On 11/8/2014 4:34 PM, rickman wrote:
On 11/8/2014 3:27 PM, Jerry Stuckle wrote:
On 11/7/2014 8:40 PM, rickman wrote:
On 11/7/2014 7:57 PM, Jerry Stuckle wrote:
On 11/7/2014 7:18 PM, rickman wrote:
On 11/7/2014 5:07 PM, Jerry Stuckle wrote:
On 11/7/2014 4:40 PM, rickman wrote:
On 11/7/2014 4:23 PM, Jerry Stuckle wrote:
On 11/7/2014 3:07 PM, rickman wrote:
On 11/7/2014 1:53 PM, Jerry Stuckle wrote:
On 11/7/2014 1:26 PM, rickman wrote:
On 11/7/2014 1:17 PM, Jerry Stuckle wrote:
On 11/7/2014 1:02 PM, rickman wrote:
On 11/7/2014 10:49 AM, Jerry Stuckle wrote:
On 11/6/2014 11:45 AM, rickman wrote:
On 11/6/2014 10:04 AM, Jerry Stuckle wrote:
On 11/5/2014 1:29 PM, rickman wrote:
On 11/4/2014 9:42 PM, Jerry Stuckle wrote:
On 11/4/2014 6:29 PM, rickman wrote:
I am working on a project for receiving a very narrow
bandwidth
signal
at 60 kHz. One of the design goals is to keep the power
consumption to
an absolute minimum. I'm trying to figure out how to
run a
pre-amplifier on less than 100 uW. So far I have found
nothing.
Any
suggestions?


I agree with Jim. We need many more specifics to
provide a
meaningful
answer. There are a lot of micropower opamps out there
now,
but
the
devil is in the details.

I've only found one detail that is giving me the devil.
That
is the
bandwidth. The signal is 60 kHz. I can't think of any
other
issues I
would have with any amp capable of amplifying this signal
with
a low
power level. What more info do you feel is needed? Can
you ask
questions? Better yet, just point me to any amp that will
meet
my two
stated requirements!


The other posts you made had the info - things like
impedance and
gain
are important, as is frequency of operation (but we already
know
that).

A couple of things to consider, however. The higher the
impedance, the
more susceptible it will be to ambient noise pickup.
You're
starting
with a very small signal and may need to add shielding to
limit
external
noise.

The other problem is you're asking for low impedance
output. Low
impedance limits noise pickup, but increases current
drain. So
how low
of an impedance do you want?

I don't follow on this. How does a low output impedance
drive the
current drain?


There are op amps with very high (in the gigaohm range)
input
impedance
and pretty low quiescent current drain. How much it draws
during use
will be greatly dependent on the output current required,
which
obviously depends on output voltage and impedance.

Consider the current used only by the amp, not the load.


I don't have time right now, but later today I'll look
through
some of
my data sheets on op amps to see what I can find.

Thanks.


Total current is not just dependent on output current; it
also is
affected by the design of the chip. Op amps are not just
single
transistor devices; a lower output impedance also means more
current to
drive the output stage, which affects other components. So
even
if you
have a high impedance load, the lower the output impedance of
the op
amp
(i.e. the more current it can source/sink at a specific
supply
voltage),
the more overall current the op amp will draw.

With that said, I did some looking around (sorry for not
getting
back to
you quicker - yesterday was pretty busy). Depending on your
needs,
there are hundreds you can choose from. I might recommend
you
check
out
http://www.mouser.com/Semiconductors...mps/_/N-6j73m/







. You can pick and choose the parameters you want. Another
one
I've
used is http://www.newark.com/operational-amplifiers.

Between the two I found several hundred possibilities, but
you
know the
details of what you want better than I do, so rather than
guess at
what
you might want, I think this would be better. It should give
you a
start.

I have done this before and found nothing. But I did it
again at
both
Mouser and Digikey and found several. One listed by Mouser
looked
especially good only to find rather than 0.75 uA of supply
current, it
had 0.75 mA of supply current. lol

But then the next part, same thing... another one... and
another... one
part I'm not sure what to make of it. The selection table
shows
supply
current of 0.034 mA and the data sheet shows 25 A! Yes,
that's
right,
the data sheet shows between 25 and 300 Amps for typical
supply
current!!! I would contact TI about this obvious typo, but
this
part is
not suitable because of the GBW which is also incorrect in the
selection
table.

