While power architectures of defense electronics are extremely diverse, all have common operational demands: they must be robust (shock, vibration, temperature extremes), highly reliable, able to power-up after periods of dormancy, and be based on components with minimal obsolescence. Design engineers can improve their power distribution quality over a wide spectrum of signal types from digital to RF power, while meeting the above application requirements, through the implementation of various passive components designed specifically to enhance performance and improve signal-to-noise ratios. In this article we will talk about low-inductance, low-ESR power filter capacitors; wet tantalum capacitors; RF/microwave directional couplers; and land grid array inductors and filters.
Tantalum & Ceramic: Bulk Capacitance and Low ESR
Starting at the low-frequency end of the spectrum, there are many types of applications that fall into this category. From wireless transmission to pulsed laser and radar, they are all characterized by a load that has a high instantaneous energy demand with only a finite amount of voltage drop in the circuit being allowable, as it will not operate below a certain cutoff voltage. The load dynamics are defined by the duration of the pulse over, what voltage must be retained, as well as repetition rate or duty cycle of the pulse. In the case of a pure DC hold-up requirement, the pulse length is infinite and the hold-up time will depend on the charge decay of the capacitor (or battery) supplying the circuit.
In dynamic loads, there are two sources of drop that occur. The first is the instantaneous voltage drop due to the internal resistance of the device, or ESR. The second is the capacitive drop, as the charge of the hold-up device decays over time. So for any pulse application, a high capacitance can be traded for a low ESR as long as the total hold-up criterion is met. However, the ESR and capacitance must be measured at the application frequency, not necessarily the component reference frequency. This means evaluating different technologies at the pulse repetition rate, the pulse width, and the pulse load itself.
Simply put, in order to achieve the necessary voltage hold-up for the duration of the pulse, the instantaneous voltage drop should be minimized, and for long-term hold-up, the capacitance should be maximized. Then, the device doing the holdup needs to be able to be recharged (or at least kept topped-up) from the DC power source during the cycle.
While technology is pushing to develop capacitors that combine high capacitance with low ESR, very often a combination of two technologies in parallel gives the optimum overall solution.
For avionic applications operating from a 28V bus, high-voltage solutions are needed such as 50V to 63V rated tantalum chip modules (now available to DSCC 09009), or wet tantalum axials for overall capacitance and ESR performance (updated ratings available in DSCC 93026 Rev P), coupled with stacked ceramic switch-mode capacitors for ultra-low ESR to reduce instantaneous voltage drop and provide high-frequency capacitance beyond the range of the tantalum devices. For example, AVX’s ST Series (TurboCap) combines the ultra-low ESR design of multilayer ceramic with a low ESL geometry to give the best overall combination of filtering and pulse hold-up with 50V capacitance ratings to 220uF.
For everyday applications, such as wireless digital transmission, the low-voltage-operation power amplifiers used (typically 3.6V) are ideally matched pulse capacitors - either a supercapacitor or high CV tantalum chip. This type of application is prevalent in secure handheld devices, either for voice/data transmission or data scanning/logging so low-profile devices (2.1 mm or less) are often required.
Low Inductance: Enabling High-Speed Decoupling
While low profile and overall physical downsizing for hand-held systems is a major driver, the major consideration is how to deliver high-current pulses at high-repetition rates to maintain effective decoupling for increasingly fast digital logic ICs. This is not just a consideration for handhelds, but for any system where high-speed data acquisition, processing, and transmission is required.
As we go further up the frequency spectrum, at 50 MHz we are in the regime of general decoupling for ICs, where capacitance and ESR remain prime considerations. As IC speeds increase, standard capacitors will exhibit increasing reactance. This means that the designer is actually placing an L-C network rather than a pure capacitor next to the IC in order to supply the necessary transient switching current, and performance will suffer as a consequence.
Because standard surface-mount capacitor technology is problematic (placing a 2-terminal chip on the PCB creates an inductive loop governed by the geometry of the terminal spacing), designs are required that negate the geometric effect and reduce equivalent series inductance (ESL).
