As we have seen in part I and part II of this series [1], there are no inherent wear-out mechanisms in ceramic capacitors. And yes, as covered in part II, the high dielectric Class 2 capacitors (X7R) can lose upwards of 85% of their capacitance with use due to applied voltage and aging, but even then, they do not catastrophically fail.
Ceramic capacitors are very useful, except where values larger than the micro-farad region are required, so for bulk decoupling applications, other types are needed. Here, we look at the alternatives.
Solid tantalum capacitors are widely used in electronics and are quite useful because of their high capacitance and moderately high-rated voltages of 75 VDC in common SMT packages. Tantalum capacitors do not have a wear-out mechanism per se, but they do have a possible catastrophic failure mechanism. When stressed under high surge voltages or currents, tantalum capacitors can fail as a short circuit. Depending on the source impedance, this may lead to a burnt-up tantalum capacitor on your board, perhaps taking your board with it.
Over the last 50 years, a 100% prevention of this problem has escaped capacitor manufacturers however, the last 20 years of testing shows that, although this is less of an issue now, the problem remains [2].
The old 1960’s military/NASA have guidelines that a resistor of around 3 Ω per volt needs to be placed in series with any tantalum capacitor that is used in ultra-reliable applications like spaceflight. Recently, the guideline has been modified to around 1 Ω per volt or by limiting the current of the circuit. This, however, defeated one of the main reasons to use a tantalum capacitor in the first place—the low equivalent series resistance (ESR) for power line filtering. Hence, I have never seen anyone do this in actual “commercial” practice. Adequate derating or surge limiting by design has been used instead [3].
It has been found that not using tantalum capacitors in high surge locations and derating the working voltage by at least 50% yields a very reliable usage scenario [2].
In other words, don’t use tantalums on the input side of a hot-plugged PCB card where you know the voltage and current surge will be high. On the other hand, usage on the output of a DC/DC converter is usually fine because the relatively slow, and possibly current limited turn-on behavior, limits the maximum possible voltage/current surge. One caveat may be a non-isolated boost converter, so think carefully about your exact circuit configuration.
To mitigate the catastrophic nature of a shorted capacitor, manufacturers have devised fusible links that are included in some types of tantalum capacitors, and for high-reliability applications, they use surge current testing to qualify either batches of capacitors or even individual units.
In other applications where the voltage/current surge is limited, such as using the tantalum in a frequency shaping network of a slow control loop where there is usually a large series resistor of 100’s to 1000’s of ohms to make a pole/zero pair, the possible surge is naturally limited to a very low value. In these cases, using a higher ratio of working voltage to derated voltage is possible with high reliability.
As with all “large value” capacitors, tantalums tend to find use in switching power supply smoothing filter applications. In these applications, the root-mean-square (RMS) ripple current can be large and, due to the tantalum’s higher ESR compared to a ceramic capacitor, there is a real danger of damaging the capacitor due to self-heating. Be sure to use a tantalum that is rated for RMS current in these applications and derate as per industry, or your in-house, requirements [3].
Note that the maximum rated voltage and the allowable RMS ripple current is usually derated with increasing case temperature. There may also be an RMS ripple current and/or ripple voltage derating with frequency, so be sure to find and study a detailed data sheet carefully.
Most common commercial tantalum capacitors are rated with a maximum case operating temperature of 85oC. Higher temperature variants are available up to 230oC.
In the last few decades, various polymer and other materials have been increasingly used to make tantalum-type capacitors. These types may be better in certain reliability aspects, such as better resistance to surge voltage failures and/or failure in an open circuit condition. Some companies even allow for a higher ratio of working voltage to rated voltage when using such components. Be sure to check your company’s policy or manufacturer’s application notes when using these “newer” tantalum components.
Finally, molded tantalum capacitors are generally very tolerant to moisture, although when you push the maximum capacitance versus case size, you will start to see moisture sensitivity levels rise, this is a very important concern with the modern high-temperature, lead-free soldering processes.
