Searching \ for '[EE] Opamp design doubt' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=opamp+design+doubt
Search entire site for: 'Opamp design doubt'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Opamp design doubt'
2010\05\10@115900 by yamanoor sairam

picon face
Dear All,
           I would like to thank Mr.Russell McMahon, Mr. Marcel Du Champ and Mr.Sergio for their kind help. I was successfully able to develop the PIC based ORP circuit. I am sorry that I was not able to thank them earlier since I have a very restricted access to the internet.

I am currently trying to use the same op amp circuit to modify such that I am able to read signals from probe between -2V to +2V. I wanted to level shift the signal such that I am able to read the signal against a scale of 0-4V. I removed the 40K resistor and I replaced the same with a 10K resistor such that my opamp gain is 2 in the first stage. Now when I set an offset voltage such that my output is 4V for a -2V input, I see that I am not able to find linearity in the output. When I provide a 2V input to my opamp, the output reads zero. When I provide a -1V input to my opamp, the output is close to 3.81V and 4.1V for     -2V. Have I adopted a wrong technique for level shifting my signal? Please find the circuit diagram in the link given below:
http://mechatronicscraze.files.wordpress.com/2010/03/orp-circuit1.jpg

Please correct me if I am wrong.

Yours Sincerely,
Sairam


     

2010\05\10@121859 by Marcel Duchamp

picon face
On 5/10/2010 8:58 AM, yamanoor sairam wrote:
{Quote hidden}

You are very close to the answer.  Try removing R9 - this will change
the gain of the input amp from +2V to +1V.  With an offset adjust value
around +1V, you should get what you are looking for.

Why the change?  You have an input voltage range of -2V to +2V ==> 4V
span.  You want an output span of +4V to 0V ==> 4V span also.  Thus, you
need a system gain of -1 which your second stage already has.

2010\05\10@122958 by Marcel Duchamp

picon face
I wrote:
> You are very close to the answer.  Try removing R9 - this will change
> the gain of the input amp from +2V to +1V.  With an offset adjust value
> around +1V, you should get what you are looking for.
>
> Why the change?  You have an input voltage range of -2V to +2V ==>  4V
> span.  You want an output span of +4V to 0V ==>  4V span also.  Thus, you
> need a system gain of -1 which your second stage already has.

Of course, I meant
"Try removing R9 - this will change the gain of the input amp from +2 to
+1."

2010\05\10@131433 by Olin Lathrop

face picon face
yamanoor sairam wrote:
> I am currently trying to use the same op amp circuit to modify such
> that I am able to read signals from probe between -2V to +2V. I
> wanted to level shift the signal such that I am able to read the
> signal against a scale of 0-4V. I removed the 40K resistor
> http://mechatronicscraze.files.wordpress.com/2010/03/orp-circuit1.jpg

There is much wrong here.  The most serious problem is that these opamps are
intended to run from a wider supply range, and have rather large headroom
requirements for both input and output.  The datasheet is vague on what
happens when you power them at only +-5V, but at +-15V they require 4V
headroom for input signals and 3V for output.  This is not likely to get
better with lower supply voltage.  You therefore should figure these amps
are good for +-1V input and +-2V output in your application.

As for the 40K resistor (I assume you mean R16), that together with R9 sets
the gain of the input stage.  Removing it makes the amp function as a
comparator around 0.

It's obvious you're just guessing how this circuit works.  You can't make
meaningful changes until you do.  First I suggest ditching the TL072 in
favor of something with less headroom required, such that it can do useful
things with +-5V supply.  Fortunately there are many opamps that fit that
requirement.  Next, find a basic text on electronics that covers opamps and
read it.  Don't do anything until you understand how your original circuit
was supposed to work (ignoring the headroom limitations of the TL072).  Then
read the datasheet of whatever opamp you chose carefully.  Only then are you
ready to make modifications to this circuit.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\10@131953 by Olin Lathrop

face picon face
Marcel Duchamp wrote:
> Of course, I meant
> "Try removing R9 - this will change the gain of the input amp from +2
> to +1."

Try again.  The original gain of the input amp is +5.

Also just removing R9 won't make a unit gain stage, but a oscillator.  TL072
are not unity gain stable, despite what some sources claim.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\10@133044 by Russell McMahon

face picon face
> > You are very close to the answer.  Try removing R9 - this will change
> > the gain of the input amp from +2V to +1V.  With an offset adjust value
> > around +1V, you should get what you are looking for.
> >
> > Why the change?  You have an input voltage range of -2V to +2V ==>  4V
> > span.  You want an output span of +4V to 0V ==>  4V span also.  Thus, you
> > need a system gain of -1 which your second stage already has.
>
> Of course, I meant
> "Try removing R9 - this will change the gain of the input amp from +2 to
> +1."

If the P version of the opamp  is essentially the same as this

                      http://www.st.com/stonline/books/pdf/docs/2298.pdf

then it's output is not guaranteed to swing to within more than about
2 volts of either supply at +/- 5V supply.
How closely it does swing when you drive it into 'forbidden' areas may
relate to how much overdrive you provide, leading to unpredicatble and
non-linear effects.

Start off running it on say 2 x 9V batteries to get the circuit
working correctly and then investigate the implications of lower power
supply voltages.

You have a rail to rail input amplifier but not R2R output.
In this case, if you only have +/- 5 Volt available you want R2R on
input and output.


  Russell McMahon.

2010\05\10@133102 by Mike Hagen
picon face
Microchip make some really nice op amps and programmable gain amps.
Singles, duals, quads.
I have had good luck with several of them.  Look for low voltage/low
power operation with close to rail to rail outputs and inputs.
Don't even have to leave Microchip.com!


On 5/10/2010 10:14 AM, Olin Lathrop wrote:
{Quote hidden}

2010\05\10@141643 by Olin Lathrop

face picon face
Mike Hagen wrote:
> Microchip make some really nice op amps and programmable gain amps.
> Singles, duals, quads.
> I have had good luck with several of them.  Look for low voltage/low
> power operation with close to rail to rail outputs and inputs.
> Don't even have to leave Microchip.com!

The Microchip opamps are great when you need something that can run off the
same supply as a PIC.  However, they are all limited to 6V power, so
wouldn't work with the OP's +-5V power.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\10@141808 by Dwayne Reid

flavicon
face
At 11:19 AM 5/10/2010, Olin Lathrop wrote:

>Also just removing R9 won't make a unit gain stage, but a oscillator.  TL072
>are not unity gain stable, despite what some sources claim.

Hmm - this looks wrong.  I've been using TL07x as unity gain buffers
for decades now.

