I recently read a technical article that scared me – that’s something which doesn’t happen often. No, it was not about negligent or rushed design, or even software bugs that went undiscovered until it was too late and very bad things happened. Instead, it was about “plain, boring” multilayer chip capacitors (MLCCs) – a widely used subclass of capacitors ¬– and some of the issues that can arise when specifying them, as well as issues related to obtaining consistency in performance.
I know passive components don’t get a lot of attention in most designs, but reality is that in analog design and the analog portion of the bill of material (BOM), there are a lot of them: resistors, capacitors, inductors, LEDs and photosensors, and transformers are the most common ones. It’s not unusual to see the number of passives on the BOM be five or more times the number of ICs. Yes, many of them are non-critical, such as a nominal 10-kΩ pullup resistor on an unterminated output, a surprising number of them are important in both obvious and much-less obvious ways. Further, as operating frequencies often reach to the GHz and multi-GHz range, their secondary and tertiary characteristics and consistency become even more important.
The article, “The ‘Relativity’ of High Q Capacitors,” in Medical Design Briefs was partially about the dimensionless quality factor Q (not to be confused with the fictional head of the research and development division of the British Secret Service in the James Bond films). It looked at capacitor design and production issues that affect the value of this parameter often considered to be a second-tier factor (capacitance, tolerance, and working voltage are generally considered to be the first-tier ones). As the article notes at its top, “For many high-power RF applications, the “Q factor” of embedded capacitors is one of the most important characteristics in the design of circuits.
This includes products such as cellular/telecom equipment, MRI coils, plasma generators, lasers, and other medical, military, and industrial electronics.” It discussed the legitimate different ways that vendors characterize Q at higher frequencies (it’s not a simple set-up or test), how small errors in test arrangement can result in relatively large errors in quantitative results, and legitimate variations of the claimed value. Other second-tier parameters include series resonant frequency (SRF) and parallel resonant frequency (PRF), whether the capacitor is designed and measured for horizontal or vertical mounting (Figures 1 and 2 ), among others.
The insertion loss of an MLCC where the electrodes are parallel to the substrate surface (Image source: Figure A Johanson Technology)
The loss for vertical mounting where the electrodes are perpendicular to the substrate surface for the same capacitor value (Image source: Figure B Johanson Technology)
After I was sufficiently on my MLCC guard, things got more frightening when the article looked at how subtle batch-to-batch changes including number of layers in the same model number unit – even from a single vendor – can change these values of supposedly “identical” capacitors, and it gets worse if you get parts from different vendors, of course. So even if you are diligent in design and specify the maximum allowable Q and ESR, what you actually get may be very different. Perhaps worse, it may vary significantly between batches, which will have an outsized impact on production, test, and performance consistency.
It’s not just capacitors that have these sorts of issues. I have always thought it ironic – or a humbling lesson in real-world physics – that the ideal transformer is initially characterized by such the simple, well-known voltage/turns relationship (Vprimary /Vsecondary = Turnsprimary /Turnssecondary ), but the situation rapidly gets more complicated. Once you start factoring in issues such as losses; self-heating; fringing; temperature coefficients and effect on wire resistance, magnetics performance, and winding arrangement (to cite just a few points), it’s a very tricky design. Then add in the realities of manufacturing variations and tolerances, and the simple idea transformer is a very complicated component, and it gets even more challenging as frequencies push in the MHz and higher ranges.
Certainly, the component design and design-in issues are greatly eased by using general-purpose multidimensional modeling and simulation tools which can simultaneously consider the linked electrical, mechanical, materials, and thermal factors (such as COMSOL) or single-purpose niche tools optimized for one component type. Still, a lot of transformer design, especially at higher power levels, relies on intuition, experience, and hands-on knowledge, plus each vendor’s property “secret” sauce.”
The good news is that there are many good passive-component information sources ranging from highly analytical academic treatises, to practical insights from vendors and engineers, to hands-on “how to do it” pieces from vendors and even experienced amateurs (the References give just a few of these.)
Have you ever been unexpectedly surprised (or “bitten”) by issues related these second- and third-tier parameters, their specifications, or variations and changes in their values?
- Extreme Analog Design: Don’t Forget Those Passives
- Go active or stay passive?
- The Zero-Ohm Resistor: A ‘Simple’ Passive Component?
- That simple sense resistor isn’t so simple
- Don’t let ESR Waste Power and Cook Capacitors
- Johanson Technology, “SRF & PRF and Their Relation to RF Capacitor Applications”
- Johanson Technology, “Q & ESR Explained”
- Johanson Technology, “Simulating the Effect of Mounting on SRF and S-Parameters for High Frequency Multi-Layer Ceramic Capacitors”
- Texas Instruments, SLUP127, “Magnetics Design 5 – Inductor and Flyback Transformer Design“
- Texas Instruments, SLUP126, “Power Transformer Design”
- University of Colorado, Fundamentals of Power Electronics, “Chapter 14 – Inductor Design”
- University of Colorado, Fundamentals of Power Electronics, “Chapter 15 – Transformer Design”
- Wikibooks, “Electronics/Transformer Design”