# How to choose the right opamp

I

#### [email protected]

Jan 1, 1970
0
Hi,
if i consider a signal with a defined bandwith, how can i choose the
right opamp to amplify it?
For example, if consider a signal with a maximum frequency of 1kHz i
think i should use an opamp with first pole equal or greater to 1kHz:
is it right?If so, what is the phase effect on the real signal?
All frequencies below 1kHz have different phase induced by opamp non
idelities, the output signal is equal to Re(|A|e^jfi), so: what shoud
have is an output time delay or the output signal is distorted
(because different frequencies have different phase shift)? How this
problem is important in real pratical cases?

E

#### Eeyore

Jan 1, 1970
0
Hi,
if i consider a signal with a defined bandwith, how can i choose the
right opamp to amplify it?

First define the voltage gain you want and the HF droop allowable at high
frequency X. Plus for audio especially, distortion is a consideration.
That (plus a data sheet) at least will tell you the Gain Bandwidth
Product you need.

Consider input noise voltage and current. Input bias and differential
input bias current are very important for DC coupled amps. Output drive
capability. Ability to swing the inputs and/or outputs to ground and/or
supply rail or possibly beyond. Power consumption. Maximum working
voltage. Package dissipation in the event of high output currents. And so
on.

at TI there's also SBOA92A.pdf. Only 94 pages that one instead of 464. A
little more condensed !

Graham

E

#### Eeyore

Jan 1, 1970
0
For example, if consider a signal with a maximum frequency of 1kHz i
think i should use an opamp with first pole equal or greater to 1kHz:
is it right?

NO. See a Bode Plot for a typical op-amp.

Example: the TL071 family has been VERY widely used for respectable
quality audio (up to 20kHz) but the pole is at around 30Hz.. See Figure
12
http://focus.ti.com/docs/prod/folders/print/tl071.html click on
datasheet.

If so, what is the phase effect on the real signal?

How much feedback theory have you learnt ?

All frequencies below 1kHz have different phase induced by opamp non
idelities, the output signal is equal to Re(|A|e^jfi), so: what shoud
have is an output time delay or the output signal is distorted
(because different frequencies have different phase shift)? How this
problem is important in real pratical cases?

Utter rubbish.

Graham

B

#### Bob Eld

Jan 1, 1970
0
Hi,
if i consider a signal with a defined bandwith, how can i choose the
right opamp to amplify it?
For example, if consider a signal with a maximum frequency of 1kHz i
think i should use an opamp with first pole equal or greater to 1kHz:
is it right?If so, what is the phase effect on the real signal?
All frequencies below 1kHz have different phase induced by opamp non
idelities, the output signal is equal to Re(|A|e^jfi), so: what shoud
have is an output time delay or the output signal is distorted
(because different frequencies have different phase shift)? How this
problem is important in real pratical cases?

Putting feedback around and amp moves the open-loop pole out to a higher
frequency. So in the closed loop case, the position of the pole or break
point depends on the gain.

If the natural pole is 10Hz and the open loop gain 100,000 times, then
closing the loop to a gain of 100 times would move the 10Hz pole to
100,000/100 X 10Hz = 10 kHz.

The phase shift of the new pole position is 45 Deg at that frequency, in
this case, 10 kHz. In other words both the frequency response and phase
response gets "flattened" out and extended by applying feedback.

Even when the closed loop bandwidth appears sufficient for a particular
application, the amplifier's slew rate may limit the amplifier's ability to
drive high frequencies at a given voltage.
Slew rate is independent of bandwidth but must be considered for a
particular application.

Usually an amplifiers bandwidth is expressed as Gain Bandwidth Product. This
number is any point along the open loop frequency response curve. In the
above example, the the gain was 100,000 times at 10 Hz. so the GBP =
1,000,000. It also implies, assuming a single pole response that unitity
open loop gain occurs at 1 MHz.

G

#### Glenn Gundlach

Jan 1, 1970
0
at TI there's also SBOA92A.pdf. Only 94 pages that one instead of 464. A
little more condensed !

Graham

Took a little hunting for it.

SBOA092A

is the correct number.

E

#### Eeyore

Jan 1, 1970
0
Glenn said:
Took a little hunting for it.

SBOA092A

is the correct number.

Hah. That's not what's on the cover ! I'd already renamed the file you
see.

Graham

E

#### Eeyore

Jan 1, 1970
0
RST Engineering (jw) said:
Let's see now, first week of December ... TERM PAPER TIME FOR FALL SEMESTER.

I think someone's going to fail this one.

Graham

I

#### [email protected]

Jan 1, 1970
0
Thanks for the infos...
at TI there's also SBOA92A.pdf. Only 94 pages that one instead of 464. A
ittle more condensed !
done!I'll read it very soon, it's interesting...
Putting feedback around and amp moves the open-loop pole out to a higher
frequency. So in the closed loop case, the position of the pole or break
point depends on the gain.
Thanks, now it's clear
My doubt is about phase effect on a real signal...I know that phase is
important for a feedback system (stability of a feedback system is
related to phase margin), but i'm not able to merge signal theory with
electronics...I've studied in signal theory that if i consider the
whole system (opamp+feedback) as black box and know only output and
input signal,output signal is equal to Re(|A|e^jfi) with A=(gain of
black box)*(amplitude of input signal), fi=(phase shift of black box)+
(phase of input signal)...If consider output signal as opamp output,
if signal is a cos with frequency equal to cut-off frequency of the
whole system, output cos has amplitude of -3dB respect to the same low
frequency cos, and phase shift of 45°....so real segnil has vaule
modified also by phase shift: is it right?
Let's see now, first week of December .... TERM PAPER TIME FOR FALL SEMESTER
I've decided to avoid this semestrer, i want to understand it very
well

E

#### Eeyore

Jan 1, 1970
0
Thanks for the infos...
done!I'll read it very soon, it's interesting...

Thanks, now it's clear
My doubt is about phase effect on a real signal...I know that phase is
important for a feedback system (stability of a feedback system is
related to phase margin), but i'm not able to merge signal theory with
electronics...I've studied in signal theory that if i consider the
whole system (opamp+feedback) as black box and know only output and
input signal,output signal is equal to Re(|A|e^jfi) with A=(gain of
black box)*(amplitude of input signal), fi=(phase shift of black box)+
(phase of input signal)...If consider output signal as opamp output,
if signal is a cos with frequency equal to cut-off frequency of the
whole system, output cos has amplitude of -3dB respect to the same low
frequency cos, and phase shift of 45°....so real segnil has vaule
modified also by phase shift: is it right?

I've decided to avoid this semestrer, i want to understand it very
well

Sensible move.

Incidentally negative feedback works for phase shift much like it works for gain.
In approximate terms suppose your op-amp had an open-loop phase shift of 10
degrees at frequency X and you apply a feedback factor of 10, then the output
phase shift will be about 1 degree. This quick calculation relies on the fact that
for small values of degrees, the relationship between sin theta and its theta is
quite linear.

Graham

Replies
3
Views
734
Replies
9
Views
1K
D
Replies
2
Views
3K
D
A
Replies
2
Views
1K
Andrew Lawrance
A