The only widespread reliability issues with tantalum capacitors that I have seen are when there is a global supply shortage. In these instances, the prevalence of counterfeit capacitors greatly increases. The only safe refuge from these issues is to stick with manufacturer-certified distributors and supply chains.
In general, solid tantalum capacitors are the most reliable of the electrolytic types (Figure 1). They have predictable performance, excellent low-temperature performance, long storage life, and when used correctly, a long service life.
Figure 1 This impedance comparison of a general purpose 10µF/10V tantalum (orange curve) versus a 2.2µF/10V ceramic capacitor (blue curve) is what keeps me coming back to tantalums time and again. That nice “low Q”, “non-ringy” impedance over several decades is just “music to the eyes” of an analog engineer.
Aluminum electrolytic types are the preferred component where very large capacitance values are needed. They are ubiquitous in nearly every electronic device where absolute size is not the first design factor. Aluminum electrolytic types are available with a large rated voltage on the order of several hundreds of volts and huge capacitance values on the order of thousands of microfarads. Specialized varieties of “super capacitors” are available with capacitance values exceeding a whole Farad, these are mainly used as battery substitutes, not bulk decoupling.
Aluminum electrolytic capacitors are the first choice in high ripple current applications, like switching power supplies. They can have very low ESR values and can support many amps of RMS ripple current. The ripple current rating is usually derated over temperature and frequency, so be sure to check the detailed specifications of the data sheet carefully for your exact application, as not all manufacturers use the same rating/derating methodology or terminology (Table 1). ESR values on the order of tens of milliohms are commonly available now, which is a switching power supply designer’s dream.
Table 1 Typical aluminum electrolytic derating multipliers. These will be different for different manufacturer’s products and intended application. For instance, capacitors intended for use after mains operated bridge rectifiers will typically reference the maximum RMS ripple current multiplier to 120 Hz instead of the 200 kHz listed here.
Aluminum electrolytic capacitors do have an expected lifetime, even when used with proper derating. Since they are typically made with a liquid electrolyte, they will dry out over time and lose capacitance and the ESR will increase. There are many high-reliability types available with defined lifetimes of up to several tens of thousands of hours, and if your requirement is for a long-lived product, I would recommend looking into these.
There are also newer polymer or organic types that promise decreased ESR and increased ripple current capability as shown in Figure 2.
Figure 2 Specialized aluminum electrolytics are available with very low ESR values now. An ultra-low ESR comes with the advantages of higher ripple current capability, but with the disadvantage of a “high-Q” impedance profile (as shown) that can cause ringing in switching power supply applications.
Generally, heat is the big killer of aluminum electrolytic capacitors because heat will tend to evaporate the electrolytic fluid in the capacitor, so adequate temperature derating is recommended along with making sure to keep self-heating with any RMS ripple current controlled. Most types are rated to either -25 oC to 85 oC or, -55 oC to 105oC. The electrolyte tends to freeze at lower temperatures, causing a raise in ESR and limiting low-temperature performance. Specialty types are available with a -55 oC to 150 oC rating.
For maximum life, voltage derating is usually recommended to be below 80% of rated maximum voltage, and RMS current is normally derated to below 75% maximum with a limit of 10 oC rise in the capacitor due to self-heating [4][5]. Some companies also specify a minimum working voltage of 20% of the rated maximum to keep the oxide layer properly formed (see below).
Because of the large physical size of most aluminum electrolytic’s, mounting issues also exasperate failures, since the first point of failure in any wet aluminum electrolyte is the leakage of the electrolyte and the only real path of the electrolyte is through the rubber end seal, it is imperative that all precautions be taken to not disturb this rubber seal during assembly or use.
The rubber seal can be damaged by bending or pulling the leads during product assembly. This can happen if the lead space isn’t exactly right on a through-hole part, or by using the capacitor bent over on a PCB and not supporting the leads during the bending operation. Figure 3 shows two ways to quickly shorten the life of an otherwise well-designed aluminum electrolytic capacitor application.
