospam.com>
wrote:
Quote:
I messed-up.
I made a pair of clamps to limit the min and max output voltage of a
*precision* programmable current source/sink.
I used a pair of diodes and the SO8-packaged dual LM6142: non-inverting
inputs to upper/lower limit references; inverting inputs to the node being
clamped; diodes between outputs and inverting inputs.
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision*
is
wrecked.
Does anybody know a dual op-amp with the standard SO8 pinout that could be
a
drop-in replacement?
I don't need stellar performance. A few MHz GBW and even uA input bias
could salvage the day.
TIA
What supply voltage?
+/- 5V
Spehro Pefhany
Guest
Thu Jan 12, 2012 12:40 am
On Wed, 11 Jan 2012 22:28:21 -0000, the renowned "Andrew Holme"
<ah_at_nospam.com> wrote:
Quote:
"Spehro Pefhany" wrote in message
news:8j2sg7h4fp68e32sl8uupgq0jifn9711dd_at_4ax.com...
On Wed, 11 Jan 2012 22:13:35 -0000, "Andrew Holme" <ah_at_nospam.com
wrote:
I messed-up.
I made a pair of clamps to limit the min and max output voltage of a
*precision* programmable current source/sink.
I used a pair of diodes and the SO8-packaged dual LM6142: non-inverting
inputs to upper/lower limit references; inverting inputs to the node being
clamped; diodes between outputs and inverting inputs.
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision*
is
wrecked.
Does anybody know a dual op-amp with the standard SO8 pinout that could be
a
drop-in replacement?
I don't need stellar performance. A few MHz GBW and even uA input bias
could salvage the day.
TIA
What supply voltage?
+/- 5V
LMC6482, perhaps, (1.5MHz GBW) about $1.00 ea/100
or maybe
LMP7702 (2.5MHz GBW) about $1.65 ea/100
Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff_at_interlog.com Info for manufacturers:
http://www.trexon.com
Embedded software/hardware/analog Info for designers:
http://www.speff.com
NT
Guest
Fri Jan 13, 2012 3:55 pm
On Jan 11, 10:13 pm, "Andrew Holme" <a...@nospam.com> wrote:
Quote:
I messed-up.
I made a pair of clamps to limit the min and max output voltage of a
*precision* programmable current source/sink.
I used a pair of diodes and the SO8-packaged dual LM6142: non-inverting
inputs to upper/lower limit references; inverting inputs to the node being
clamped; diodes between outputs and inverting inputs.
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision* is
wrecked.
Does anybody know a dual op-amp with the standard SO8 pinout that could be a
drop-in replacement?
I don't need stellar performance. A few MHz GBW and even uA input bias
could salvage the day.
TIA
Something's wrong with this question. Opamps have huge gain, and so
long as the output isnt hard against a rail, there will not be
anywhere near a diode drop of differential input. When you add nfb to
bring gain down, this makes teh gain of the whole circuit lower, but
does not change the actual ic's gain, which is still extremely high,
and the minimal differential input v still remains true.
NT
Phil Hobbs
Guest
Sat Jan 14, 2012 8:58 pm
whit3rd wrote:
Quote:
On Friday, January 13, 2012 5:55:45 AM UTC-8, NT wrote:
On Jan 11, 10:13 pm, "Andrew Holme" <a....@nospam.com> wrote:
I messed-up.
I made a pair of clamps to limit the min and max output voltage ...
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision* is
wrecked.
Something's wrong with this question. Opamps have huge gain, and so
long as the output isnt hard against a rail, there will not be
anywhere near a diode drop of differential input.
That statement assumes stability with the outputs out of saturation, and only
in some negative-feedback situations does that circumstance hold.
Comparator or Schmitt-trigger uses of operational amplifiers are
not uncommon, and will misbehave badly if the input doesn't have a
suitable differential-input range. The question is a good one (in
my opinion, it's a classic!).
CMOS inputs can exhibit threshold drift if they spend a lot of time with
large differential inputs. Bipolars can have similar problems if you
zener a BE junction. For a comparator application, it's usually best to
use a comparator.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Tim Williams
Guest
Sat Jan 14, 2012 9:26 pm
"Phil Hobbs" <pcdhSpamMeSenseless_at_electrooptical.net> wrote in message
news:4F11DE63.A7A0DB41_at_electrooptical.net...
Quote:
CMOS inputs can exhibit threshold drift if they spend a lot of time with
large differential inputs.
Hmm, is that due to stray ions by any chance?
I recall reading sodium ions trapped in the gate oxide used to cause
threshold drift with input flux (i.e., apply a voltage for along time and
Vgs(th) apparently changes*). Extremely high purity everything
(especially wash water) more-or-less cured that, at least allowing
transistors to be used for general purposes.
