Maxim Integrated recently posted a series of application notes chronicling how there’s more going on than you’d think in even the simplest “passive” components. Nothing’s safe: capacitors, resistors, and even printed circuit boards can all behave in non-ideal ways, and that can bite you in the reflow-oven if you’re not aware of them.
You might already know that capacitors have an equivalent series resistance that limits how fast they can discharge, and an equivalent inductance that models departures from ideal behavior at higher frequencies. But did you know that ceramic capacitors can also act like voltage sources, acting piezoelectrically under physical stress?
For resistors, you’ll also have to reckon with temperature dependence as well as the same range of piezoelectric and inductance characteristics that capacitors display. Worse, resistors can display variable resistance under higher voltages, and actually produce a small amount of random noise: Johnson Noise that depends on the value of the resistance.
Finally, the third article in the series tackles the PCB, summarizing a lot of potential manufacturing defects to look out for, as well as covering the parasitic capacitance, leakage currents, and frequency dependence that the actual fiberglass layers themselves can introduce into your circuit.
If you’re having a feeling of déjà-vu, the same series of articles ran in 2013 in Electronic Design but they’re good enough that we hope you won’t mind the redundant repetition all over again. And if you’re already quibbling with exactly what they mean by “passive”, we feel your pain: they’re really talking about parasitic effects, but we’ll let that slide too. We’re in a giving mood today.
[via Dangerous Prototypes]
Ceramic caps acting piezoelectrically? I am so going to try that!
Good article, please keep stuff like this coming. Despite what some folks around here say, we’re not all uber engineers with 50 years of experience. Some of us are still starting out and articles like this are incredibly helpful. Thanks!!!
Yup, that was a nightmare when building RF oscillators: ceramic caps (and coils) have a strong microphonicity so that if you talk loudly in front of them you can hear yourself in a receiver even if the oscillator has no modulation circuit. If you want to try it just build the simplest single transistor free oscillator (no xtal), then tune a radio to its frequency (or a strong enough harmonic), then gently hit the breadboard/pcb and you’ll hear the hit coming from the radio loudspeaker. Now put the circuit in front of the radio, raise the audio volume and you have some good larsen effect …without a mic!
Explanation: Caps and coils slightly change their value according to mechanical solicitations; if they’re used in a LC oscillator, which is the case here, it will change its frequency according with the vibrations, effectively behaving like a low efficiency microphone. This also applies to xtals which are themselves piezoelectric devices, but they’re built to minimize this effect so it’s harder to detect.
If you’ve ever heard a switching supply make a high-pitched noise, it is almost entirely from the piezoelectric effect causing ceramic capacitors to vibrate at the switching frequency or some other harmonic.
I remember learning about the piezoelectric qualities of ceramic caps the hard way. One of the first real product designs I ever did was a GPS receiver for kids. It was straightforward: a GPS chipset, some power circuitry, and a passive LCD. The design was very space-constrained, and on an early spin of the board there was a PCB library issue that led to some 0603 ceramic caps being populated with 0805s. As it turned out, these were right behind the LCD, and there was so little clearance that the LCD glass actually contacted the glass. Piezoelectric element plus big plate equals bug speaker. You could tell what mode the GPS chip was in by the tones emitted by the screen. The client was not happy, but after some poking and prodding we figured out what had happened, swapped out the caps and all was well.
Ugh…the LCD contacted the *caps*. *Big* speaker.
Bug speaker actually makes sense in this case. Glad it was a relatively easy fix.
This also happens fairly commonly in QI wireless charging pads as something to do with the implementation is just the right frequency for the caps to act piezoelectricity, usually only kids can hear it though.
As an older engineer I had to learn of a lot of these effects the hard way. They can be some of the most vexing problems to chase. What amazes me with younger engineers (fresh out of school) is having to constantly remind them that a capacitor or resistor is not an ideal component and circuit boards have parasitic capacitance and inductance.
This is the kind of info I only learned at a tech college, never at university. Our college lab instructors would berate us for ugly wiring on our breadboards and make us redo it, whereas at uni all we got was some poor overworked grad student who knew less about the lab than us!
Microphonics is still alive and well https://en.wikipedia.org/wiki/Microphonics
Some LEDs have been known to generate sufficient voltage that they glitched the micro if someone shone a bright torch at them, or for that matter a camera flash.
Lesson learned, if you aren’t using input mode set *all* unused pins to high Z.
I seem to remember this happened with the RaPi2 when people took photos with a flash bulb.
How can you be sure that wasn’t caused by the massive electrical signal produced in a camera flash (as opposed to the light from the flash charging an LED)?
Because the camera in question was not electrically connected to the Pi.
Don’t you mean low Z?
>if you aren’t using input mode set *all* unused pins to high Z.
Input mode is high Z. Do you mean drive the pins to low so that they won;t float? Telling people to float CMOS inputs is bad advice.
I’ve seen circuit boards specifically designed around that. A very low power sensor node that harvests power from a few red LED (basically using them as a cheap solar panel) and uses that power to broadcast a temperature reading over RF in a large enough interval (something like once every few seconds).
As a demonstration, I’ve always wanted to combine two parasitic characteristics of ceramic caps: piezoelectricity and voltage-capacitance dependence. Glue a plate to the top of a few caps to amplify the piezo motion, then construct an RC oscillator with high transient currents to excite the caps. Next, vary the DC bias to adjust the capacitance and thus the frequency of oscillation. Bam, instant VCO plus audio output in one component!
Meh, if you use good design practices this stuff will rarely bite you. I work with high voltage stuff around microwave RF daily and rarely have too many issues with passives. But we make it a point to use non-inductive resistors when they are called for, caps that are made for high-voltage / high frequency etc.
Being aware is a good thing, but unless your hobby is building LNAs at microwave frequencies operating in radars right next to a high voltage microwave tubes, most of your problems aren’t going to be due to Johnson noise, stray capacitance and or inductance or worth fretting over ESR of your caps.
Getting smart about “ground” and ground planes and the gotchas there would be something to spend some time learning about. I’ve built some pretty complex high frequency digital and even RF stuff on breadboards that would make most high frequency guys cringe and have it work. But I have also had simple digital panel meters flip out on simple analog signals where some randomly placed ferrite rings saved the day.
I guess what I’m saying, be aware, but don’t start pulling out the physics books first when your blinky LED doesn’t blink the first time