Searching \ for '[EE] operation amp power pin' 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/power.htm?key=power
Search entire site for: 'operation amp power pin'.

Exact match. Not showing close matches.
PICList Thread
'[EE] operation amp power pin'
2007\09\17@180246 by Andre Abelian

flavicon
face
Hi to all,

one of our engineer recommends to connect opamp power pins thru 5 ohm resistors and 0.1 uF cap
then another engineer disagrees. I personally never used it like that.
when I asked the purpose the answer was better filteration.
my question is does this make sense to connect all opamps thru 5 ohm resistor?

any feedback will appreciate.

Andre

2007\09\17@181649 by Harold Hallikainen

face
flavicon
face

> Hi to all,
>
> one of our engineer recommends to connect opamp power pins thru 5 ohm
> resistors and 0.1 uF cap
> then another engineer disagrees. I personally never used it like that.
> when I asked the purpose the answer was better filteration.
> my question is does this make sense to connect all opamps thru 5 ohm
> resistor?
>
> any feedback will appreciate.
>
> Andre
>


I haven't done it, but I can see where it could be useful. The series R
and the shunt C form a low pass filter, so noise on the supply bus does
not get to the op amp. Further, without resistors somewhere, there's a
possibility of the supply bus ringing when currents switch. The resistors
absorb the ringing. If the op amp draws a fair amount of current for a
considerable time (long enough to pull significant charge out of the
capacitor), the supply voltage to the op amp will drop. This MIGHT be a
problem.

Harold


--
FCC Rules Updated Daily at http://www.hallikainen.com - Advertising
opportunities available!

2007\09\17@181743 by David VanHorn

picon face
On 9/17/07, Andre Abelian <spam_OUTaabelianTakeThisOuTspammason-electric.com> wrote:
> Hi to all,
>
> one of our engineer recommends to connect opamp power pins thru 5 ohm resistors and 0.1 uF cap
> then another engineer disagrees. I personally never used it like that.
> when I asked the purpose the answer was better filteration.
> my question is does this make sense to connect all opamps thru 5 ohm resistor?



one thing that this approach does, is to help attenuate noise that
might be riding on your system's power rails.  Another thing it does,
is raise the impedace of the supply to the opamp some, and that will
degrade it's ability to drive large currents.

It works, I've used it, but if you have problems that come from bad
layout, or coupling, then this won't solve them.

2007\09\17@182451 by Dario Greggio

face picon face
David VanHorn wrote:

> It works, I've used it,  [...]

same here.

--
Ciao, Dario

2007\09\17@182525 by Bob Axtell

face picon face
Andre Abelian wrote:
> Hi to all,
>
> one of our engineer recommends to connect opamp power pins thru 5 ohm resistors and 0.1 uF cap
> then another engineer disagrees. I personally never used it like that.
> when I asked the purpose the answer was better filteration.
> my question is does this make sense to connect all opamps thru 5 ohm resistor?
>
> any feedback will appreciate.
>
> Andre
>
>  
They are both right.

What he is trying to do is to isolate the OP-AMP from digital noise in
the power supply. This was needed in years past,
but modern opamps have a superb power-noise suppression mechanism that
makes those measures no longer necessary.

But it won't hurt anything. I think 0.1uF is too thin, frankly, for the
older devices. For very noisy systems, I would use 1-5
uF tantalum.

--Bob A

2007\09\17@182922 by Jinx

face picon face


> my question is does this make sense to connect all opamps thru
> 5 ohm resistor?

Andre, I've seen some examples but never thought to try it. Generally
I filter the power supply itself and that seems to be adequate for my
projects

You should read this (below is my extract concerning RC filters)

http://www.elecdesign.com/Articles/Index.cfm?ArticleID=1478&pg=3

Proper Design Techniques Solve High-Speed Op-Amp Stability Problems

Successful high-speed circuit designs need the right combination of
grounding, power-supply bypassing, and decoupling.

One solution is to add a small resistance in series with each IC's power-
supply line(s) (Fig. 4a). This provides a simple low-pass RC filter. A
typical value for this resistor is 10 Ω or less. In some applications, this
simple RC filter's gradual rolloff characteristic of 6 dB/octave may not be
enough to stabilize the circuit. An even more serious problem with this
decoupling method is that the series resistor deregulates the IC's power
supply, because the voltage at the op amp's power-supply pins varies
with supply current. This, in effect, reduces the op amp's PSRR and
introduces dc offset errors. If the resistor is increased to make the RC
filter more effective, the voltage drop could become severe enough to
shut down the op amp during large-signal periods when it must deliver
high output currents.