Just looking now: TI's data sheet
<http://focus.ti.com/lit/ds/symlink/tl071.pdf> says that the chips
are unity-gain stable.  See page 9 of the PDF file I referenced above.

dwayne

--
Dwayne Reid   <spam_OUTdwaynerTakeThisOuTspamplanet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(780) 489-3199 voice          (780) 487-6397 fax
http://www.trinity-electronics.com
Custom Electronics Design and Manufacturing

2010\05\10@142727 by Mike Hagen

picon face
Since you state -2V to +2V input and want a 0-4V output, you have to
have an overall gain of 1, no matter how many op amps.
4V change on input, 4V change on output.

If you want the same direction (sign of gain), you are going to have to
add another inverted gain stage?
If you are going to read it with a Pic, you can fix the inversion there?

I am assuming you want the input on the non-inverting side of an op amp
for high impedance?

Keep going...
Mike

2010\05\10@152659 by Olin Lathrop

face picon face
Dwayne Reid wrote:
> At 11:19 AM 5/10/2010, Olin Lathrop wrote:
>
>> Also just removing R9 won't make a unit gain stage, but a
>> oscillator.  TL072 are not unity gain stable, despite what some
>> sources claim.
>
> Hmm - this looks wrong.  I've been using TL07x as unity gain buffers
> for decades now.
>
> Just looking now: TI's data sheet
> <http://focus.ti.com/lit/ds/symlink/tl071.pdf> says that the chips
> are unity-gain stable.  See page 9 of the PDF file I referenced above.

Yes, that's why I say "despite what some sources claim".  I have seen these
things oscillate in some cases.  Maybe those weren't genuine TI parts.  If
you specify genuine TI parts then you should be able to use them as unity
gain buffers.  However, there are various second sources out there.  If you
allow your purchasing people to pick where they get these amps from and who
makes them, you'd be better off not relying on unity gain stability.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\10@192248 by Marcel Duchamp

picon face
On 5/10/2010 9:18 AM, Marcel Duchamp wrote:

> You are very close to the answer.  Try removing R9 - this will change
> the gain of the input amp from +2 to +1.  With an offset adjust value
> around +1V, you should get what you are looking for.
>
> Why the change?  You have an input voltage range of -2V to +2V ==>  4V
> span.  You want an output span of +4V to 0V ==>  4V span also.  Thus, you
> need a system gain of -1 which your second stage already has.

Sairam, by now, you have seen several conflicting suggestions.  I still
stand by what I wrote.  I knew of the common mode range versus power
supply levels limitation but it was my impression that you were
successful with using it with the ORP application.  That may have been
incorrect on my part.  But if you were successful before using it then
it should work here as well.  It is something for you to look into to
verify correct operation.

The data sheet is quite clear about unity gain operation.  There should
be absolutely no problem with that.  Removing R9 will indeed make the
first stage function with a voltage gain of unity.

2010\05\11@030637 by yamanoor sairam

picon face
Dear All,

I thank you very much for the support you have extended. I must thank Mr.Marcel Duchamp, Mr.Russell McMahon, Mr. Dwayne Reid, Mr.Mike Hagen, Mr. Olin Lathrop for their opinion and suggestions. I was able to test the output of the opamp for the range of -2V to +2V. When I removed the resistor R9 and adjusted the offset voltage to 1V, I was able to infer that the circuit spans between -2V to 2V and my output is 0-4V.

However, I tried to understand the concept of offset voltage in an opamp from the link below:
http://www.ecircuitcenter.com/Circuits/op_voff/op_voff2.htm

I am not able to understand the concept entirely. Can anyone throw some light on this?

I must thank Mr.Marcel for having helped me through out the development. I am still studying the stability of the ORP system.

I must thank Mr.Russell and Mr. Olin for their opinion on the stability of the TL07X amplifiers and I shall ensure that I check their findings completely.

Mr.Olin also pointed out the approach that I have taken for designing opamp circuits. I regret that I have a very poor knowledge about opamps and I am trying to improve myself by designing a few circuits from scratch.

Mr.Olin, would you mind recommending a book like this to me?
http://books.google.co.in/books?id=0zqU01lKPCEC&dq=Op+Amps+For+Everyone&printsec=frontcover&source=bn&hl=en&ei=NgDpS5T-M5O1rAez0fnzCQ&sa=X&oi=book_result&ct=result&resnum=4&ved=0CCUQ6AEwAw#v=onepage&q&f=false

Mr. Mike, thanks for the suggestion on Microchip op amps.  

Thank you.

Yours Sincerely,
Sairam



________________________________
From: Marcel Duchamp <.....marcel.duchampKILLspamspam@spam@sbcglobal.net>
To: Microcontroller discussion list - Public. <piclistspamKILLspammit.edu>
Sent: Tue, May 11, 2010 4:52:37 AM
Subject: Re: [EE] Opamp design doubt

On 5/10/2010 9:18 AM, Marcel Duchamp wrote:

> You are very close to the answer.  Try removing R9 - this will change
> the gain of the input amp from +2 to +1.  With an offset adjust value
> around +1V, you should get what you are looking for.
>
> Why the change?  You have an input voltage range of -2V to +2V ==>  4V
> span.  You want an output span of +4V to 0V ==>  4V span also.  Thus, you
> need a system gain of -1 which your second stage already has.

Sairam, by now, you have seen several conflicting suggestions.  I still
stand by what I wrote.  I knew of the common mode range versus power
supply levels limitation but it was my impression that you were
successful with using it with the ORP application.  That may have been
incorrect on my part.  But if you were successful before using it then
it should work here as well.  It is something for you to look into to
verify correct operation.

The data sheet is quite clear about unity gain operation.  There should
be absolutely no problem with that.  Removing R9 will indeed make the
first stage function with a voltage gain of unity.

2010\05\11@034239 by Russell McMahon

face picon face
The following addresses the moderately important issue re data sheet
reliability and vendor reliability.

>>> TL072 are not unity gain stable, despite what some
>>> sources claim.

>> Hmm - this looks wrong.  I've been using TL07x as unity gain buffers
>> for decades now.

>> Just looking now: TI's data sheet
>> <http://focus.ti.com/lit/ds/symlink/tl071.pdf> says that the chips
>> are unity-gain stable.  See page 9 of the PDF file I referenced above.

> Yes, that's why I say "despite what some sources claim".  I have seen these
> things oscillate in some cases.  Maybe those weren't genuine TI parts.

Digikey list only TI, ST and Diodes Inc parts.

           <http://www.diodes.com/datasheets/TL072.pdf>
           <www.st.com/stonline/books/pdf/docs/2298.pdf>
           <http://focus.ti.com/lit/ds/symlink/tl071.pdf>

all give information on unity gain operation.
This seems to be part of the formal TL072 specification.
It would SEEM that *any reputable* TL072 part should work in this mode.
ie datasheet says so, assume it's true.

> If you allow your purchasing people to pick where they
> get these amps from and who makes them, you'd be
> better off not relying on unity gain stability.

Maybe you did :-).