Same thing at Digikey, everything in the selection table that
meets
these two requirements is a mistake.


A couple of things.

First of all, I've found minor errors in the listings at
Mouser (I
don't
use Digikey much), but never real glaring errors. And this
is th
first
time I've seen a TI datasheet that far off. Looks like someone
dropped
a decimal point . However, I've found Mouser is
interested in
correcting errors; they are input by humans, after all, at some
point in
time, and errors do creep in.

Yes, when you list millions of parts there will be errors. I
have
written digikey many times about listing errors and they always
thank
me. I'm sure Mouser is no different.


Secondly, the current shown is going to be max current, which
will
depend on the output impedance (and the amount that has to be
sourced/sunk). It's not going to pull this all the time; I
would
expect
your actual current draw to be much less since you're 1) going
into a
high impedance load and 2) not going from rail to rail.

I find the opposite. The current listed is under specified
conditions
which usually *do not* include output drive. In fact, it
usually
listed
as a quiescent current.


Well, yes and no. Op amps typically sink more than they source,
and
the
sink current does not come from the chip. Source current at the
output
is supplied by the chip, of course.

And I've found a wide difference between how op amp specs are
listed;
some show quiescent current, some show average current under
typical
operating conditions. Some even show maximum current which
can be
drawn. So I'll retract that statement above. Wasn't thinking
clearly.


Also, if you use a bipolar supply, then current drain should be
less
because you'll be operating near ground, instead of the
midpoint
of a
single supply voltage (where the output would be at 1/2 Vcc).
Some of
these are quite low voltage, and I would think a couple of the
larger
lithium coin batteries should last quite a while.

Not sure how the ground level would affect the bias currents.
When
the
supply voltage is lowered the GBW lowers as well.


If the output is at ground level, no current will be pulled from
either
rail (at the output). Shifting above or below that will draw a
little
current, reference zero. However, if you're running a single
ended
supply, your output will be at 1/2 Vcc, and will always be
pulling
some
current to maintain that level. The signal will change that
slightly,
increasing or decreasing. But unless you have a square wave with
a 50%
duty cycle, you'll end up needing more current from the single
ended
supply.

What you are saying is only true if your load is ground connected.
The
load for this circuit will be a voltage source through a high
impedance.
The input is differential and to make it as sensitive as
possible a
bias will be applied to one input sufficient to offset the input
bias
voltage. So in reality the load will be biased to approx 1/2 Vcc.


True, but with a bipolar supply, the input is referenced to ground
and
no current flows with no input. The output is also referenced to
ground, so no current flows their, either. And with both input and
output at ground potential, there is less current flowing
internally.

I just explained a scenario where the load will draw current from
the
amp regardless of power supply arrangement. You are making an
assumption that the input and output are ground referenced. That is
independent of the supply arrangement.


But if they aren't ground referenced, then they must be referenced
to an
artificial ground, i.e. 1/2 Vcc. And creating that artificial ground
will require a certain amount of current.

For instance - it's common to bias the input of an op amp running
from a
single ended supply at 1/2 Vcc. This is generally done with a
couple of
resistors, in various configurations. But you will always have a
small
current through those resistors. The lower the impedance of the
input,
the lower the resistors must be.

Output in this case will also be referenced at 1/2 Vcc, which
means the
op amp output is conducting some current all of the time. Even if
the
output is capacitive coupled to the load, internally the op amp must
draw some current to maintain that 1/2 Vcc. Again, the amount of
current is dependent on the output impedance, but it is still there.

With a bipolar supply, the op amp doesn't draw input current with no
input signal, and doesn't have to source or sink any current when you
have 0V output.

I hope this is a bit clearer.

It is not a question of clear. It isn't relevant to the power
consumption of the opamp. No matter what the reference, somebody,
somewhere even if it is in the power supply, is using power sometime
unless there are no voltages on any of the resistors in the
design. But
none of that is relevant to the power consumed by the opamp when in
the
quiescent state.