By using a smaller component, the loop inductance can be reduced, but this also reduces the available capacitance for a given dielectric/voltage combination. Three methods for overcoming this are:
- Reducing the loop for a given size part. With low inductance ceramic chip (LICC) capacitors, the terminations are on the sides, rather than on the ends of the part, effectively halving the inductance for a given package size/capacitance/voltage rating.
- Separating the inductive and capacitive properties of the device. Interdigitated chips are designed to provide internal inductance cancellation when the parts are operated at high-speed ASIC frequencies in the 50-500 MHz range.
- Having extremely short signal loops when the capacitor is configured on the PCB. Land grid array (LGA) products use a termination system coupled with a vertical electrode configuration to provide extremely small signal paths that also provide low inductance at maximum capacitance.
The effect of using LGA-configured capacitors in high-speed ASIC decoupling cannot be understated. In relation to avionic systems, with the steadily increasing content of high-speed applications in aircraft and unmanned vehicles (UAVs) including telemetry, satellite tracking, forward-looking infrared (FLIR), imaging, and more, in-flight data rates are steadily increasing and it has been a principal that these new technologies can also be supplied in tin-lead termination and voltage conditioned versions for use in mission-critical avionic applications. LGA technology now enables a two-terminal device requiring only standard SMT assembly to achieve ESL performance below 20 nH previously limited to flip-chip technology.
RF Capacitor Technologies
For the RF signal side of high-speed communications, capacitor technology has continued to develop for lower ESR, higher power density, and smaller size. Basic RF capacitors are based on class 1 porcelain dielectric with NP0 (C0G or MIL BP) temperature characteristics. NP0 ceramics offer one of the most stable capacitor dielectrics available. Capacitance change with temperature is 0 ±30 ppm/°C, which is less than ±0.3% Δ C from -55 °C to +125 °C. Capacitance drift or hysteresis for NP0 is negligible at less than ±0.05%. Typical capacitance change with life is less than ±0.1% with no aging characteristics.
NP0 formulations usually have a “Q” in excess of 1000 and show little change in capacitance or “Q” with frequency. Their dielectric absorption is typically less than 0.6% and for this reason, can often be used as a mica replacement.
Typical “workhorse” examples are BP dielectric CDR 31-35 (MIL-PRF-55681), which have a series resonant frequency (SRF) ranging from 10 MHz to 4.2 GHz and is available down to 0805 size. While 0603 size is available in commercial or SCD versions, these are now nearing the limit of downsizing for microcircuit applications.
For higher-power RF applications, larger versions are available that minimize ESR and maximize Q to improve the power dissipation performance in high-power RF and microwave circuit designs. These have the highest working voltage (at 4kV) for designs requiring large biasing voltages and/or RF voltages, and retain high capacitance values ranging from 10pf-6800pf. The devices remain an ideal solution for high-power RF applications, such as MRI equipment, high-power industrial amplifiers and test equipment, antenna tuning/ impedance matching, and inductive heating devices.
Another set of devices are single-layer capacitors (SLC). These are capable of providing a SRF up to 30 GHz with 1pF capacitance values for optical applications, or with an additional MLCC in a micro-stack chip for extreme broadband. They are hybrid circuit compatible, with most applications being epoxy mount/wire bond.
One of the benefits of the SLC design is that it is a single layer. Mulitlayer capacitor devices are characterized by a sharp resonance in their impedance curves (essentially tuned devices), but due to their multilayer construction, they have higher frequency resonances that can be problematic in RF design. The single layer by comparison, has no harmonic resonances. New designs are available that incorporate the 10-GHz performance of a traditional SLC with low-frequency, low-ESR characteristics, providing ultra-broadband signal filtering.
An ideal “wish list” device would combine the stability of glass, a compact SMD design, and the clean response of an SLC. A technology is available that does just that. Thin film technology is based on the use of highly stable dielectrics (e.g. SiO2) deposited on a stable alumina base, and in wafer/dice form, so it can be used to make highly stable capacitors down to 0201 size.