I'm not actually 100% sure what the above advice means as I'd hope
that with a properly run purchasing system one should be able to rely
on one's purchasing people to do just that. ie they may not be able to
write programs or design hardware but a professionally run purchasing
department can perform almost equally impressive miracles in their own
specialist area. Needless to say, many fall short of that mark. In
another lifetime I worked for a major corporate, where being totally
in control of purchasing was a major aim - and one of the companies
that I deal with in China also aims at providing consistent component
supply in that environment - a good trick indeed if you can do it.

In a US environment you'd hope that as long as you bought known-label
parts from reputable suppliers that you could rely on data sheets.
There are indeed occasional major blowouts in the system (eg bad
motherboard caps of some years ago, known fake smps ICs of certain
sorts, some Motorola power transistors being widely faked  etc), but
overall if you can't trust the system to a reasonable extent then data
sheets become utterly meaningless. If you start mistrusting data
sheets to the stage where you recommend that a major specification not
be relied on then you may equally adjudge a given PIC as unlikely to
run at full rated clock speed, or to meet sleep current specs or to
provide claimed RC oscillator stability over temperature etc. Where
would you then draw the line ?



   Russell

2010\05\11@034715 by yamanoor sairam

picon face
Dear All,

I have a question regarding the calibration of probes. As the ORP probes age, they tend to lose their accuracy of reading due to the aging of probes. The ORP probe's accuracy is checked using solutions that have potentials in the range of +200mV, +400mV etc.


Can I adjust the offset voltage using a precision trim pot to make small adjustments in the offset voltage? Should I make minor adjustments in my code for offsets in voltage?

Yours Sincerely,
Sairam



________________________________
From: Marcel Duchamp <.....marcel.duchampKILLspamspam.....sbcglobal.net>
To: Microcontroller discussion list - Public. <EraseMEpiclistspam_OUTspamTakeThisOuTmit.edu>
Sent: Tue, May 11, 2010 4:52:37 AM
Subject: Re: [EE] Opamp design doubt

On 5/10/2010 9:18 AM, Marcel Duchamp wrote:

> You are very close to the answer.  Try removing R9 - this will change
> the gain of the input amp from +2 to +1.  With an offset adjust value
> around +1V, you should get what you are looking for.
>
> Why the change?  You have an input voltage range of -2V to +2V ==>  4V
> span.  You want an output span of +4V to 0V ==>  4V span also.  Thus, you
> need a system gain of -1 which your second stage already has.

Sairam, by now, you have seen several conflicting suggestions.  I still
stand by what I wrote.  I knew of the common mode range versus power
supply levels limitation but it was my impression that you were
successful with using it with the ORP application.  That may have been
incorrect on my part.  But if you were successful before using it then
it should work here as well.  It is something for you to look into to
verify correct operation.

The data sheet is quite clear about unity gain operation.  There should
be absolutely no problem with that.  Removing R9 will indeed make the
first stage function with a voltage gain of unity.

2010\05\11@073400 by Olin Lathrop

face picon face
Marcel Duchamp wrote:
> But if you were successful before using it then
> it should work here as well.

Wrong.  One instance of something working out of spec is no proof of
anything except of that one item working.  The next unit could react quite
differently.  Also, did you test the one "working" unit over the full
temperature range, phases of the moon, and proximity levels to dead fish?


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\11@075615 by Olin Lathrop

face picon face
yamanoor sairam wrote:
> However, I tried to understand the concept of offset voltage in an
> opamp from the link below:

First you have to understand how a ideal opamp works.  It takes the voltage
on the + input minus the voltage on the - input, then multiplies that by a
large number (the gain).  Often you consider the gain of a ideal opamp to be
infinite.  You get useful linear results with this by using feedback.

So with a perfect opamp, if the - input is a 1.000 volts and the + input is
at 0.999 volts, then the output will slam to the low supply rail.  If the +
input then moves to 1.001 volts, then the output slams to the high supply
rail.  For a ideal opamp, this is still true for the + input being 1.0001,
1.00001 and 1.000001 volts since those are all greater than the 1.0 volts
the - input is being held at.

However, real opamps have a error tolerance when comparing the + and -
inputs.  This is called the "input offset voltage", and is usually a few
millivolts.  Let's say your opamp has a 2mV input offset voltage.  If the -
input is still held at 1.000 volts, then you don't know what the amp will do
if the + input is within 2mV of that, meaning from 0.998 to 1.002 volts.
The + input has to be outside that range for the opamp to reliably determine
that it is higher or lower than the - input.

The equation for a ideal opamp is

 Out = Gain(PosIn - NegIn)

When the input offset voltage of a real opamp is taken into account it
becomes:

 Out = Gain(PosIn - NegIn + Err)

Where Err can vary from amp to amp, over temperature, and possibly other
parameters.  You don't know ahead of time what Err is except that it is
bounded to be within the input offset voltage.  In the example above, Err
can therefore be anywhere from -2mV to +2mV.

How the input offset voltage effects various circuits and how to deal with
it are long topics beyond the scope of this post.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\11@080053 by Olin Lathrop

face picon face
Russell McMahon wrote:
> If you start mistrusting data
> sheets to the stage where you recommend that a major specification not
> be relied on then you may equally adjudge a given PIC as unlikely to
> run at full rated clock speed, or to meet sleep current specs or to
> provide claimed RC oscillator stability over temperature etc. Where
> would you then draw the line ?

You're right, I should not have brought up the unit gain stability issue.

I had a bad experience with TL072 in this role many years back, but at this
point I can't remember enough specifics to give any useful advice.  I
apologize for the confusion.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\11@080814 by Olin Lathrop

face picon face
yamanoor sairam wrote:
> Can I adjust the offset voltage using a precision trim pot to make
> small adjustments in the offset voltage? Should I make minor
> adjustments in my code for offsets in voltage?

No matter how you adjust of any offset error in the system, you first have
to supply a known calibrated input so that the offset voltage can be
measured.  If the measurement is processed in a digital system like a PIC,
my first preference is to deal with it there.  You could store the
correction value in EEPROM so that it is retain accross power down, like a
trimpot setting would be.

The reason I prefer not to add a trimpot is because these cost real money,
take real board space, can add noise, drift, can get bumped, or diddle by
idle fingers.  Of course for the digital correction there has to be some
communication means to tell the PIC a calibrated input is being presented.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\11@111143 by William \Chops\ Westfield

face picon face
>> offset voltage

Isn't this particular application (-2 to 2V yields 0 to 4V) a classic  
example of an "analog adder" for which there is a standard circuit?  
Any op-amp book should yield a decent set of circuits aimed at analog  
computing rather than "amplification."

