Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Another fine example:
with OPA128 from TI you can measure currents like 100fA @DC....
theoretically, but when you try, you discover that circuit is actually
wonderful seismometer. Every outside truk will give you a peak.
Reason is piezoelectrical effect in silicon cristall himself
and you cann't do nothing with this...
It's a nightmare.
On the bench everithing is working perfectly, You installa your PCB inside the box and mount it on your machine and nothing works !
Manly is due to the change of capacity between the walls of the box and the resonating elements of the PCB ( oscillators, tuned circuits, ecc ).
Vibrating the distance between the PCB and the cover change and then change also the capacitance between the two elements.
Solutions are to use THICK walls, use many screws to lock your PCB inside the box, avoid ( if possible ) to think to unscreened narrow band filters and use the filter for the PLL with the highest frequency.
Often is a cut and try work.
For your reference, I want to add that I had on several occasions a lot of trouble due to the use of high capacitance chipcaps in sensitive circuits such as e.g. PLL components and decouplings even. Especially the material Y5V just act a a microphone.... use X7R or better NPO wherever possible.
Piezoelectric effects are very pronounced in high dielectric capacitors. I know of someone who used the high K capacitors as transducers for a fish finding sonar. Magnetostriction is another problem with some inductor cores. The best bet for oscillators is air core inductors and the low k dielectric capacitors.
Everything said here is perfect, mainly the post from Flatulent.
To avoid the mechanical effect I've been using epoxy to glue the big parts as tantalum caps, choque bypass, etc... to the pcb. Also any wire should be glued using some epoxy. If lumped coil is used as part of the oscillator's resonator, it should be enclosed in some (plastic) package and this should also be glued.
I work for a wireless company. For us microphonics is defined as mechanical sensitivity of a circuit to frequencies below tens of kHz. Above this frequency generally the physical vibration is attenuated. For us, the main reason for this is, the components used in the circuit and sometimes the PCB board itself. The underlying reason is the piezo-electric effect( mechanical movement transformed into electrical and vice versa). Therea re a number of design rules we apply to reduce the effects of this phenomena.
1) we solidly mount the probable circuits to dense(inelastic) metal boxes.
2) We usually use lots of screws on the cases of the RF modules to reduce mechanical eleasticity.
3) We try to keep the compare frequencies of PLLs very high and use wide bandwidth for loops.
4) choose caps and resonators accordingly to reduce the effects.
5) Sometimes we have to shockmount some critical components.
In the old days the main problem was the tubes/valves. I have seen a special mallet with rubber ends given away by a tube/valve manufacturer for tapping them to see where the problem was.
The main problem today would be the capacitor problems described in the pervious posts.
In the past when there were several power transoformers, there was the problem of mounting them so that their magnetic fields interatcted by producing oscillating forces between the transformers at the mains frequency and the chassis worked as a sound board.
I decided to add this comment to an old thread for future readers... from my own personal experiences in aerospace vibration testing and wireless meter reading transceivers.
I once discovered (circa '94) one of a dozen types of residential westinghouse electric power meters that created microphonics with a 1GHz narrow band transmitter caused only by the rotating disc in the meter which is suspended by a magnetic air bearing. None of the RF experts could find the cause after years of isolated spurious noise in burst mode communication. ( only one vintage model of meters had delray plastic insulation on the needle of the magnetic air bearing, which was lossy at microwave and modulated the TX path via the circular disc only when rotated with the TX running, ie it did not affect the receiver.. so the plastic had a modulation effect on strong near field path loss.. solution.. retrofit only those meters (0.1% of installed base) p.s. I was just the production guy in an r&d shop.
In short any rapid change in permittivity or permeability of an object in direct transmission of a significant power or high gain or loss fluctuation in the signal path can be detected as acoustic noise even up to 1MHz in tight crystalline structures ( piezo effect)
p.s. dont rely on large ceramic caps or split gap chokes in high gain microphonic sensitive circuits. They are microphonic.