Yes and no. Op amps are by design bipolar devices; they go plus and
minus from some value. It can be zero volts (ground), or it can be
some
value between Vcc and ground. In the latter case, an artificial ground
must be established; by definition this takes current to establish a
voltage between Vcc and ground.

Actually your characterization of opamps is not accurate. A very few
are designed to use dual supplies but most can work with unipolar
supplies. Basically they don't know where the ground is and they don't
care. The quiescent power the opamp dissipates is not related to where
ground is.


I suggest you build an op amp out of discreet components to understand
how it works. It was one of the projects in an EE class. It was very
educational to see just how they work.

As I said before, "But none of that is relevant to the power consumed by
the opamp when in the quiescent state."


Actually, it does. See above.


I'm trying to pick an opamp. I have no need to evaluate the rest
of the
design when the problem is trying to find an amp with sufficient
GBW and
low quiescent current.


But when you're talking very low power like you are, low quiescent
current is very dependent on the power supply, as well as input and
output impedance and the chip used.

It is certainly not dependent on the power supply configuration. I can
design the rest of the circuit. I'm just looking for a low power
device.


Once again, it does. But I see I won't convince you.

No, you won't. If you are talking about the minute differences in
internal biasing, etc, then I can't rule out all effects absolutely, but
if you had a valid point you would be able to explain it other than just
stating the fact repeatedly.


I have tried explaining it several times. However, you haven't accepted
the explanations. I can't help that.


The real point is that this is *LOAD* current, not amplifier
current and
is independent of the amplifier and so considered separately since
selection of the amplifier has no impact on it.


That is true - to an extent. When the op amp is sinking
current, the
current comes from the load, not the op amp. However, internally
there
must still be current flowing to provide that sink.

You aren't grasping the issue. It is not about which direction the
current is flowing, it is about what is responsible for setting the
amount of current. The load determines the current that flows in
or out
of the load and is independent of the opamp characteristics or the
power
supply arrangement. What I can control by picking the opamp is the
current that flows through the opamp that is independent of the
load or
input.


But the direction is important, also. You wanted to know how much
the
op amp itself will draw; when sourcing the load, you will find more
current on Vcc then when the op amp is sinking the load.4

No, I am not asking what the opamp will "draw". I'm asking about low
power amplifiers. I'd be happy knowing how it is done in the chips in
the radio controlled clocks since I'm pretty sure I'm not going to
find
a standard opamp that will do this.


You can; there are a number with that match your requirements. But you
need to design the entire circuit around your requirements, not just
the
op amp.

I only found one that was even close. As I said, the specs listed in
the selection guilds were all in error, usually by three orders of
magnitude.


Interesting - I don't know which ones you checked, but I've never found
that many errors in Mouser's data sheets. Not to say there aren't
errors - I have found several over the years. But not the high
percentage you've found.

It is not a high percentage. There are over 10,000 amps in Mouser's
list. I found around half a dozen or so that were listed incorrectly.
Search for amps that are under 40 uA quiescent current and GBW of 6 or
more.


It's a high percentage of those with the specs you need, which was my point.


But when the op amp is sourcing current, the amplifier has to
provide
the source current plus the drive current. Now you can bias the
output
so that the op amp is always sinking, but then you have a steady
drain
from the standby current.

When you're using a bipolar supply and input is at ground, the
output
will also be at ground, and very little current will be flowing.

You can create an artificial ground at 1/2 Vcc, but even creating
that
artificial ground draws current.

The difference in current drain is not important in the vast
majority of
cases. But they become important when you're talking the low
drain you
wish.

None of this is relevant to the issue of picking an amplifier.


It is if you're trying to minimize power usage to the maximum
extent. I
admit my op amp theory is around 40 years old, but I don't think the
laws of physics have changed in that time

No, and the laws of relevancy haven't changed either. The power
consumed by the load doesn't impact my selection of amplifier.


In many cases, it really doesn't matter much. But it does, when you're
down in the microamp range. That's what you don't understand.

Ok, thanks for your comments.







--
==================
Remove the "x" from my email address
Jerry, AI0K

==================