BillW

2010\05\11@125452 by Rich

picon face
The circuit you show is for an op amp without offset adjustment pins.  Think
of the gain of the op amp as the slope of a linear curve of the form y=mx+b.
The slope m, is the gain which is set by the ratio of feedback to input
resistors.  The offset is the y intercept (b) of the linear curve which is
determined by the voltage at the non-inverting input.  By applying a voltage
to the non-inverting input the slope (gain) is unaffected but it is
referenced to the y intercept (b), which is the offset voltage applied to
the non-inverting pin.



{Original Message removed}

2010\05\11@130607 by Rich

picon face
If you show the op amp circuit configuration I will be happy to explain how
to do this.


----- Original Message -----
From: "yamanoor sairam" <yamanoorsaispamspam_OUTyahoo.com>
To: "Microcontroller discussion list - Public." <@spam@piclistKILLspamspammit.edu>
Sent: Tuesday, May 11, 2010 3:47 AM
Subject: Re: [EE] Opamp design doubt


{Quote hidden}

> --

2010\05\11@155719 by Rich

picon face
That is true, Russell.  In fact most op amps configured as unity gain are
not stable near zero.  That is why there are special very high open loop
gain devices called buffers or followers, like the LM 310, for example.


{Original Message removed}

2010\05\11@194827 by Sean Breheny

face picon face
Hi Rich,

Not stable near zero what? I don't understand your comment. Near zero
volts output?

Sean


On Tue, May 11, 2010 at 3:56 PM, Rich <spamBeGonergrazia1spamBeGonespamrochester.rr.com> wrote:
> That is true, Russell.  In fact most op amps configured as unity gain are
> not stable near zero.  That is why there are special very high open loop
> gain devices called buffers or followers, like the LM 310, for example.
>
>

2010\05\11@203730 by Gerhard Fiedler

picon face
Olin Lathrop wrote:

> yamanoor sairam wrote:
>> However, I tried to understand the concept of offset voltage in an
>> opamp from the link below:
>
> First you have to understand how a ideal opamp works.  It takes the
> voltage on the + input minus the voltage on the - input, then
> multiplies that by a large number (the gain).  Often you consider the
> gain of a ideal opamp to be infinite.  You get useful linear results
> with this by using feedback.

Rather than working with the mental image of an infinite gain, I found
it quite useful in my early days to think of the opamp as a device that
sets its output so that both inputs are equal. This image has its
limitations, but it works well for circuits that don't run into
saturation and are mostly DC. It helped me a great deal understanding
the different circuits. (This still works when adding offset voltage and
current.)

Gerhard

2010\05\11@231248 by Rich

picon face
Sean, I may have been a bit misleading.  Thank you for catching it.  I meant
to say near zero volts output, as you pointed out.  Most op amps do not have
the very high open loop gain, which unity gain buffers do have.  The buffer
example I gave was for the LM310. The only spec I could find on open loop
gain was was super OLG.   Op amps can be configured as unity gain easily by
connecting the output directly to the inverting input.  Sometimes resistors
are placed in the feedback loop and in the non-inverting input to balance
the currents.  That configuration is generally stable for relatively large
signal, and the higher the open loop gain of the device the better.  At
small signals I believe the jth valley currents can contribute to drift.
When one is operating in the low millivolts (near zero) there can develop a
small drift band which is okay if your measurement is not small enough to be
compromised.  Of course, we are talking about a DC configuration. The OP37
is a precision op amp which is fairly reliable as a unity gain buffer.  It
has a 1.8 million open loop gain. Drift is 0.2 uV/C where the LM310 is 10
uV/C so it is really stable.   Even the long term drift spec is good.  I
have designed amplifiers that operate in the microvolts and even the board
leakage can cause drift and affect the performance so the offset currents
need to be balanced.  I designed a 50 femtoamp amplifier once and I could
not even mount the parts in a PCB.  So I have become keenly aware of the
problems of small signal and drift and configuration.


{Original Message removed}

2010\05\12@004331 by Sean Breheny

face picon face
Thanks for explaining, Rich - that makes sense. I think that the
stability which Olin was talking about was in the sense of control
loop stability, not stability of offset voltage. I have never designed
an amplifier which works with 50 fA current input (!) before, but I
can appreciate how lots of things we normally take for granted must be
questioned at those signal levels!

Sean


On Tue, May 11, 2010 at 11:12 PM, Rich <TakeThisOuTrgrazia1EraseMEspamspam_OUTrochester.rr.com> wrote:
{Quote hidden}

> {Original Message removed}

2010\05\12@011941 by Rich

picon face

It was the first time I worked that low.  It was one of those "nightmares"
that you can enjoy, just for the challenge :-)


{Original Message removed}

2010\05\12@015011 by Russell McMahon

face picon face
> >> However, I tried to understand the concept of offset voltage in an
> >> opamp from the link below:

> Rather than working with the mental image of an infinite gain, I found
> it quite useful in my early days to think of the opamp as a device that
> sets its output so that both inputs are equal. This image has its
> limitations, but it works well for circuits that don't run into
> saturation and are mostly DC.

This is an EXTREMELY powerful and useful concept.
If a circuit's feedback loops (ie circuitry) are correctly configured
then for practical purpose the inverting and non inverting inputs are
at the same voltage. In practice there will be a VERY small difference
due to the finite gain of the amplifier but for practical purposes -
if you can measure a voltage between the two inputs then there is a
fault condition of some sort.

_______________

The other very useful 'rules' are as below:
Known as  Kirchhoff's voltage and current laws.
(He has other laws which electrical engineers tend to be less aware of)

     http://en.wikipedia.org/wiki/Kirchhoff's_circuit_laws

Applying these to many circuits allows useful calculations to be done
re what voltages exist etc.

- The sum of currents into a passive circuit node (point) is zero.

ie current in and out must balance or charge would accumulate or drain
away. I said "... passive ..." as their are conditions where charge
does in fact accumulate or drain away. Connections to cnd inductors
are two examples. Also some active circuitry, but in all cases, the
net current flow will lead to a voltage change somewhere to maintain
stability and long term this is not maintainable.

- Voltages summed around a loop sum to zero.

eg A circuit may consist of a battery a load resistor and a MOSFET in a "loop".

Starting at ground and going 'up' in voltage through the battery and
down in voltage steps via battery leads, connector resistance, load
resistor, MOSFET on resistance etc and back to ground the sum of ups
and down in voltage will sum to zero. If this were not so the voltage
at the point could seem to vary just by proposing different loops
which included it.

Both the above are fundamental to basic circuit understanding and are
extremely useful.


         Russell.

2010\05\12@083105 by Olin Lathrop

face picon face
Gerhard Fiedler wrote:
> Rather than working with the mental image of an infinite gain, I found
> it quite useful in my early days to think of the opamp as a device
> that sets its output so that both inputs are equal.

I think this is a bad way to teach opamps and deliberately avoid it.  The
problem is that this rule of thumb sounds deceptively simple and
authorotative, and people don't think about the limitations when they apply
it.  To know when to apply it, you need to understand opamps to the point
where this crutch is no longer necessary.

