TCXO Failure Analysis
serd.es
serd.es
This is especially true for TCXOs, which also have the entire loose crystal in them on top of the controller die, and for MEMS mics, which are designed to be sensitive to vibration. But it's also true for things like common CMOS image sensors, which are "exposed die", but not mechanically sensitive otherwise.
Bond wires that are hanging midair instead of being pinned in place by package epoxy don't vibe with ultrasonic cleaning methods.
The risks are usually small, mind. Which is why prototyping teams and repair shops often use ultrasonic cleaning regardless. But in actual mass manufacturing, you really don't want to risk that extra 1% failure rate. So you either ask the vendors for "safe" values and dance around those energies and frequencies, or avoid ultrasonics altogether.
Quite to the contrary, they DO vibe. Destructively :\
After that, we moved the ultrasonic cleaner to a back room.
I've also heard that one shouldn't trim the leads of a through-hole XO before soldering it into the board, since the mechanical shock of the lead breaking can ring the whole package and similarly shake it apart. I'm curious if anyone here has seen that in practice!
I’ve never put a through hole crystal into production so I can’t say anything about this conjecture.
However the larger surface mount crystals are not hard to hand solder if you get a package with side wettable flanks and make the pads reasonably large. It’s something I’d recommend considering.
Rule of thumb: one simple xtal per board in small manufacturing runs (4 digits or less) means you're fine.
The larger your manufacturing runs are, and the more sensitive components you have on your boards, the more careful you want to be. Components can easily make the difference between 0.2% failure rate and 2% failure rate, and that 2% failure rate bites when you push units by hundreds of thousands.
Of course, there's always a chance of you getting a perfect match of the exact intensity and frequency used on a given manufacturing line, which you didn't know, with what happens to kill your specific components at a disproportionate rate, which you also didn't know. But it's a pretty low chance. Feeling lucky?
Because yes, it's not actually worth the engineering/support effort for you, your manufacturer and your part vendor to actually put the thinking cap on and characterize all of that shit for a typical low volume run. So luck it is.
When a batch with 20 MHz surface-mount crystals, in a package similar to the one in the article, were accidentally run through an ultrasonic cleaner, the failure rate was immediately noticeable, in the single-digit percent.
Leads of through-hole components are usually trimmed before assembly, on both manual and automated assembly lines, (e.g. https://www.youtube.com/watch?v=cjVY8lb0LG8) and I've never seen this prohibited in a datasheet, but ultrasonic cleaning is usually prohibited.
So if someone is telling you not to trim the lead...I'll let you draw your own conclusion.
The base abbreviation is "Xtal" (for crystal) and predates modern electronics by quite a bit (was already used before 1900 in geology etc). The author linking this to Xmas (indirectly, "Christ") via the the greek Chi (Χ) is very likely correct.
In electronics this weird abbreviation (X for crystal) is further helped by the fact that "C" is completely taken by "capacitor" (an even more important passive component).
"X" because "xtal", and "Y" because of the distinct shape of a tuning fork.
Pedantically most of them aren't crystal oscillators, merely crystal resonators. Oscillators begin oscillating on their own when a DC voltage is applied, they usually are 3-pin or 4-pin devices with power input & oscillating outputs. 2-pin crystal resonators merely act as high-Q filters in an oscillator circuit, they still need other components to drive the oscillation.
I tried :D
At least that's my recollection on the mechanics of these types of resonators; the general frequency of period being proportional to length and thus your observation holding is fairly universal, though.
The 32kHz tuning forks btw use a torsional vibration mode thus not needing to be as massive as linear scaling would leave you to assume.
Given the teeny tiny endmill the author was using, I suspect they were using a small mill with a very fast spindle. Maybe something like a Taig or a Sherline.
Edit -- I see on another post the author has a Sherline 5400 mini mill.
Looking at the layout I'm very confident this controller is fully analog, with just a few gates of digital doing control stuff. There's only a few dozen to maaaaybe low hundreds of gates of digital in the whole die, not nearly enough to be any kind of MCU.
