Similar topics
Recommended
John Caldwell

While working on a low-noise amplifier circuit for an upcoming TI Designs reference design, I was caught off-guard by some interesting behavior. Casually moving the printed circuit board (PCB) on my workbench caused the output voltage to jump! Fascinated, I decided to perform a highly scientific test: I tapped on the PCB repeatedly while observing the output voltage on an oscilloscope.

Figure 1: Circuit output produced by tapping on the PCB

 

The seven spikes in the output voltage shown in Figure 1 are the result of my tapping on the PCB. Many physical interactions with a PCB can cause the output of a circuit to change. For example, placing stress on the package of an op amp can change its offset voltage. However, this circuit was EXTREMELY sensitive to vibration, and op amps typically do not show this level of sensitivity. With this in mind, I turned my attention to the ceramic capacitors on the PCB.

Multi-layer ceramic capacitors are incredibly useful. They offer a unique combination of low equivalent series resistance (ESR) and equivalent series inductance (ESL) and high volumetric efficiency. Their construction consists of multiple layers of metal electrodes inside a ceramic dielectric material as shown in Figure 2.

                                                                                       

Figure 2: The typical structure of a multilayer ceramic capacitor

Barium Titanate (BaTiO3) is commonly used in the dielectric of ceramic capacitors because it can have a relative permittivity greater than 3000 [1]. Typically, as you shrink the physical size of a ceramic capacitor, increasing the capacitance requires a larger quantity of BaTiO3 to be used in the dielectric. Aside from high permittivity, BaTiO3 has another interesting property: it is highly piezoelectric. This makes it an excellent choice for piezoelectric microphones and guitar pickups!

The piezoelectric effect is the creation of a voltage due to an applied mechanical stress [2]. Figure 3 shows a ceramic capacitor soldered to a PCB. When downward pressure is applied (red arrow) the PCB deforms, causing the dielectric to be stretched or compressed by the end caps (blue arrows). When I tapped on my PCB, I applied a mechanical stress to the ceramic capacitors, causing a piezoelectric response in the dielectric and producing an output voltage.

Figure 3: Mechanical stress on the PCB is coupled to the dielectric through the capacitor end caps

 

Piezoelectricity can be a major problem for electronics installed in high-vibration environments. In such applications the need for high capacitance, low ESR and ESL, and small size may lead an engineer to select a high-k ceramic capacitor (X7R, Y5V, Z5U, etc.), which contains a large percentage of BaTiO3 [3]. A common example is the capacitor placed on the reference input to an ADC. The circuit may work fine in the lab, where it isn’t being shaken vigorously. Once it is installed in an environment with vibration, significant errors in the ADC readings might appear. Power supply designers are also aware of the converse piezoelectric effect, where the ripple voltage across the capacitor causes it to “sing” or vibrate.

For my low-noise amplifier circuit, I chose to investigate a few different solutions to this problem:

  1. Soft-termination ceramics: These are ceramic capacitors with a flexible material inside the end-caps for stress relief. They were introduced for automotive applications, where PCB flexing can cause capacitors to fail.
  2. Tantalum capacitors: Tantalum capacitors reportedly do not exhibit microphonic effects [4]. However, they do have some disadvantages. They are polarized and typically have higher ESR and ESL than ceramics of similar size and capacitance.  
  3. Film capacitors: Some customers have reported using film capacitors in high vibration environments with satisfactory results. The downside is that film capacitors are typically larger than ceramic or tantalum types and can be very expensive. 

These solutions are component-level and don’t include possible modifications to the PCB, such as strain-relief cutouts. In my next post, I’ll test each of these capacitors in the same circuit and compare how sensitive they are to vibration. 

Additional resources:

Read more about an ADC TI Designs reference design, TIPD115.

Download a datasheet on OPA211, a low-noise and low-power precision operational amplifier

Footnotes

[1] Kahn, M., Multilayer Ceramic Capacitors – Materials and Manufacture, http://www.avx.com/docs/techinfo/mlcmat.pdf

[2] Piezoelectricity, http://en.wikipedia.org/wiki/Piezoelectricity

[3] Caldwell, J., More about understanding the distortion mechanism of high-K MLCCs, http://www.edn.com/design/analog/4426318/More-about-understanding-the-distortion-mechanism-of-high-K-MLCCs

[4] Cain, J., Comparison of Multilayer Ceramic and Tantalum Capacitors, http://www.avx.com/docs/techinfo/mlc-tant.pdf

Anonymous

  • The pyroelectric sensitivity is indeed an interesting point, thanks Niel. As John mentioned already it will be a challenge to seperate this parameter from mechanical stress artifacts. I've already some ideas which might be helpful in this.

    Anyway I wanted to point out an article from our colleague who looked at this phenomenon from the other perspective. He experienced audible noise from MLCC's which is basically the same phenomenon. Here is the link: www.analog-eetimes.com/.../reducing-mlccs-piezoelectric-effects-and-audible-noise.html;news_id=222903370

  • Hi Mike, I'll have to do some digging for a document on offset voltage and stress, but this is a fairly well-known effect, especially in precision voltage references. Any mechanism which applies pressure to the die can cause a shift in offset voltage (even the mold compound of the op amp package!). Limiting this effect requires proper layout of the PCB by locating sensitive circuits in low stress areas like the outside edges of the PCB or including strain relief cut-outs on the board. Chopper and auto-zero op amps are much more resistant to these effects because they are cancelling their offset in real time. I haven't personally seen microphone effects in switched-cap filters, but the basic principles hold true. Rather than the capacitor acting as a variable resistance in the filter, it will act like a variable resistance and a voltage source due to the microphonic effects. And yes, the 2.2uF caps were definitely in the signal path and not the power supply. Although the power supply is always part of the signal path, in this application the PSRR of the amplifier would suppress any microphonic effects in the bypass capacitors.

  • Very interesting discussion.  Do you have a reference for the sensitivity of an op-amp's bias voltage to stress?  You mentioned this as an aside.  Also, speaking of IC performance, any comments on these effects with switch-capacitor filters?   I had a problem that I attributed to these for a high-shock instrumentation application.  And finally, I presume your 2.2uF cap was in the signal chain and not just power supply decoupling.

  • That's a really interesting point Neil! I may need to write another blog post in the future on that effect. The challenge will be to separate the pyroelectric effects from the piezoelectric ones in any experiment I run because a rapid change in temperature will produce mechanical stress on the dielectric as well. I suspect that attaching long leads to the capacitor and keeping it off the PCB would help reduce this effect. Thanks for your comment!

  • This behavior of high- titanate ceramic capacitors is not widely known and it can create problems that are puzzling; but knowing about this will help designers to avoid these types of capacitors in applications where piezoelectric effects can create microphonics.

    By the way-- high- titanate ceramic capacitors are also pyroelectric, responding to sudden changes in temperature.