Replacing the resistor with a small inductor of low internal resistance
creates a more effective decoupling filter (Fig. 4b). Small ferrite beads
are commonly used for this purpose as they're very low cost and have
no significant series resistance to degrade the op amp's PSRR. Ferrite
beads can be attached to a short length of bus wire and soldered into
the board as a through-hole component

2007\09\17@190437 by Zik Saleeba

face picon face
I've seen a similar approach for a MEMS rate gyro which suggested
using a filter like this in front of an LDO regulator. Presumably this
gives the advantage of supply filtering without losing the accuracy of
regulation.

I do worry that transient loads could potentially cause a big drop
across the resistor and cause regulation to fail. I think this kind of
circuit is probably only appropriate where the load is very
consistent.

Cheers,
Zik

2007\09\17@194515 by RICH

picon face
The power pin of an op amp should be connected to the lowest possible source
impedance for the best operation with respect to noise, transients, and
response.  I do not understand how raising the power supply impedance will
improve anything.

{Original Message removed}

2007\09\17@202612 by Gerhard Fiedler

picon face
Zik Saleeba wrote:

> I do worry that transient loads could potentially cause a big drop across
> the resistor and cause regulation to fail. I think this kind of circuit
> is probably only appropriate where the load is very consistent.

Or slowly changing or low currents, like many measurement applications.

Gerhard

2007\09\17@220141 by Russell McMahon

face
flavicon
face
>>> one of our engineer recommends to connect opamp power pins thru 5
>>> ohm
>>> resistors and 0.1 uF cap

> The power pin of an op amp should be connected to the lowest
> possible source
> impedance for the best operation with respect to noise, transients,
> and
> response.  I do not understand how raising the power supply
> impedance will
> improve anything.

The object, and hopefully not the result, is not to raise the
impedance overall but to provide an isolated low impedance node for
the IC supply that is decoupled from the presumed noisy system power
supply.
To make calculation easy take a presumed signal at 10 Mhz say and do a
simplistic analysis The 0.1 uF cap has an impedance of 1/(2PifC) =
~=0.15 ohm.
This forms an ~= 30:1 voltage divider with the 5R resistor so the 10
Mhz signal is attenuated by about 30 dB. More complex and less
effective in practice, but a good start IF you need it. Adding a 0.001
in parallel and maybe a 10 uF tantalum or SOLID Al electrolytric. as
well will probably help.

FWIW this is one place where a properly voltage rated tantalum can be
used to advantage and ALMOST with safety :-). (Due to the significant
supply impedance feeding it).





           Russell


2007\09\17@221745 by Russell McMahon

face
flavicon
face
> I've seen a similar approach for a MEMS rate gyro which suggested
> using a filter like this in front of an LDO regulator. Presumably
> this
> gives the advantage of supply filtering without losing the accuracy
> of
> regulation.

If you are REALLY serious you do this in front of a local LDO :-).

> I do worry that transient loads could potentially cause a big drop
> across the resistor and cause regulation to fail. I think this kind
> of
> circuit is probably only appropriate where the load is very
> consistent.

You need to weigh up the effect of variations "inside" the power
supply filter against those outside it and, if you use it, design the
local impedance level (aka cap capacity, trace impedances, ... etc)
such that the result is much better when doing it than when not. If
the variations in local pin voltage caused by local current changes
are less than would otherwise be there at the frequencies of interest
then you (usually) win.

The local variations that may cause problems are liable to be
occurring all across the system from other ICs etc and are dealt with
by the distributed filtering. Here you are making a design decision
that you can provide superior local filtering for the noise produced
by the local node compared to the filtering you get from the system
bus compared to the noise produced on that bus. [[eg locking yourself
in a tyger proof cage is liable to protect you against tygers as long
as there is not a tyger locked inside the cage with you.]]

You can almost always design such a 'local filtering' system to be
superior to not having one, but sometimes it does not make economic or
practical sense to do so.