It then adds additional elements: photolithography and plasma-enhanced chemical vapor deposition (PECVD) processing. The photolith gives extremely precise geometry resulting in extremely high accuracy and tight tolerance, while the low-temperature PECVD process enables the benefits of high-conductivity conductors for the electrode layers, which in turn provides optimum power=handling characteristics.
This system is characterized by extremely stable dielectric, single-layer construction that eliminates harmonics. It is readily modeled and extremely reproducible, as designs breadboarded on the bench will be precisely reproduced in mass production at the manufacturing location, month-to-month and year-to-year. Because the parts are discrete circuit elements, there are no upfront design costs and full design flexibility is maintained throughout the program lifetime.
Because of their precision, these devices can be used to fine-tune any application or modification, or even for last-minute tuning for FCC compliance etc. Another advantage of thin-film precision is the ability to supply custom capacitance values if the circuit tuning requires this.
Low-noise amplifier (LNA) applications are among the more critical sections in the receiver circuitry, and to maximize the performance, it is essential to have stable biasing and accurate impedance matching. Thin film combines discrete capacitors and inductors with high Q, low ESR, accurate capacitance (±.01pf) and inductance values (±.1nH). This not only improves the quality of the LNA, but also the yield in manufacturing by eliminating the fine-tuning of circuits in production.
The same components can be used to accomplish the critical matching of the input and output of a power amplifier. By using low-loss thin-film capacitors and inductors, more power can be sent to the amplifier transferred to the antenna. This results in improved performance and increased efficiency of the power amplifier as well as improving temperature performance.
Antenna matching itself is a critical design issue. The available real estate for the antenna is continually decreasing, which generally leads to a non-ideal form factor design. This situation will almost always require an impedance matching circuit for the antenna. Thin-film capacitors and inductors are ideal for this application, providing an accurate match of impedance to the antenna to maximize energy transfer under all conditions, minimizing losses from the PA or to the LNA.
Beyond high-accuracy microminiature capacitors and inductors, thin-film PECVD technology also lends itself to integration. By combining both a capacitor and inductor element on the substrate, an LC low-pass filter (LPF) can be formed, as shown in Figure 1. These can be made in the same small form factors (0402 and up) and use very little board space while saving cost through component count reduction. These thin-film filters provide high out-of-band attenuation (>30dB) while maintaining the lowest insertion loss available to the RF designer (<.3dB). PAs and LNAs can also be used to isolate the frequency of interest on the output of the mixer after conversion. The filters are internally matched to 50Ω so no external matching is necessary, and the conductor materials used make them capable of handling up to 3W continuous power.
A directional coupler is a device that samples an RF/microwave signal while minimizing loss to the signal. Thin-film devices, based on back-to back inductors, produce very high directivity (isolation-coupling), low-insertion-loss directional couplers. These couplers offer the highest amount of directivity found on the market today, in small package sizes down to 0402. In Figure 2, the coupler is being used to sample the output and send the sample to a gain control circuit for the power amplifier. As with the LPF, the couplers are also capable of handling up to 3W continuous power.
Directional couplers work on the principal of field coupling. The electric field produced by a transmission line in series with the signal is coupled onto an adjacent conductor through the air or dielectric medium. Coupler elements can be included within low-temperature co-fired ceramic (LTCC) modules, as the technology allows lumped elements, rather than coupled lines, to produce directional couplers to 10 dB. However, thin-film technology has a number of advantages in this area; the finer line widths maximize the coupling coefficient, making available hybrid couplers to 3dB in 0603 size.
This coupler, with port configuration shown in Figure 3, is designed to couple 3 dB of power, half to another channel, with the addition of a 90o phase shift to the signal. This can be very useful in designs utilizing an I-Q architecture where the channels are 90 degrees out of phase. By using a hybrid coupler on the output of the oscillator, the local oscillator (LO) can be generated for both I and Q sections. It can also be instrumental when using two amplifiers to improve the linearity by splitting the power between the two circuits and then recombining after amplification. This reduces harmonic emissions, improves efficiency, and increases gain from an amplifier.
This article was written by Chris Reynolds, Application Engineer at AVX Corp., Fountain Inn, SC. For more information, Click Here