TCXO Failure Analysis

At first, I felt smart about knowing what a TCXO is. Then, it went downhill from there. Great analysis. I figured it would have been the heater component that failed, then reading the comments here, I realized I'd conflated TCXO with OCXO. Similar but not.

I tried :D

A lot of "exposed bonded die" packages caution against using ultrasonic cleaning.

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.

I've always been cautioned against ultrasonic cleaning of boards that have crystal oscillators, and indeed it's in most XO datasheets.

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!

On the origin of OXCO (for oven controlled crystal oscillator):

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).

The digital part of TCXO is interesting. It must be some simple microncontroller with lookup table that steers the frequency back to nominal value. These days you really have computation in many basic components, from crystals to flash memories.

Interesting how the depackaging was done - curious what the mill setup was looked like. It seems like achieving .001” on manual mills isn’t uncommon; which would be about 25 micrometers, so in line with the depth of passes that were being taken here. I can see how the magnified view of the part would be helpful.

The divide by two is to get the quartz small enough to fit that package.

Interesting thread.

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?

Interesting writeup!

Today I learned about TCXO.

If anyone else are curious, that component cost about $2 per piece.

That's a very cute domain name. Thank you whoever wrote this up and posted it, I'm in the process of building something that has a crystal on it and I did not realize this was a risk.

This article is a good reminder that essentially everything in your computer boils down to really precise mechanical engineering.

Can't comment on the wire bonding quality but yes you're not supposed to sonic wash anything with an oscillator. This includes ultra and mega sonic. I had always thought it was because you could damage the crystal or mems structures, so color me surprised to see this failure mode, though there still could be a shift in frequency that the scoping wasn't able to see.

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.