           Russell


What the hammer? what the chain?
In what furnace was thy brain?
What the anvil? what dread grasp
Dare its deadly terrors clasp?



http://en.wikisource.org/wiki/The_Tyger
http://en.wikipedia.org/wiki/The_Tyger

2007\09\17@225345 by RICH

picon face
I can see the reasoning.  I think if it works and someone wants to use it,
fine.  But I have been there and designed for high noise and transients.  I
have discovered a couple of things that I will share with anyone who is
interested.  Some may disagree and that is ok.  But I hope we can be civil
about our opinions. I have discovered that switchers have better immunity to
line noise and transients than do linear supplies.  I do not use linear
supplies in high noise environments like pump rooms, for example, where
pumps and compressors are banging away.  I use switchers and design my own.
I find that designing the power supply for the lowest impedance and using
low ESR capacitors has better noise immunity that a purchased linear supply.
If possible, it is better to design your own supply, but that is not always
possible.  If you have a spectrum analyzer and can identify the exact
frequency or frequencies of the noise it can provide useful information
because you can see what you are up against and design accordingly.


{Original Message removed}

2007\09\18@001623 by Russell McMahon

face
flavicon
face
> I can see the reasoning.  I think if it works and someone wants to
> use it,
> fine.

It is, or was, quite common practice. Appropriate in some cases.
Totally inappropriate in others. It's a matter of engineering - use
the best solution available, all things considered. Vector sum all of
(noise levels, cost, board area, asthetics, personal preference, ... )
and optimise to suit your personal preferences.

>But I hope we can be civil about our opinions.

No need for civility - this is engineering :-).
ie as above, different people choose different solutions for differing
reasons after having weighed all factors using scaling factors that
seem right to them. In many cases there is no one best solution. Some
solutions may be so manifestly deficient in some areas that it would
require an exceptional personal weighting of other factors to make
them the one you'd choose.

eg One silly example:    A shunt zener regulator may be the cheapest
and most compact solution to a problem. It would usually have little
else going for it. There are zillions of zener shunt regulator power
suppllies out there and some of them may even be the best solution.

Another even sillier example. A 7805*.
:-)

> I have discovered that switchers have better immunity to
> line noise and transients than do linear supplies.

That may well be true for many versions of each subset. One valid
approach is what yuou have become comfortable with. Another is to
adapt the linear regulator's environmenmt in some way. Sometimes this
will be "better" than using a swuitcher. Othertimes it wont.




           Russell

* Fighting words !
[[A 7805 / LM340 is cheap. It has many deficiencies wrt other
solutions. But it's cheap.]]


2007\09\18@003226 by Vasile Surducan

face picon face
If will be just one OA and one resistor, it does not hurt nor throwing
you in the heaven.
If you'll have an OA and a comparator and you'll do it the same with
both (from a single resistor) will throw you in hell. Do you know why
?
This "filtering" methode is often used by hw engineers designing logic
stuff running at high speed (AKA CPLD, FPGA, DSP). The better choice
instead a resistor is a 600ohm/100MHz and 0 ohm DC ferrite suppressor.
As long you don't know from where is comming the aggressor signal is
usless because it can kill your analogic signal through ground as
well.

Vasile

On 9/17/07, Andre Abelian <.....aabelianKILLspamspam@spam@mason-electric.com> wrote:
{Quote hidden}

> -

2007\09\18@020351 by wouter van ooijen

face picon face
> one of our engineer recommends to connect opamp power pins
> thru 5 ohm resistors and 0.1 uF cap then another engineer
> disagrees. I personally never used it like that. when I asked
> the purpose the answer was better filteration.
> my question is does this make sense to connect all opamps
> thru 5 ohm resistor?

my first action would be: quantify! how much noise can you tolerate on
the opamps output, how much noise is on the power rail, and how much
does the opamp couple  power to output (forgot the correct term), and
how much does a 5r / 100n filter reduce this?

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu



2007\09\18@025552 by Xiaofan Chen

face picon face
On 9/18/07, wouter van ooijen <wouterspamKILLspamvoti.nl> wrote:
> > one of our engineer recommends to connect opamp power pins
> > thru 5 ohm resistors and 0.1 uF cap then another engineer
> > disagrees. I personally never used it like that. when I asked
> > the purpose the answer was better filteration.
> > my question is does this make sense to connect all opamps
> > thru 5 ohm resistor?
>
> my first action would be: quantify! how much noise can you tolerate on
> the opamps output, how much noise is on the power rail, and how much
> does the opamp couple  power to output (forgot the correct term), and
> how much does a 5r / 100n filter reduce this?
>
Wouter:
I think this is *surprisingly* not so easy to do. Give an example and show
us how you do that using hand-calculation or Spice simulation or whatever
tools you have.