One of the questions I ask when interviewing EE candidates is to draw a
simple opamp circuit with hysteresis.  It's just the opamp and two resistors
(output to +, and + to ground).  I tell them this is a ideal opamp powered
from -5 and +5 volts, and both resistors are 10Kohms.  The problem is to
explain what happens as the input ramps from -5 to +5 volts.  You'd be
surprised how many people mess this up, and how many of those mumble about
the opamp keeping the two inputs at the same voltage.

Some opamp *circuits* will try to keep the - input at the same voltage as
the + input.  The opamp itself merely performs:

 Out = Gain(PosIn - NegIn)

If you can't understand how that causes the circuit to have the - input
follow the + input in some cases, you won't be able to know which cases that
applies to and which cases it doesn't.  In other words, the real rule isn't
any harder to understand than when it's OK to apply the simple rule.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\12@085454 by Olin Lathrop

face picon face
Sean Breheny wrote:
> I think that the
> stability which Olin was talking about was in the sense of control
> loop stability, not stability of offset voltage.

Right.  "Unity gain stable" refers to the the opamp being stable (not
oscillating) in a unity gain configuration.  The way such things work, the
higher the gain the lower the negative feeback and the more "hyper" the
opamp can be before it becomes unstable as a control loop.  Not all opamps
are specified as unity gain stable.  Those that aren't usually have a spec
"stable down to xxx gain".

Basically the opamp needs to be more damped to be stable at unity gain.
Therefore you give up some bandwidth if running the opamp at a higher gain.
Everything is a tradeoff, so some opamps are built with the higher bandwidth
capability at the expense of not being usable in low gain configurations.
Some opamps, particular from long ago, have "compensation" pins that you put
a capacitor accross to set your own tradeoff.

Some of the Microchip opamps are good examples of this tradeoff.  The
MCP604x series is specifically mentioned to be unity gain stable.  It has a
typical gain*bandwidth product of 14KHz.  The MCP614x series is basically
the same thing but only stable for gains of 10 or more, and the datasheet
mentions this distinction prominently.  In return you get 100KHz
gain*bandwidth.  I strongly suspect the 604x and 614x are the same design
with different internal compenstation setting.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\12@091023 by Michael Watterson

face picon face
Olin Lathrop wrote:
>
>
> One of the questions I ask when interviewing EE candidates is to draw a
> simple opamp circuit with hysteresis.  It's just the opamp and two resistors
> (output to +, and + to ground).  I tell them this is a ideal opamp powered
> from -5 and +5 volts, and both resistors are 10Kohms.  The problem is to
> explain what happens as the input ramps from -5 to +5 volts.  
But I'd mentally think of that as a comparator, implementing a Schmitt
trigger.

An op-amp is only an op-amp if it has predominantly negative  feedback
that  gives  a closed loop  gain a tiny fraction of open loop gain.
Otherwise it's just an IC "gain block".

To me and Op Amp is NOT a TL 072 or 741. Those are high gain
differential input amplifiers that let you build an  op-amp.

But what ever you call it, Alice and HD ought to spot you have drawn a
Schmitt trigger, not a Linear Amplifier of accurately defined gain (
Operational Amplifier). I have an old valve based op-amp schematic some
place...

It's odd that some people don't spot that though certain ICs are sold
for "op-amp" or "comparator" applications, in some cases an open
collector/open drain comparator might be a better "op-amp" and if you
have a dual amp IC sold as "Op-Amp" 1 might be a x10 amp and the other a
comparator for a zero-crossing detector. Depending on surrounding circuit.

So maybe your post is a little pedantic and your EE candidates a bit
dumb if they don't catch that your description is misleading. Which
achieves your weeding out aim :)



2010\05\12@092501 by Russell McMahon

face picon face
>> Rather than working with the mental image of an infinite gain, I found
>> it quite useful in my early days to think of the opamp as a device
>> that sets its output so that both inputs are equal.

> I think this is a bad way to teach opamps and deliberately avoid it.

I don't :-).
I think it is useful PROVIDED the conditions under which it should
approach true are understood.
ie you can't just rote learn it - you need to understand WHY it is
often a good generalisation.

I said.

>> This is an EXTREMELY powerful and useful concept.
>> If a circuit's feedback loops (ie circuitry) are correctly configured ...

ie negative feedback is a necessity for this to be true. Understanding
why is crucial.

Negative feedback was not present in Olin's "comparator" (rather than
true opamp) example with expected results.

> The
> problem is that this rule of thumb sounds deceptively simple and
> authorotative, and people don't think about the limitations when they apply
> it.

Yes. As above. We agree :-).
But it's a useful tool. It allows you to eg apply Kirchoff (aka
applied common sense) as if things were wholly passive when the need
to do so applies.

> To know when to apply it, you need to understand opamps to the point
> where this crutch is no longer necessary.

As above. Point understood. Student needs at least Oak Leaves to handle it.

> One of the questions I ask when interviewing EE candidates is to draw a
> simple opamp circuit with hysteresis.  It's just the opamp and two resistors
> (output to +, and + to ground).  I tell them this is a ideal opamp powered
> from -5 and +5 volts, and both resistors are 10Kohms.  The problem is to
> explain what happens as the input ramps from -5 to +5 volts.  You'd be
> surprised how many people mess this up, and how many of those mumble about
> the opamp keeping the two inputs at the same voltage.

Yes. No Oak Leaves. This is however a vanishingly unusual way to use
an opamp. It's a comparator circuit, which an opamp will handle OK
enough if it has to, but there is no "amp" in the usual sense.

For extra points, give them a circuit with nagar\tive AND positive
feedback (eg finite gain opamp with hysteresis) and set them loose.


> Some

for most values of some,

> opamp *circuits* will try to keep the - input at the same voltage as
> the + input.  The opamp itself merely performs:
>
>  Out = Gain(PosIn - NegIn)

Yes. We don't, of course, disagree on the basic principles, just the
perspective.
Not unusual :-).


  Russell

2010\05\12@093432 by Russell McMahon

face picon face
> An op-amp is only an op-amp if it has predominantly negative  feedback
> that  gives  a closed loop  gain a tiny fraction of open loop gain.
> Otherwise it's just an IC "gain block".

> To me and Op Amp is NOT a TL 072 or 741. Those are high gain
> differential input amplifiers that let you build an  op-amp.

> But what ever you call it, Alice and HD ought to spot you have drawn a
> Schmitt trigger, not a Linear Amplifier of accurately defined gain (
> Operational Amplifier). I have an old valve based op-amp schematic some
> place...

For extra points ... :-).
Turn a Scmitt trigger into a linear amplifier, almost :-).

