Before working as an applications engineer, I worked as an IC test development engineer here at TI. One of my projects was to characterize an I2C temperature sensor. After writing some software, I threw together a hand-wired prototype board. I was in a hurry, so I left off that pesky decoupling capacitor. Who needs it, right?
I collected data for about a week, and none of my results matched expectations. I made numerous changes in an attempt to improve performance, but nothing worked. Finally, I decided to add the decoupling capacitor. As you might expect, this solved the issue.
This got me thinking…do we always need decoupling capacitors? What do they really do?
One way to answer the question is to show what happens when you don’t use proper decoupling.
Figure 1 shows a buffer circuit driving an R-C load with and without decoupling capacitors (C1 and C2). Notice that the output signal has a high frequency (3.8MHz) oscillation for the circuit without decoupling. Poor stability, poor transient response, start-up problems, and other anomalies are common challenges with amplifiers that do not have decoupling capacitors.
Figure 1: Buffer with and without decoupling (measured results)
Figure 2 illustrates why decoupling is important. Note that the inductance of the power supply trace will limit the transient current.
The decoupling capacitor is very close to the device, so it has a very low inductance path for current flow. During transients, the capacitor can supply very large amounts of current to the device for a very short duration.
The device without decoupling does not have a mechanism to provide the transient currents, so the amplifier’s internal nodes will droop – often referred to as a glitch. The internal power supply glitches on the device without decoupling cause inconsistent operation, because the internal nodes are not properly biased.
Figure 2: Current flow with and without decoupling
In addition to using a decoupling capacitor, you should also use a short low impedance connection between the decoupling capacitor, the power supply, and the ground connection.
Figure 3 compares a good decoupling layout to a bad one. You should always try to keep the decoupling connections short and avoid vias in the decoupling path, because vias add inductance. Most data sheets recommend a decoupling capacitor value. If no recommendation is given, use 0.1uF.
Figure 3: Good vs. bad PCB layout
Using a properly connected decoupling capacitor can save you a lot of trouble. Even if your circuit works on the bench without decoupling, it could have issues when you go into production from process variation and other real world influences.
Learn from my mistake; don’t fall into the no-decoupling trap!
A special thanks to my colleagues Ichiro Itoi and Tim Green for your insights into decoupling and real-world measured results.
In this occasion, I will just say that this subject is much more complex than you presented it here!
Another perspective is to treat the opamp as being supplied mostly from the decoupling capacitor; that the PCB power traces (due to their length) are only for d.c. supply and are sized just for low resistive drop. Depends on the load current and size and number of capacitors you can place next to the opamp though, but for driving ADC inputs the load is often very light at d.c.. Taking it further, you can even insert a resistor or inductor in the power supply trace as a cheap form of isolation from the system supply if you can't afford a local regulator.
This experiment data is very useful for junior engineer. Thank you very much.
Thank you for the post. I was wondering if you have other useful posts regarding low noise design?
I wrote a text book on noise "Operational Amplifier Noise, by Art Kay". Also, below is a link to a short video on the subject. I will probably do some further blogs on this subject.
Thank you for very useful data., How to choose voltage rating of the capacitor? Because I heard different answers from different people.
I am usually very conservative when it comes to choosing the voltage rating on capacitors. I choose the rating to be at lease 2x the maximum expected voltage on the capacitor. So for example, if I am using a 15V supply I will make sure the capacitor is rated for 30V or more. Generally, it is easy to find a 0.1uF X7R capacitor in a small package (e.g. 0603, or 0508) that will meet my requirements. The voltage rating has three main concerns. 1) exceeding it can damage the capacitor, 2) approaching the rating can de-rate the capacitance value, 3) the voltage rating may be de-rated over temperature. In cases where you cannot use the conservative approach you need to get detailed specifications on the capacitor and understand if the rating will work for your application over termperature.
All content and materials on this site are provided "as is". TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with regard to these materials, including but not limited to all implied warranties and conditions of merchantability, fitness for a particular purpose, title and non-infringement of any third party intellectual property right. No license, either express or implied, by estoppel or otherwise, is granted by TI. Use of the information on this site may require a license from a third party, or a license from TI.
TI is a global semiconductor design and manufacturing company. Innovate with 100,000+ analog ICs andembedded processors, along with software, tools and the industry’s largest sales/support staff.