Rendered at 12:17:58 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
ACCount37 23 hours ago [-]
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.
superxpro12 22 hours ago [-]
"...dont vibe with ultrasonic cleaning...."
Quite to the contrary, they DO vibe. Destructively :\
dlcarrier 18 hours ago [-]
I worked somewhere that accidentally ultrasonically washed a product batch with crystal oscillators in them, and the failure rate was probably even higher than 1%. It was very immediately noticeable when testing the product. There was also an above-average failure rate in the PMIC inductors.
After that, we moved the ultrasonic cleaner to a back room.
ACCount37 17 hours ago [-]
Just one XO per board is usually quite safe, but you do get those freaky highly sensitive outliers every once in a while. There are methods to reduce the risks.
myself248 1 days ago [-]
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!
Aurornis 23 hours ago [-]
> 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.
summa_tech 20 hours ago [-]
Sometimes, the bigger physical size of through-hole crystals gives them a higher Q. I, too, prefer surface-mount everything but have been defeated on that sadly ;(
dlcarrier 17 hours ago [-]
I worked on a product that included pre-trimmed HC-49/S through-hole crystal oscillators, and not a single crystal failed. It was a low-volume product, but there were still probably tens too hundreds of thousands of them built.
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.
the__alchemist 24 hours ago [-]
I went down this rabbit hole a few years ago, and couldn't find an actionable answer on if this is OK or not. Sounded like "No, you shouldn't", but almost every PCB I've designed (or used?) has at least one, and I know ultrasonic cleaning is a thing, so I'm not sure how to reconcile these.
ACCount37 23 hours ago [-]
There is no single answer. It depends on the exact components, their sensitivities, frequencies and energies used, and how much failure risk are you willing to take.
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.
garaetjjte 22 hours ago [-]
On the other hand I heard that one shouldn't trim leads after soldering as it might crack solder joints...
codehero 21 hours ago [-]
Depends on the quality of the solder joint.
Poor quality solder joints do not survive mechanical shock.
If you are fighting a GND pin that sinks a lot of heat, using leadfree solder and you aren't that skilled...don't trim that lead flush with the PCB.
Otherwise, if you are sure that the solder has wicked into the hole, trim away.
So if someone is telling you not to trim the lead...I'll let you draw your own conclusion.
jnd-cz 21 hours ago [-]
This depends how close to the solder joint (or to board) you are trimming. If you're already cutting solder together with the component lead then it's too close and can affect the quality. I'm sure the NASA soldering manuals show this in great detail.
jacquesm 24 hours ago [-]
Oh, that's a good one, I can see how that would put a lot of g's on the package. I think this will be a factor depending on the weight of the total assembly. If that weight is significant it will dampen the shockwave.
myrmidon 23 hours ago [-]
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).
ACCount37 23 hours ago [-]
Amusingly, "crystal oscillator" can be both "X" and "Y" in schematics.
"X" because "xtal", and "Y" because of the distinct shape of a tuning fork.
SAI_Peregrinus 23 hours ago [-]
Even more amusingly, only low-frequency crystals (very often 32.768kHz) are tuning-fork cut, high-frequency resonators use other shapes.
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.
kens 20 hours ago [-]
The article asks about the etymology of X for crystal. I looked into that a while ago. The abbreviation "xtal" has been used for "crystal" since the 1800s in medicine, geology, and chemistry, and then electronics copied the usage. This comes from the earlier use of X for the "christ" sound, as in "xmas", which goes back to the 16th century. As the article suggests, the Greek chi (Χ) is the root.
geocrasher 21 hours ago [-]
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
namibj 24 hours ago [-]
The divide by two is to get the quartz small enough to fit that package.
superxpro12 22 hours ago [-]
Is there some kind of inverse relationship between resonant frequency and crystal size?
namibj 22 hours ago [-]
Yeah, the resonant waves tend to be surface flexing (the top side expanding the bottom contracting and thus the sheet as a whole bending; and ofc vise-versa) which has a natural frequency related to the lever arm length amplifying the moving mass vs. the more centrally located elastic crystal lattice deformation forces acting as the spring.
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.
rmast 23 hours ago [-]
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.
showerst 23 hours ago [-]
A thou on any decent mill is no problem.
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.
jnd-cz 21 hours ago [-]
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.
jacquesm 21 hours ago [-]
Yes, the typical way this works is that the lookup table is programmed during device calibration and that the microcontroller has a temperature sensor attached and uses a varicap to drive one of the two capacitors attached to the crystal (usually through a coupling capacitor to avoid loading up the circuit too much).
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.
avian 20 hours ago [-]
Is there really a microcontroller in there? As in a general purpose microprocessor core executing machine code in ROM? Any references for that?
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.
jacquesm 19 hours ago [-]
That microcontroller costs a small fraction of the precision ground crystal it is boxed in with.
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.
explodingwaffle 18 hours ago [-]
just because it is digital doesn't mean it has to be a microcontroller though, right?? i see no reason this wouldnt just be a state machine or whatever out of plain old logic.
jacquesm 17 hours ago [-]
Well, I've been working with a number of these devices, different brands but the same or slightly more functionality and they all have little controllers in them. Some are documented and you can talk to them directly (usually I^2C) others are 'black boxes', you can tell there is something living on the other side of the nominally 'NC' pin but not what and you don't have control over it.
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.
dlcarrier 17 hours ago [-]
There's definitely enough transistors in there for a microcontroller. It only takes a few thousand. If you're building a custom integrated circuit in the first place, the cost of a microcontroller core is relatively low, and often the cheapest option. The alternative is to write the logic in a hardware description language (HDL) like Verilog, and implement it with logic gates.
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.
adrian_b 19 hours ago [-]
If you implement a temperature-calibration curve by analog means, it will drift in time, unless you use very high-quality and expensive components.
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.
jacquesm 24 hours ago [-]
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.
_Microft 16 hours ago [-]
That's a website of Andrew Zonenberg if that name rings a bell.
It did not, thank you for the pointer, amazing stuff he's working on.
We also appear to be buying from the same supplier (Leo Bodnar).
kentrf 1 days ago [-]
Interesting writeup!
Today I learned about TCXO.
If anyone else are curious, that component cost about $2 per piece.
the__alchemist 24 hours ago [-]
Yea! Useful if you need precise timing under temp swings. I use them for UAS LoRa radios. Def more expensive than a normal XO!
flyinghamster 22 hours ago [-]
It's also something you want to look for if you're buying an SDR. Getting one with a TCXO will eliminate frequency drift, and the better-made SDRs will also have little or no need for frequency correction.
jacquesm 24 hours ago [-]
But cheaper than OCXO and far less power consumption too. Also (much) less stable.
galangalalgol 24 hours ago [-]
Compensating for the temperature will never be as accurate as actually controlling it (O is for ovenized). I keep reading about chip scale atomic clocks coming down in price but I've yet to see them as the oscillator in anything mass produced.
rasz 23 hours ago [-]
When 2G started being decommissioned ebay was suddenly flooded with super cheap rubidium frequency standards from parted out base stations.
jacquesm 21 hours ago [-]
I'd love to have some.
jnd-cz 21 hours ago [-]
Also cheap OCXOs.
JanoMartinez 21 hours ago [-]
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?
Neywiny 1 days ago [-]
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.
jmole 22 hours ago [-]
This article is a good reminder that essentially everything in your computer boils down to really precise mechanical engineering.
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.
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.
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.
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.
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.
https://news.ycombinator.com/user?id=azonenberg
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.