The use of CMOS "digital" inverters as linear amplifiers is well
known. eg place a 1M resistor from input to output with a 100k input
resistor from a lowish impedance source and you have a ~~~= 10x linear
amplifier, subject to some real world stuff.

BUT take an eg 74C14 / CD40106 / ... or similar Schmitt trigger
inverter and use the same circuit. At audio frequencies linear AC gain
appears to happen. Observe output with an oscilloscope and be suitable
impressed. The gate produces PWM at as high a frequency as it can
manage given stray capacitance and this is usually high enough a
frequency to allow "audio" to be produced as is. Add a sniff of RC
output filter to get rid of the switching noise. Add a few pF at the
input to bring the frequency down into the gate's intended switching
range. Quite a fun result, although not something you'd usually expect
to do with a Schmitt inverter.



          Russell

2010\05\12@093727 by Olin Lathrop

face picon face
Michael Watterson wrote:
> But I'd mentally think of that as a comparator, implementing a Schmitt
> trigger.

Right.  That's the point.  Too many EE candidates apply the linear circuit
rules without thinking whether they actually have a linear circuit.  Even
worse, some don't realize that their rule only applies to a limited set of
circuits and just blindly follow the inapplicable rule.

> An op-amp is only an op-amp if it has predominantly negative  feedback
> that  gives  a closed loop  gain a tiny fraction of open loop gain.

No, that's just one way to use a opamp.  A opamp is merely a circuit
building block that (in the ideal case) does:

 Out = Gain(PosIn - NegIn)

The rest is how you use that basic capability within your circuit.  The
basic opamp building block, however, remains just that.  It works the same
way regardless how the external circuit utilizes its capabilities.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\12@094728 by Olin Lathrop

face picon face
Russell McMahon wrote:
> This is however a vanishingly unusual way to use
> an opamp.

For some values of "unusual".  However, the point is to see how a candidate
thinks about electronics, not just to test their ability to fill in knee
jerk forumlas.

Of course this is just one test of several during a technical interview.
All tests however are designed to identify those that really undestand
electronics.  I've got no use for knee jerkers.  Let them go work for
Ratheon or someplace, as long as it's somplace else.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\12@103240 by Michael Watterson

face picon face
Russell McMahon wrote:
>
> For extra points ... :-).
> Turn a Scmitt trigger into a linear amplifier, almost :-).
"How do I get to Skibbereen?"  I ask

"Ah Sur", the famer mumbles, "Sure, you'd not be wanting to start from
here."

By removing the 74HC14 and putting in a 74HC04? (or some such similar  
action)


2010\05\12@104929 by Spehro Pefhany

picon face
At 10:31 AM 12/05/2010, you wrote:
>Russell McMahon wrote:
> >
> > For extra points ... :-).
> > Turn a Scmitt trigger into a linear amplifier, almost :-).
>"How do I get to Skibbereen?"  I ask
>
>"Ah Sur", the famer mumbles, "Sure, you'd not be wanting to start from
>here."
>
>By removing the 74HC14 and putting in a 74HC04? (or some such similar
>action)

_74HCU04_ not 74HC04

>Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
RemoveMEspeffspamTakeThisOuTinterlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com



2010\05\12@112039 by Russell McMahon

face picon face
>> > For extra points ... :-).
>> > Turn a Scmitt trigger into a linear amplifier, almost :-).
>>"How do I get to Skibbereen?"  I ask

>>"Ah Sur", the farmer mumbles, "Sure, you'd not be wanting to start from
>>here."

   http://en.wikipedia.org/wiki/Skibbereen
   http://en.wikipedia.org/wiki/Skibbereen_(song)

>>By removing the 74HC14 and putting in a 74HC04? (or some such similar
>>action)
>
> _74HCU04_ not 74HC04

74HC14 not 74xxx04 anything.
Being digital is what it's about in this case.
FWIW the PWM could be used to input "digitised" analog to a processor,
should one wish.

I can't immediately think of a really good [tm] reason for doing this
in areal world situation if linear amplification was all that was
required, but if all you had available was a spare Schmitt gate this
technique may have it's place. And, it must be about as cheap an A2D
as you can get :-)


R

2010\05\12@114732 by Michael Watterson

face picon face
Russell McMahon wrote:
>
> 74HC14 not 74xxx04 anything.
> Being digital is what it's about in this case.
> FWIW the PWM could be used to input "digitised" analog to a processor,
> should one wish.
>
> I can't immediately think of a really good [tm] reason for doing this
> in areal world situation if linear amplification was all that was
> required, but if all you had available was a spare Schmitt gate this
> technique may have it's place. And, it must be about as cheap an A2D
> as you can get :-)
>
>
>  R
Well... It turns out that for say QAM16 etc RF linear amplifiers have to
be REALLY linear and typically are used at 1/10th of the -1dB CW
compression power limit.

A polar amplifier is the solution. You have a Class C RF amp driven by
phase modulated carrier and its power rail comes from a PWM PSU such
that the RF output level is defined by PWM duty cycle.

making this linear and putting delay on phse modulated path to match the
PWM latency and low pass filter delay of the Amplitude term controlling
PWM..

I think it can be done with digital delay  after a phase modulated NCO
in an FPGA driving Class C amp and PWM generated in the FPGA also, so
the PWM PSU is a pair of FET switches from 12V supply, PWM transformer
(plus rectifier and filter) gives then 5V to 90V to drive FET Class C RF
amp, driven almost direct by FPGA.
Train it with a sawtooth at base band, measure phase and amplitude of RF
via sampling, Zero IF mixers and ADC to FPGA to store a predistortion
lookup table and decide the delay need on phase output.

maybe the unused 74HC14 can go in it. Probably not.

But it's an illustration of how a purely on/of nonlinear system can be
arranged to be more linear than a linear amp. With in the case of  100W
RF amp, maybe saving 90W or more on PSU.


2010\05\12@155647 by Gerhard Fiedler

picon face
Olin Lathrop wrote:

> Gerhard Fiedler wrote:
>> Rather than working with the mental image of an infinite gain, I
>> found it quite useful in my early days to think of the opamp as a
>> device that sets its output so that both inputs are equal.
>
> I think this is a bad way to teach opamps and deliberately avoid it.
> The problem is that this rule of thumb sounds deceptively simple and
> authorotative, and people don't think about the limitations when they
> apply it.  To know when to apply it, you need to understand opamps to
> the point where this crutch is no longer necessary.
>
> One of the questions I ask when interviewing EE candidates is to draw
> a simple opamp circuit with hysteresis.  It's just the opamp and two
> resistors (output to +, and + to ground).  I tell them this is a
> ideal opamp powered from -5 and +5 volts, and both resistors are
> 10Kohms.  The problem is to explain what happens as the input ramps
> from -5 to +5 volts.  You'd be surprised how many people mess this
> up, and how many of those mumble about the opamp keeping the two
> inputs at the same voltage.

