# How to use Decoupling capacitors for Op amp in dual rail

#### picstudent

Joined Feb 3, 2009
91
Hello

I have op amp OP07 and TL084 in a board. Some 2to 30 in a board with a MCU onboard.

How to provide the decoupling capacitors for these op amps. +12 and -12 are supply rails?

We need to decouple both rails or one is needed? How the the optimum value is decided?

Thanks

#### mik3

Joined Feb 4, 2008
4,843
Put a 100nF capacitor between the +ve rail and ground, another one between -ve rail and ground and one between +ve and -ve rails.

#### picstudent

Joined Feb 3, 2009
91
Put a 100nF capacitor between the +ve rail and ground, another one between -ve rail and ground and one between +ve and -ve rails.
So three non polar capacitors .1uF for each op amp. I have some 30 op amp in a PCB. So huge part count.

What is really the function of de copupling capacitors in Op amp circuits? What are the consequences in general if not provided?

Thanks

#### gotumal

Joined Mar 24, 2008
99
Any IC consume power depending on the output load condition (drive power). So when load curent increases suddenly, a voltage dip may occure at the power pins of IC. During that time, decoupling capacitor provides the enough charge not to have voltage dip.

Depending on freqency of operation (analog, digital (in KHz, MHz)) value can be decided like 0.001µF, 0.01µF, 0.1µF etc.

#### SgtWookie

Joined Jul 17, 2007
22,221
Well, actually - just two 100nF nonpolarized caps per opamp.
One from Vcc to ground, and one from Vee to ground, as close to the supply pins as you can.

xox

#### mik3

Joined Feb 4, 2008
4,843
Well, with op amps, if you have a good low noise power supply you don't really need decoupling capacitors because they don't switch rapidly and cause noise on the pcb tracks.

#### davidgrahamsites

Joined Jul 6, 2011
1
Why the extra between V+ and V-? Is this in case of a fast transition between upper and lower voltage rails with a signal carrying a higher load?

#### crutschow

Joined Mar 14, 2008
28,563
Well, with op amps, if you have a good low noise power supply you don't really need decoupling capacitors because they don't switch rapidly and cause noise on the pcb tracks.
Not entirely true. The decoupling is not to prevent noise on the power rails, it's to stabilize the op amp. Some op amps (especially higher frequency ones) will become unstable or show deviations in their frequency response if their power rails are not properly decoupled.

#### crutschow

Joined Mar 14, 2008
28,563
Why the extra between V+ and V-? Is this in case of a fast transition between upper and lower voltage rails with a signal carrying a higher load?
If the power rails are properly decoupled to ground, then you shouldn't need the third cap. The power rail AC currents are from each rail to ground through the output load (plus or minus rail depending upon whether the output polarity is positive or negative), so the primary currents to be decoupled are from each of the rails to ground, not between rails.

#### ErnieM

Joined Apr 24, 2011
8,220
Caps are damn cheap. Add then every place (even that third cap) if you are laying out a PC board. If you find you don't need them all you just don't populate some spots.

If the power rails are properly decoupled to ground, then you shouldn't need the third cap.
I've never had cause to need that third cap either.

Well, with op amps, if you have a good low noise power supply you don't really need decoupling capacitors because they don't switch rapidly and cause noise on the pcb tracks.
Sure, leave em out, if you don't mind extraneous oscillations and other weird artifacts.

#### MrChips

Joined Oct 2, 2009
25,057
This question keeps popping up. On top of that you have a board mixing analog with digital. This can lead to serious problems depending on the specifics.

A new student in electronics may not appreciate the importance of proper power and ground layout and decoupling. You would be wise to listen to the experts in the field and follow their advice.

First good advice: install a 100nF capacitor between power and ground for every chip. Keep leads as short as possible.

For critical analog circuitry such as op amps, ADC, etc. install a 10uF electrolytic in parallel with 100nF between power and ground at each device.

If you have mixed signal board, i.e. both analog and digital, do not power the analog circuitry with the same power rails as the digital. Keep analog power rails separate from digital.

Know the difference between a star connection and a party-line or daisy-chain connection.
Avoid running daisy-chain analog power rails

#### wakibaki

Joined Jun 12, 2012
7
This is a vexed question.