Regards,
Xiaofan

2007\09\18@031428 by Russell McMahon

face
flavicon
face
> my first action would be: quantify! how much noise can you tolerate
> on
> the opamps output, how much noise is on the power rail, and how much
> does the opamp couple  power to output (forgot the correct term),

Power supply rejection ratio.

Useful tutorial thereon. Note that they mention a choke, as Vasile
mentioned, but do not mention using a "less real" series resistor.
They do mention a decoupling cap per power supply pin.

       http://www.national.com/nationaledge/apr01/ab102.html


> and how much does a 5r / 100n filter reduce this?

My roughas boots example suggests it's not more than 30 dB at 10 MHz.
So probably say 10 to 20 dB in the real world, for some values of
real. There's some environments where a 10 dB reduction in power
supply induced noise would be most welcome. Whether any diadvantages
outweight the gain is application specific.


       Russell

2007\09\18@031916 by Russell McMahon

face
flavicon
face
>> my first action would be: quantify! how much noise can you tolerate
>> on
>> the opamps output, how much noise is on the power rail, and how
>> much
>> does the opamp couple  power to output (forgot the correct term),
>> and
>> how much does a 5r / 100n filter reduce this?

> I think this is *surprisingly* not so easy to do. Give an example
> and show
> us how you do that using hand-calculation or Spice simulation or
> whatever
> tools you have.

See below, from an earlier post of mine:
This is a single frequency calculation aimed at getting an order of
magnitude feel for the noise reduction. It assumes that the capacitor
impedance is substantially lower than the IC impedance. Also that the
0.1 uF makes a better filter for the ICs own noise than the
distributed decoupling on the system power rail.

Spice etc can do a better job of this but this gives a feel for what
may be achievable.


       Russell
______________

The object, and hopefully not the result, is not to raise the
impedance overall but to provide an isolated low impedance node for
the IC supply that is decoupled from the presumed noisy system power
supply.

To make calculation easy take a presumed signal at 10 Mhz say and do a
simplistic analysis The 0.1 uF cap has an impedance of 1/(2PifC) =
~=0.15 ohm.

This forms an ~= 30:1 voltage divider with the 5R resistor so the 10
Mhz signal is attenuated by about 30 dB. More complex and less
effective in practice, but a good start IF you need it. Adding a 0.001
in parallel and maybe a 10 uF tantalum or SOLID Al electrolytic. as
well will probably help.

2007\09\18@040009 by wouter van ooijen

face picon face
> I think this is *surprisingly* not so easy to do. Give an
> example and show us how you do that using hand-calculation or
> Spice simulation or whatever tools you have.

Which of the steps is not easy to do, and which amount of error would
you expect?

I don't pretend to have the ultimate answer in such electrical questions
(I am a software guy), but IME even an attempt to qualify things can
make a lot of things clear.

That is, if power-line induced noise on the output is the problem. Maybe
it is stability, in which case I rest my case.

OTOH I have seen electronic designers put in things which they 'always
did' and which made perfect sense 2 or 3 generations of chips ago. No
big problem, unless that practice causes other problems (cost, size,
reliability, etc).

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu



2007\09\18@040012 by wouter van ooijen

face picon face
> So probably say 10 to 20 dB in the real world, for some values of
> real. There's some environments where a 10 dB reduction in power
> supply induced noise would be most welcome. Whether any diadvantages
> outweight the gain is application specific.

That's a avlue for one of the figures. Two more to go for a napkin
approximation :)

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu



2007\09\18@041325 by Xiaofan Chen

face picon face
On 9/18/07, wouter van ooijen <.....wouterKILLspamspam.....voti.nl> wrote:
> > I think this is *surprisingly* not so easy to do. Give an
> > example and show us how you do that using hand-calculation or
> > Spice simulation or whatever tools you have.
>
> Which of the steps is not easy to do, and which amount of error would
> you expect?
>
> I don't pretend to have the ultimate answer in such electrical questions
> (I am a software guy), but IME even an attempt to qualify things can
> make a lot of things clear.

The problem is that you write "quantify!" and that is certainly different
than "qualify".
On 9/18/07, wouter van ooijen <EraseMEwouterspam_OUTspamTakeThisOuTvoti.nl> wrote:
> my first action would be: quantify!