You didn't cite the crucial condition from my original email:

>> This image has its limitations, but it works well for circuits that
>> don't run into saturation and are mostly DC.

"Don't run into saturation" is the one here. This is a simple condition
that doesn't need an EE degree to understand.

I was 14 when I used this mental image, and I was no EE back then. It
did, however, allow me with what I knew about electronics to understand
how most /amplifier/ circuits are supposed to work. And it did in no way
make it more difficult for me to add deeper understanding about the
various non-ideal characteristics real-world opamps have.

> Some opamp *circuits* will try to keep the - input at the same voltage as
> the + input.  The opamp itself merely performs:
>
>   Out = Gain(PosIn - NegIn)

Well, no, it doesn't. If it did, several common circuits wouldn't work
as intended (including the Schmitt-Trigger you used in your example),
and my mental image would apply much wider than it does with actual
opamps.

The differences between the ideal output function you wrote and the real
one are for the most part the same that limit the applicability of my
mental image -- but these reasons are also what limits the applicability
of this simple formula. (OTOH, I was at 14 perfectly capable of applying
this mental image of a device that /tries/ to make both its inputs equal
to positive-feedback circuits like Schmitt-Triggers and understand how
they work. If an EE can't explain such a circuit, I think that /any/
mental image of it is probably lost on the person.)

> If you can't understand how that causes the circuit to have the -
> input follow the + input in some cases, you won't be able to know
> which cases that applies to and which cases it doesn't.  In other
> words, the real rule isn't any harder to understand than when it's OK
> to apply the simple rule.

I think you misunderstood the difference between the "real" rule and my
simplification.

1) What you call "real rule" is quite far away from a real-world useful
mathematical model of the transfer function of a real-world opamp. So
it's not the real rule, it is a simplification.

2) The /only/ thing it doesn't simplify is the gain. Everything else
that influences the transfer function is simplified out.

3) The /only/ thing my simplification simplifies compared to your
simplification is the gain. As long as the gain is high enough for the
circuit in question (and for Schmitt-Triggers, for example, you don't
need a very high gain for the circuit to work), your simplification and
my simplification are pretty much the same -- just looking at the issue
from different angles.

My simplification assumes an ideal opamp, with infinite gain. Your
transfer function assumes an ideal opamp, with a constant, given gain.
Both are not real, and for both you need to know the limitations of the
applicability of the simplification if you get anywhere near the limits.

Gerhard

2010\05\12@223607 by yamanoor sairam

picon face
Dear All,

Thanks for your contribution to this thread. Your posts were quite informative!

I used the circuit that was available at http://www.diy-labs.com


If you show the op amp circuit configuration I will be happy to explain
how
to do this.

I have posted a jpeg image of the same in the link below:
http://mechatronicscraze.wordpress.com/pic-orp/

I am currently experimenting what Mr.Olin had suggested on the calibration part.

Sairam




________________________________
From: Rich <rgrazia1EraseMEspam.....rochester.rr.com>
To: Microcontroller discussion list - Public. <EraseMEpiclistspammit.edu>
Sent: Tue, May 11, 2010 10:35:48 PM
Subject: Re: [EE] Opamp design doubt

If you show the op amp circuit configuration I will be happy to explain how
to do this.


{Original Message removed}

2010\05\12@234905 by Russell McMahon

face picon face
> http://mechatronicscraze.wordpress.com/pic-orp/
>
> I am currently experimenting what Mr.Olin had suggested on the calibration part.

If you do not also address the fact that you are operating the op-amp
outside it's guaranteed output voltage range it may cause you problems
in future. The fact that the circuit appears to be operating correctly
at present is 'lucky' at best.

If this is a 'one-off' and you are happy to fiddle with it to make it
work and to never be certain that it is working correctly, then using
the opamp as is may be acceptable to you. But if want to be certain
that the opamp is working inside the manufacturers guaranteed output
voltage range you need to either reduce output voltage swing to say
0-2V (loss of 1 bit of measurement resolution) or increase the +ve
supply voltage or further change the circuit.


 Russell

2010\05\13@022051 by yamanoor sairam

picon face
>But if want to be certain that the opamp is working inside the manufacturers guaranteed output
>voltage range you need to either reduce output voltage swing to say
>0-2V (loss of 1 bit of measurement resolution) or increase the +ve
>supply voltage or further change the circuit


Hello Mr. Russell,

Thanks for your mail.

Increase the + ve supply voltage alone? Did you mean that I can increase the voltage swing by increasing the +5V to +7V or greater?

All ORP probes and monitors operate over a range of +2V to -2V. I chose this amplifiers over other options such as LM356. What are the factors that I should consider while choosing an opamp?

As Mr.Olin had suggested, when I set the gain of the opamp using a pot for a pre determined offset voltage, the error should be corrected by storing the offset values in the EEPROM?

In future, if I am interested to check the calibration of the probe, should I write a code that find the offset and stores it in EEPROM?

Yours Sincerely,
Sairam



________________________________
From: Russell McMahon <RemoveMEapptechnzEraseMEspamEraseMEgmail.com>
To: Microcontroller discussion list - Public. <RemoveMEpiclistspam_OUTspamKILLspammit.edu>
Sent: Thu, May 13, 2010 9:18:45 AM
Subject: Re: [EE] Opamp design doubt

> http://mechatronicscraze.wordpress.com/pic-orp/
>
> I am currently experimenting what Mr.Olin had suggested on the calibration part.

If you do not also address the fact that you are operating the op-amp
outside it's guaranteed output voltage range it may cause you problems
in future. The fact that the circuit appears to be operating correctly
at present is 'lucky' at best.

If this is a 'one-off' and you are happy to fiddle with it to make it
work and to never be certain that it is working correctly, then using
the opamp as is may be acceptable to you. But if want to be certain
that the opamp is working inside the manufacturers guaranteed output
voltage range you need to either reduce output voltage swing to say
0-2V (loss of 1 bit of measurement resolution) or increase the +ve
supply voltage or further change the circuit.


 Russell

2010\05\13@075319 by Olin Lathrop

face picon face
yamanoor sairam wrote:
> Increase the + ve supply voltage alone? Did you mean that I can
> increase the voltage swing by increasing the +5V to +7V or greater?