And there's a whole bunch of really large resistors and other analog stuff.
I do see a structure that might be some kind of ROM for piecewise linear calibration curves, but it's too obscured by wiring on top for me to draw any definitive conclusions without delayering the chip.
This is nice because it will help to keep the crystal on track but a lot depends on the time constants of the control circuit whether or not the Allan Deviation of the circuit as a whole is going to be acceptable across all applicable timescales.
As a domain this is both fascinating and far more complex than I had ever imagined it to be, but having spent the last couple of months researching this and building (small) prototypes I've learned enough to have holy respect for anything that doesn't have an atomic clock in it and that does better than 10^-7. That is a serious engineering challenge especially when you're on a tiny budget.
If you can use a GPS disciplined oscillator then that's one possible solution, but there to you may see short term deviations that are unacceptable even if the system is long term very precise.
I find it baffling that this would be cost effective. Maybe by dropping in a CPU core and software you save some design cost vs. a more specialized IC. But it must be more expensive per unit to manufacture in a process where you can fit in all those transistors. And these things are manufactured in such quantities that design costs must be a pretty minimal part of the final part price.
This used to be implemented as a purely analog control loop, i.e. opamps and such. After all TCXOs predate the age of ubiquitous CPUs by decades. Even if there is a need for a factory-programmed temperature calibration curve, there are techniques where it can be implemented in a pure analog way, or in a dedicated digital circuit where the transistor count will be much lower vs. a general purpose CPU core.
You need a way to calibrate the device after the package is sealed, that implies some smarts or you're going to end up with a whole raft of extra pins and that would be costlier than the microcontroller!
I'm sure there are alternative ways but in this day and age cpus and small amounts of flash + memory are priced a little bit above the sand they're made of. I have whole units packaged and with far larger capabilities for $3 Q1, and that's with a whole lot of assembly and other costly detailing.
Microchip, one particular embedded controller manufacturer lists their SMD packaged PIC16F15213-I/SN which is much more powerful than what you need here for $0.33, Q100 that drops to $ 0,27400. This is a complete device, not an unpackaged die, which would retail for a small fraction of that.
Control loops and analog stuff works well, but not if you also want to be able to do calibration after the fact package is sealed, I'm not aware of any tech that would be fully analog but that would have the same flexibility and long term stability, never mind mechanical stability (microphony, talking to a crystal is probably the cheapest and easiest way to get FM modulation!). Note that this is different precision wise from a device that simply measures the temperature and does a compensation based on that, the device you are looking at in this article is easily an order of magnitude better.
I also have a couple of very fancy ones that you can compensate and whose NV memory you can write to directly. Those are pretty expensive, $100 or thereabouts but the precision is unreal for a non-governed device.
The microcontroller approach uses a fixed number of transistors, with cheap mask ROM scaling with complexity, whereas the HDL approach scales its transistor usage with complexity. The HDL approach usually runs much, much faster, is far less error prone, and takes longer to develop.
Which approach is better depends a lot on the application.
Calibrations done with a microcontroller have replaced those done with analog components in most applications, because the total cost is reduced in this way.
Even a relatively powerful 32-bit ARM microcontroller costs a fraction of a dollar. Good analog components, with guaranteed behavior in temperature and in time, are usually more expensive than microcontrollers.
We also appear to be buying from the same supplier (Leo Bodnar).
Today I learned about TCXO.
If anyone else are curious, that component cost about $2 per piece.
I had heard similar cautions about ultrasonic cleaning with certain MEMS sensors, but I didn't realize the same concern could apply to exposed-die CMOS sensors as well.
Is the main risk the resonance frequencies of the bond wires, or more about mechanical stress propagating through the package?
I tried looking at an exemplar ECS tcxo datasheet and didn't see anything in there about washing which is surprising but it also doesn't say not to crush it with a hammer so maybe it was assumed. That's bad on them.
As for SMA to 0.1" headers: yes these are very cursed. But RF designers love putting SMAs for every connector on an eval board (power, enable, whatever) and those come in handy.