Xiaofan

2007\09\18@042202 by Alan B. Pearce

face picon face
> one of our engineer recommends to connect opamp power pins thru
> 5 ohm resistors and 0.1 uF cap then another engineer disagrees.

This rather sounds like the engineer advocating the filter had a problem in
a circuit, and this fixed it. It may well be that the circuit concerned had
other problems causing some form of high frequency motor boating or
instability, and filtering the power rails was enough to break the loop.

2007\09\18@043746 by Russell McMahon

face
flavicon
face
>> So probably say 10 to 20 dB in the real world, for some values of
>> real. There's some environments where a 10 dB reduction in power
>> supply induced noise would be most welcome. Whether any
>> disadvantages
>> outweigh the gain is application specific.

> That's a value for one of the figures. Two more to go for a napkin
> approximation :)

I'm puzzled.
What are the other two figures required?

I'd rate the 30 dB as napkin and the 10-20 dB as informed guess, which
rates somewhere above napkin or BOTE and below guesstimate.



           Russell
>
> Wouter van Ooijen
>
> -- -------------------------------------------
> Van Ooijen Technische Informatica: http://www.voti.nl
> consultancy, development, PICmicro products
> docent Hogeschool van Utrecht: http://www.voti.nl/hvu
>
>
>
> --

2007\09\18@050106 by Vasile Surducan

face picon face
On 9/17/07, Xiaofan Chen <xiaofancspamspam_OUTgmail.com> wrote:
> On 9/18/07, wouter van ooijen <@spam@wouterKILLspamspamvoti.nl> wrote:
> > > one of our engineer recommends to connect opamp power pins
> > > thru 5 ohm resistors and 0.1 uF cap then another engineer
> > > disagrees. I personally never used it like that. when I asked
> > > the purpose the answer was better filteration.
> > > my question is does this make sense to connect all opamps
> > > thru 5 ohm resistor?
> >
> > my first action would be: quantify! how much noise can you tolerate on
> > the opamps output, how much noise is on the power rail, and how much
> > does the opamp couple  power to output (forgot the correct term), and
> > how much does a 5r / 100n filter reduce this?
> >
> Wouter:
> I think this is *surprisingly* not so easy to do. Give an example and show
> us how you do that using hand-calculation or Spice simulation or whatever
> tools you have.

Xiofan, do you know what PSRR is, right ?
Surprisingly is quite easy if you know the supply rail noise and PSRR.

2007\09\18@055947 by Xiaofan Chen

face picon face
On 9/18/07, Vasile Surducan <KILLspampiclist9KILLspamspamgmail.com> wrote:
> > > my first action would be: quantify! how much noise can you tolerate on
> > > the opamps output, how much noise is on the power rail, and how much
> > > does the opamp couple  power to output (forgot the correct term), and
> > > how much does a 5r / 100n filter reduce this?
> > >
> > Wouter:
> > I think this is *surprisingly* not so easy to do. Give an example and show
> > us how you do that using hand-calculation or Spice simulation or whatever
> > tools you have.
>
> Xiofan, do you know what PSRR is, right ?
> Surprisingly is quite easy if you know the supply rail noise and PSRR.

Maybe I should not use the word "surprisingly". My point is that
noise is not easy to quantify. I am certainly not good at RF and
constantly bugged by EMC/EMI issues.

Xiaofan

2007\09\18@060622 by Russell McMahon

face
flavicon
face
> That's a avlue for one of the figures. Two more to go for a napkin
> approximation :)


Ah

> > my first action would be: quantify! how much noise can you
> > tolerate on
> > the opamps output, how much noise is on the power rail, and how
> > much
> > does the opamp couple  power to output (forgot the correct term),
> > and
> > how much does a 5r / 100n filter reduce this?

So

> > how much noise can you tolerate on the opamps output

V uV. Known

> how much noise is on the power rail

W uV. Known / measuraeable.

> and how much does the opamp couple  power to output

Spec sheet (or measure). Known.

> > how much does a 5r / 100n filter reduce this?

10-20 dB.

Envelope ready for inspection, Sir!


       Russell

2007\09\18@064140 by wouter van ooijen

face picon face
> The problem is that you write "quantify!" and that is

I definitely meant '(attempt to) quantify'. I don't mean 1 db accuracy,
just a rought estimate, to check whether there is a problem at all, and
whether power filtering might contribute to a solution.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu



2007\09\18@070506 by wouter van ooijen

face picon face
> Envelope ready for inspection, Sir!