Presumably, that's quantified as the change in Vgs(th) due to flux being
significantly less than the guaranteed variation in threshold voltage by
manufacturing and temperature. But if there's still a couple ions
remaining, I can see it would still change by a couple milivolts, no? Not
important on your power transistor, but a big deal with precision,
non-chopper CMOS amps.
*Say, at constant input bias, in the linear (= resistive) region, this
would vary the channel resistance. Does this not make a "transmemristor"?
Tim
--
Deep Friar: a very philosophical monk.
Website:
http://webpages.charter.net/dawill/tmoranwms
whit3rd
Guest
Sat Jan 14, 2012 9:54 pm
On Friday, January 13, 2012 5:55:45 AM UTC-8, NT wrote:
Quote:
On Jan 11, 10:13 pm, "Andrew Holme" <a....@nospam.com> wrote:
I messed-up.
I made a pair of clamps to limit the min and max output voltage ...
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision* is
wrecked.
Quote:
Something's wrong with this question. Opamps have huge gain, and so
long as the output isnt hard against a rail, there will not be
anywhere near a diode drop of differential input.
That statement assumes stability with the outputs out of saturation, and only
in some negative-feedback situations does that circumstance hold.
Comparator or Schmitt-trigger uses of operational amplifiers are
not uncommon, and will misbehave badly if the input doesn't have a
suitable differential-input range. The question is a good one (in
my opinion, it's a classic!).
Allan Herriman
Guest
Sun Jan 15, 2012 12:57 am
On Sat, 14 Jan 2012 14:26:42 -0600, Tim Williams wrote:
Quote:
"Phil Hobbs" <pcdhSpamMeSenseless_at_electrooptical.net> wrote in message
news:4F11DE63.A7A0DB41_at_electrooptical.net...
CMOS inputs can exhibit threshold drift if they spend a lot of time
with large differential inputs.
Hmm, is that due to stray ions by any chance?
I recall reading sodium ions trapped in the gate oxide used to cause
threshold drift with input flux (i.e., apply a voltage for along time
and Vgs(th) apparently changes*). Extremely high purity everything
(especially wash water) more-or-less cured that, at least allowing
transistors to be used for general purposes.
Presumably, that's quantified as the change in Vgs(th) due to flux being
significantly less than the guaranteed variation in threshold voltage by
manufacturing and temperature. But if there's still a couple ions
remaining, I can see it would still change by a couple milivolts, no?
Not important on your power transistor, but a big deal with precision,
non-chopper CMOS amps.
This effect also occurs in digital processes used to implement CMOS ring
oscillators or other delay lines consisting of an inverter chain. If the
oscillator is stopped somehow, the steady state bias will cause the
threshold voltage in every second P-channel fet (the ones with -ve gate
bias) to change, and the duty cycle (once the oscillator is gated back
on) may go outside of spec.
If the oscillator runs continuously, the effect will be roughly even over
all the P fets, and the duty cycle isn't disturbed.
http://www.xilinx.com/support/documentation/white_papers/wp224.pdf
Regards,
Allan
Spehro Pefhany
Guest
Sun Jan 15, 2012 4:47 am
On Sat, 14 Jan 2012 14:58:27 -0500, the renowned Phil Hobbs
<pcdhSpamMeSenseless_at_electrooptical.net> wrote:
Quote:
whit3rd wrote:
On Friday, January 13, 2012 5:55:45 AM UTC-8, NT wrote:
On Jan 11, 10:13 pm, "Andrew Holme" <a....@nospam.com> wrote:
I messed-up.
I made a pair of clamps to limit the min and max output voltage ...
Trouble is, the LM6142 has large input currents when VIN(differential)
exceeds a diode drop. So, when my upper clamping limit is 3V and the node
is 1V, I get current flowing through the op-amp inputs and my *precision* is
wrecked.
Something's wrong with this question. Opamps have huge gain, and so
long as the output isnt hard against a rail, there will not be
anywhere near a diode drop of differential input.
That statement assumes stability with the outputs out of saturation, and only
in some negative-feedback situations does that circumstance hold.
Comparator or Schmitt-trigger uses of operational amplifiers are
not uncommon, and will misbehave badly if the input doesn't have a
suitable differential-input range. The question is a good one (in
my opinion, it's a classic!).
CMOS inputs can exhibit threshold drift if they spend a lot of time with
large differential inputs.
I thought they pretty much licked that problem. In any case, the OP
seems to be happy with mV of Vos, so I doubt even a CA3140 would pose
much of a problem.
Quote:
Bipolars can have similar problems if you
zener a BE junction. For a comparator application, it's usually best to
use a comparator.
I assume this application is a bit like a comparator when the input is
outside the limits, but acts as an op-amp when the input exceeds the
limit, so a comparator would have other issues.
Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff_at_interlog.com Info for manufacturers:
http://www.trexon.com
Embedded software/hardware/analog Info for designers:
http://www.speff.com