This was explained in earlier posts by several people.  You need to read the
manufacturer's datasheet.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\13@080922 by Olin Lathrop

face picon face
Gerhard Fiedler wrote:
>>   Out = Gain(PosIn - NegIn)
>
> Well, no, it doesn't. If it did, several common circuits wouldn't work
> as intended (including the Schmitt-Trigger you used in your example),

No, the above equation shows what the opamp does in all cases, even when
used in a circuit with hysteresis.  I did leave out clipping at the power
rails, but thought it was understood a opamp can't generate a voltage
outside the range it is given.  So to be more correct (but more cluttered
and a little less easy to see the forest for the trees), here is what a
ideal opamp does:

 Out = min(PosPwr, max(NegPwr, Gain(PosIn - NegIn)))

In other words, it's the simple formula from earier with the result clipped
to the supply rails.  This applies to Schmitt trigger circuits just as well
as linear amplification circuits.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\13@083101 by Russell McMahon

face picon face
Sairam,

Russell said:
>>But if want to be certain that the opamp is working inside the manufacturers guaranteed output
>>voltage range you need to either reduce output voltage swing to say
>>0-2V (loss of 1 bit of measurement resolution) or increase the +ve
>>supply voltage or further change the circuit

___________

> Increase the + ve supply voltage alone? Did you mean that I can increase the voltage swing by increasing the +5V to +7V or greater?

Swing won't change, but it will 'fix' the problem.

It's not the SWING you want to change - you just need to ensure that
the opamp specification allow you to use the 'swing' that you already
have.
At present the op amp output goes within 1V of Vdd and the
specification sheet, while not as clear as it could be, very very
strongly suggests that this is an illegal condition for typical
versions of the IC. SOME may work there but others may not.

IF you can tolerate an eg 0-2 V output swing you can easily achieve it
very simply with the present system. Input is unity gain buffer =
Amplifier_1 = A1.
A2 connected also as unity gain buffer (inverting input to output).

Below R1 = R2 = some convenient & appropriate value.

A1 output via R1 to A2 + input.
A2+ input via R2 to +2 Volt point.
Vout = 1+ Vin/2.
So swing is halved and output is translated up by 1V.
For -2/+2 input you get 0-2V output.
You lose 1 bit resolution compared to 0-4V output.
You have to make +2V from some source. Or a higher voltage with a different R2.

> All ORP probes and monitors operate over a range of +2V to -2V. I chose this amplifiers over other options such as LM356. What are the factors that I should consider while choosing an opamp?

There are 2

- All datasheet parameters should never be violated under any
conditions of valid input.

- Other technical needs meet your need.

ie the present opamp does almost all this EXCEPT Vout is too high to
be safe based on the datasheets we have quoted so far. Adding 2V to
Vdd or possibly as little as 1V will cure this. Vdd = 7V looks good.

> As Mr.Olin had suggested, when I set the gain of the opamp using a pot for a pre determined offset voltage, the error should be corrected by storing the offset values in the EEPROM?
&
> In future, if I am interested to check the calibration of the probe, should I write a code that find the offset and stores it in EEPROM?

That's "up to you" - ie substituting programming for hardware
adjustment. An entirely valid option if it suits you.


  Russell

PS: Just fyi -
I know how hard it is to get language you aren't usd to correct. You
do very well. Re names -
'Mr Olin' & 'Mr Russell' is polite, but 'Olin' and 'Russell' is fine
in this context.
If you need to use "Mr" in a formal situation then use the last name
(surname, family name). eg Mr Lathrop or Mr McMahon - BUT just use
Olin and Russell here.

2010\05\13@134521 by Gerhard Fiedler

picon face
Olin Lathrop wrote:

> Gerhard Fiedler wrote:
>>>   Out = Gain(PosIn - NegIn)
>>
>> Well, no, it doesn't. If it did, several common circuits wouldn't work
>> as intended (including the Schmitt-Trigger you used in your example),
>
> No, the above equation shows what the opamp does in all cases, even
> when used in a circuit with hysteresis.  I did leave out clipping at
> the power rails, but thought it was understood a opamp can't generate
> a voltage outside the range it is given.  

You say "it was understood". If everything that is understood is really
understood, what was your point with your criticism in your previous
message? :)

> So to be more correct (but more cluttered and a little less easy to
> see the forest for the trees), here is what a ideal opamp does:
>
> Out = min(PosPwr, max(NegPwr, Gain(PosIn - NegIn)))
>
> In other words, it's the simple formula from earier with the result
> clipped to the supply rails.  This applies to Schmitt trigger
> circuits just as well as linear amplification circuits.

Again, it applies of course, but just as my mental image applies, with
the one additional condition that the gain must be high enough to not be
relevant. Which is given for the Schmitt-Trigger circuit.

Your point was that what I called my mental image has flaws that your
equation doesn't. You still have failed to provide one (except for an
anecdote of a totally incompetent EE for which both probably were not
suited).

And for the record, this is still far away from a real transfer function
of a real-world opamp. It should be obvious to you; you know enough
about opamps to be able to imagine a /real/ transfer function of an
opamp. I gave enough clues in my last email (which you'd have to look up
if you want them :)

Gerhard

2010\05\13@141501 by Olin Lathrop

face picon face
Gerhard Fiedler wrote:
>>>>   Out = Gain(PosIn - NegIn)
>>>
>>> Well, no, it doesn't. If it did, several common circuits wouldn't
>>> work as intended (including the Schmitt-Trigger you used in your
>>> example),
>>
>> No, the above equation shows what the opamp does in all cases, even
>> when used in a circuit with hysteresis.  I did leave out clipping at
>> the power rails, but thought it was understood a opamp can't generate
>> a voltage outside the range it is given.
>
> You say "it was understood". If everything that is understood is
> really understood, what was your point with your criticism in your
> previous message? :)

I am now totally lost as to what your point is, except possibly just to
argue.

When you said "no, it doesn't" above, were you referring to my leaving out
that the output will be clipped to the supply range, or something else?


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\05\14@113223 by Gerhard Fiedler

picon face
Olin Lathrop wrote:

> Gerhard Fiedler wrote:
> I am now totally lost as to what your point is, ...

I don't doubt. You stated something a few emails back about something I
wrote without providing good reason. You'd really have to go back and
read it if you are lost, and I'm sure you won't be lost anymore if you
did that.

> When you said "no, it doesn't" above, were you referring to my leaving
> out that the output will be clipped to the supply range, or something
> else?

I was referring to something you wrote. You really would have to either
remember or re-read what you wrote so you don't get lost in a
conversation.

Here's a hint... You wrote:

:: The opamp itself merely performs:
::  
::   Out = Gain(PosIn - NegIn)

You were trying to make a point why what I called a "mental image"
doesn't work well, arguing that it is not real enough, claiming that
your transfer function was the real deal. My point was (which you surely
had understood if you had read more than half of what I wrote) that this
transfer function is just as well a simplification of a real opamp's
transfer function as what I called a "mental image" -- with the only
difference that this image works with an ideal opamp with infinite gain,
whereas your transfer function describes an ideal opamp with finite
gain. Both are huge simplifications compared to real opamps, and in most
situations a beginner finds himself in the difference between the two
models (infinite vs finite gain) is not relevant.

Leaving out the clipping is just one of many simplifications, as I'm
sure you know.

Gerhard

More... (looser matching)
- Last day of these posts
- In 2010 , 2011 only
- Today
- New search...