Yeah, when the original questioner fills in the 'known values' :)

In my employed days I have spent a surprising number of hours hunting
through the organisation for figures like this, only to find out that
no-one knew for sure or dare to say. But this was mostly for more
informatics-oriented figures, but also things like strutural stiffness,
backlash, and radiation resistance.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu



2007\09\18@073245 by wouter van ooijen

face picon face
> Maybe I should not use the word "surprisingly". My point is
> that noise is not easy to quantify. I am certainly not good
> at RF and constantly bugged by EMC/EMI issues.

In retrospect I jumped two (or maybe more) steps in my answer which I
would have taken first.

If you were not talking to a reasonably experienced engineer get one.

1. If these two components do not pose a real problem (cost, size,
reliability, electrical, ...) and some reasonably experienced engineer
states that they should be there, by all means keep them there and think
no further.

If the problem is just cost make a very rough estimate to check whether
the saved cost can outweight the time you and others are gona spent
trying to remove the components.

2. if they are a real problem get an appropriate engineer to explain why
they should be there. if it is noise from the power coupled to the
output get yourself a pencil and a napkin and start asking the right
questions.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu



2007\09\18@144126 by Vasile Surducan

face picon face
On 9/18/07, Xiaofan Chen <RemoveMExiaofancTakeThisOuTspamgmail.com> wrote:
> On 9/18/07, Vasile Surducan <spamBeGonepiclist9spamBeGonespamgmail.com> wrote:
> > > > my first action would be: quantify! how much noise can you tolerate on
> > > > the opamps output, how much noise is on the power rail, and how much
> > > > does the opamp couple  power to output (forgot the correct term), and
> > > > how much does a 5r / 100n filter reduce this?
> > > >
> > > Wouter:
> > > I think this is *surprisingly* not so easy to do. Give an example and show
> > > us how you do that using hand-calculation or Spice simulation or whatever
> > > tools you have.
> >
> > Xiofan, do you know what PSRR is, right ?
> > Surprisingly is quite easy if you know the supply rail noise and PSRR.
>
> Maybe I should not use the word "surprisingly". My point is that
> noise is not easy to quantify. I am certainly not good at RF and
> constantly bugged by EMC/EMI issues.

Besides philosophy, hopefully well represented on this list, the noise
filtering problem for an OA is a simple one:
1. check from datasheet the frequency bandwith PSRR is able to reject
2. measure the supply noise using a high bandwith scope (500Mhz-1Ghz)
or a spectrum analyzer
3. if your supply has a noise frequency component much higher that
could be rejected by PSRR, it must be rejected by a suplementary
filter
4. define clear which is the aggresor signal (the supply or the load),
in this case is the supply if OA is running in linear application and
both the supply and the load if OA is running as a comparator in open
loop or with positive reaction
5. build the reject filter (LC, RC or pi network) in accordance with 3.

As a good advice: if your supply is a digital one (noise between 50mV
to 150mV) than do not use it for OA supply unless is your last option.

Vasile

2007\09\18@193049 by Xiaofan Chen

face picon face
On 9/19/07, Vasile Surducan <TakeThisOuTpiclist9EraseMEspamspam_OUTgmail.com> wrote:
{Quote hidden}

This is what I thought after my post (not as clear as you write). Thanks.
You certainly are an expert in RF and high speed digital. I have never
used an FPGA or doing something high speed. I only worked with
low frequency stuff and some DC/DC converters.

A high bandwidth high bandwith scope is a very good tools (many
can do FFT as well) but it is out of the reach for many hobbyists.

Regards,
Xiaofan

2007\09\18@194333 by Mark Rages

face picon face
On 9/18/07, Vasile Surducan <piclist9EraseMEspam.....gmail.com> wrote:
> As a good advice: if your supply is a digital one (noise between 50mV
> to 150mV) than do not use it for OA supply unless is your last option.
>
> Vasile
> --

Something I haven't seen mentioned on this thread is that you can also
improve supply noise by filtering at the noisy component.   If you
have problems with a component causing supply noise with another
component, try an RC filter on both components.

Regards,
Mark
markrages@gmail
--
Mark Rages, Engineer
Midwest Telecine LLC
EraseMEmarkragesspammidwesttelecine.com

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...