Figure 3 Aluminum electrolytic capacitors can be damaged by applying stress to the leads in the rubber end seal, ultimately causing electrolyte leakage. Damage may occur if you bend the leads flat against the body when mounting the capacitor laying down on a PCB (left), or if you mount the capacitor into a footprint with the wrong lead spacing (right). Be sure to support the leads in a fixture or with pliers before applying ANY bending action on the leads.
Another mounting issue is when a large can-type electrolytic sits on a PCB. Even shipping the PCB from one location to another can cause the capacitor to vibrate and work-harden the leads, leading to surprisingly early lead breakage or rubber seal damage. Whenever you have the opportunity to have several tall aluminum electrolytics together on a PCB, you can mount them at 90-degree angles to each other and then glue them all together with a dab of flexible adhesive (Figure 4). Anything you can do to help keep the capacitor from vibrating undamped will prevent the leads from work-hardening and failing.
Figure 4 Don’t forget to use some adhesive on any tall aluminum electrolytic capacitors to prevent vibration damage. Applying a flexible adhesive to groups of capacitors (left) or even on a single capacitor and some close-by components (right) can significantly improve the survivability of your product.
When I worked at Hewlett-Packard, we used to shake/vibration test every instrument during the design phase and ALL aluminum electroytics would easily rattle off the boards even during the low shipping-level vibration tests. Hence, we just got used to providing mechanical support for these components in every product at the design stage. Don’t ignore this important yet low-cost way to immediately improve the reliability of your products. Better yet, if you don’t have vibration testing as part of your product qualification process, start it as soon as possible.
An important factor in the care and feeding of aluminum electrolytic capacitors is cleaning. The rubber end seals are particularly sensitive to chlorinated or petrol-based cleaning solvents. This is not as much an issue as it was 30 years ago when Freon-based degreasers were used to clean boards, but it can still be a problem during touch-up or rework if you are not careful. The worst part is the failure from contamination will not show up immediately, but in the field. Appropriate use of water and isopropyl alcohol are generally approved cleaning solvents. As always, follow the capacitor manufacturer’s recommendations.
Aluminum Electrolytic Lifetime & Storage
I have many electronic instruments that are over thirty years old and all of them use electrolytic capacitors throughout, and these instruments still function fine. I have radios that are 45 years old, and they all work fine.
I have also seen industrial and consumer items using off-brand capacitors fail in just a few years, although not usually catastrophically. A few years back I walked into a Telcom room and looked at a rack of just a few years old line cards, you could see that every one of the off-brand aluminum electrolytic capacitors on these boards showed signs of ‘bulging’ indicating an overheated condition and electrolyte loss. If there ever was any truth to the saying: “You get what you pay for.” This was it.
Lastly, there is a possible storage life issue. After sitting unused, the oxide layer inside the capacitor can degrade, causing a catastrophic increase in leakage current if turned on immediately with a low-impedance source. This can either blow the mains fuses or damage the capacitor. The process of “re-forming” the oxide layer is required to prevent such catastrophic failures. Reforming is done by following the manufacturer’s recommendation. Figure 5 shows the recommendations from Vishay / BC Components on storage and reforming [5].
Figure 5 Vishay / BC Components’ typical storage life and reforming recommendations. Be sure to check your manufacturer’s exact recommendations. Source: Vishay / BC Components
We should note that as with surge failures in tantalum capacitors, storage problems in aluminum Electrolytic capacitors, in my experience, are more of a random occurrence that usually affects some, but not necessarily all of a given population. This explains why some pieces of equipment can run intermittently for 40 years or more without any issues, while others can have a failure after a short period of storage. In my experience, you are just as likely to see a failure due to lack of capacitance because of loss of electrolyte as you are to see a failure due to the oxide layer disappearing in stored equipment.
Knowing about this issue and having seen it firsthand, I make sure to turn my least used lab instruments on at least once a year for a couple of hours, just to keep the capacitors formed.
Just as the “Great Ceramic Capacitor Shortage of 2017” caused a lot of manufacturing issues for end users due to part substitutions both known and unknown, from time to time a “Capacitor Plague” hits the aluminum electrolytic market. The last one was in the 2002 to 2005 time frame [6]. All you can do to protect yourself and your products is to use reputable manufacturers who have stable manufacturing bases.