Not very long ago I designed a board following the 'one 100n per rail per chip' regime. Just today I came across this: http://www.eetimes.com/design/audio-design/4218273/Op-amps-in-small-signal-audio-design---Part-3--Selecting-the-right-op-amp?pageNumber=1 - from Doug Self.

The meat of the article in this context is this quote:

'The 5532 and 5534 type op-amps require adequate supply decoupling if they are to remain stable, otherwise they appear to be subject to some sort of internal oscillation that degrades linearity without being visible on a normal oscilloscope. The essential requirement is that the positive and negative rails should be decoupled with a 100 nF capacitor between them, at a distance of not more than a few millimeters from the op-amp; normally one such capacitor is fitted per package as close to it as possible.

It is not necessary, and often not desirable, to have two capacitors going to ground; every capacitor between a supply rail and ground carries the risk of injecting rail noise into the ground.'

Live and learn, huh?

w

#### crutschow

Joined Mar 14, 2008
28,563
..................................
It is not necessary, and often not desirable, to have two capacitors going to ground; every capacitor between a supply rail and ground carries the risk of injecting rail noise into the ground.'
I disagree with that as a general statement. If the op amp output is driving any load to ground then you need decoupling capacitors from ground to each supply rail to provide a direct return path for the high frequency load currents from ground to the rails (which are the sources for the load current). Otherwise the output may be distorted or oscillate.

#### PaulEE

Joined Dec 23, 2011
423
picstudent, how large is this board?

When decoupling op-amps, it is good practice (and design) to use both a small capacitance value (with as short a board trace as you can) as well as a larger bulk capacitance (traces not as critical). These capacitors serve to filter/smooth out/reduce high frequency (small caps) and low frequency (big caps) noise. These caps, along with an op-amps power supply rejection ratio, govern how much power supply noise will eventually get to the output of your circuit.

If you know that the signals in your circuit will be predominantly DC and that small amounts of noise or blips are acceptable at your output, then no, don't sprinkle the board with 60 capacitors. A few evenly spaced bulk capacitances and a few small caps would be fine.

If you know for a fact that your circuit will be dealing with signals such as low frequency sine waves, I would still do the above.

If, however, you are dealing with higher frequency signals, whether directly (sine waves) or indirectly (square waves whose harmonics are high frequencies, and whose properties are abrupt on/off), and noise/blips are unacceptable at the output, then you must properly decouple everything.

As far as the stability of the op-amp...there was a time when an op-amp was not internally-compensated, as in, its bandwidth was limited only by its stray capacitances and not through any other means, these op-amps and circuits tended to oscillate. The solution to this was to add a compensation capacitor, or, in plain English, a capacitor to either compensation pins or to the feedback circuit that attenuated/filtered/limited the device with respect to higher frequencies.

Nowadays, op-amps like the 741 and OP07 DO have internal capacitances that generally prevent random oscillations from occurring.

I do not know what your circuit is physically doing, but in looking at the gain bandwidth product and slew rate of the TL084, which I believe is just a quad TL081, I hope your circuit is designed in such a way as to limit the bandwidth such that just your signals are happily dealt with and nothing above this range...otherwise, stringing a bunch of these together will be a disaster.

Remember the criterion for oscillation...input/output 0* or 360* with one another (in phase) and gain >=1 (>1 or 2 in real life) amplifying element.

The op-amps you use will spell out the gain and phase characteristics and you can judge whether your circuit may or may not oscillate.

Meanwhile, I haven't even mentioned input and output capacitances and inductances. I'm still trying to figure out why you've got 30 op-amp chips on a board in the first place...

Hopefully this helped a bit.

By the way:
http://www.analog.com/static/imported-files/tutorials/MT-043.pdf

#### ramancini8

Joined Jul 18, 2012
473
A third reason for adding decoupling caps is that the power supply rejection ratio is low for most op amps, it decreases with increasing noise frequency, and digitally generated noise is high frequency. Keeping noise out of the op amps generally requires a small resistor in series with the decoupling cap.

EDN Access--09.12.97 Decoupling capacitors: use them or fail is an EDN column that discusses decoupling in detail.