I make measuring instruments, and these have very long expected lifetimes. Hence, I have always derated my capacitors appropriately, and have used the 105oC / very long-life aluminum electrolytic types of known, reputable manufacturers, but especially in switching power supply applications where the ripple current is large. Proper use of aluminum electrolytic capacitors can measure into many decades of years of useful, and industry proven end product lifetimes.
When paralleling capacitors to get a larger ripple current rating, you have to be careful that the current is balanced in the parallel capacitors to prevent selective heating and possibly early failure in one of the parallel capacitors. The first cause of current imbalance can be trace resistance or inductance. This is especially true when trying to balance ultra-low ESR capacitors, i.e., types that are in the tens of milliohms.
The second issue is that the ESR between any set of capacitors may be as much as 2:1 from another capacitor, especially as they age, causing a ripple current imbalance between the capacitors in a parallel group (Figure 6).
Figure 6 Paralleling capacitors to increase the ripple current capability can suffer from many problems that result in uneven current sharing between the capacitors. The ESR of the capacitors can vary as much as 2:1 even in new capacitors, causing a current-sharing imbalance. In ultra-low ESR capacitors, the wiring resistance or impedance can be a source of overall current imbalance if not carefully thought out in advance.
The third issue is mounting in the final product. I have seen individual components buried in large groups of components, yet supposedly in a cooling airflow that was effectively blocked from that cooling air because of the upwind components in the way. This will cause hotter areas depending on location. This is especially true for a stack of large capacitors next to each other. The surest way to deal with this is to make sure that the temperature rise is minimized in the capacitors in the first place.
In other words, if you start with two capacitors rated at 1 amp RMS ripple current and combine them, you can’t expect to make a parallel combination that can handle 2 amps RMS current. Given a 2:1 ESR imbalance between many new capacitors, you can at best rate the parallel combination at 1.5 amps RMS current. Then you also have to consider the wiring and lifetime ESR effects into your calculations [4][5].
When paralleling capacitors, be careful to fully analyze the resulting circuit and take into account the circuit wiring, the inherent mismatch between the capacitor’s initial ESR, lifetime expected ESR, and any possible operating temperature differences due to mounting in the final product. A thermal camera used when qualifying any new product can help find these issues before they become problems in the field.
[1] Previous review of Ceramic Capacitors,
[2] Tomáš Zedníček, AVX Corporation, “Voltage Derating Rules for Solid Tantalum and Niobium Capacitors”, https://www.kyocera-avx.com/docs/techinfo/VoltageDeratingRulesforSolidTantalumandNiobiumCapacitors.pdf
[3] David Mattingly, AVX Corporation, “Increasing Reliability Of SMD Tantalum Capacitors In Low Impedance Applications”, 1994.
[3] A comprehensive derating guide: “IPC-9592 Performance Parameters for Power Conversion Devices”, https://shop.ipc.org/document-numbers/ipc-9592
[4] This old Sprague catalog is about as good an explanation of proper usage and pitfalls of using Aluminum Electrolytic as you can find, https://archive.org/details/bitsavers_spragueSprors1990_119953213/page/270/mode/2up
[5] Vishay BC components Product Catalog, “Introduction, Basic Concepts, and Definitions: Aluminum Electrolytic Capacitors”, Document Number: 28356. https://www.vishay.com/docs/28356/alucapsintrobcc.pdf
[6] The Aluminum Electrolytic Capacitor Plague of 2002-2005, https://en.wikipedia.org/wiki/Capacitor_plague
Great article with a wealth of info! Every board designer should have a copy. Question – was this a typo under Figure 4? – “Applying a flexible adhesive to groups of capacitors (left) or even on a single capacitor and some close-by components (right) can significantly reduce the survivability of your product.” REDUCE seems to contradict the points made previous to figure 4. Thanks for clarifying!
Eivad – Ouch, that’s truly embarrassing for me. Thanks for pointing the errant ‘reduce’ out – I’ll see if it can be corrected.
Thanks for the quick correction and clarification.
You must Sign in or Register to post a comment.