Every development board I have ever used that includes a 32.786Khz crystal has always had its can soldered directly to board.

What is the reason for this ? Is this to minimize radiated energy from the crystal during its operation (assuming that the pad its soldered to is ground) ?

If the can was not soldered to the board, what would be the effect ?


If you look at this image, the crystal has 3 solder points. Two for the legs, and one for the case/shell/can (green arrow).

My question is about the reason for soldering the case/shell/can to the board.

enter image description here

  • \$\begingroup\$ Note that you often find tth cans strapped down onto a pad by a stripped wire soldered on both side. \$\endgroup\$ – PlasmaHH Mar 6 '15 at 9:47

"Radiated energy" is the right idea, but your question suggests that you are thinking about electromagnetic energy here. It's rather mechanical. 32768Hz is ultrasound with an arial wavelength of about 1cm, and a crystal oscillator works via mechanical vibration. If you want to have constant Q and constant frequency response, fixing the case mechanically makes sense.


32kHz and 32.768kHz crystals are physically large. The lower the resonant frequency, the larger the crystal - the higher the resonant frequency, the smaller the crystal. So you might expect the case and leads to be larger than higher frequency crystals which obviously have leads that are strong enough to support the can, crystal, and all.

However what you find instead is that the can is just barely larger than the crystal itself, and the leads are very, very thin - not strong enough to support the weight of the can by themselves.

The reason these low frequency crystals have such seemingly poor packaging methods is due to the watch industry.

There are 1.2 billion watches sold each year. The majority of them are inexpensive digital watches, requiring a small, 32kHz crystal. The crystal can is securely fixed inside the watch body using pressure or glue, while the leads are soldered to the circuit board. There is no need for strong, supportive leads.

So to keep the cost of hundreds of millions of crystals low, the leads are thin, the can is small and thin, and it's produced as inexpensively as possible. This enables cheap, small, lightweight watches.

As a result, these crystals are extraordinarily inexpensive - in their existing form. You can get more expensive 32kHz crystals with better mechanical fixing, but they cost 10 to 100 times more in quantity than these inexpensive watch crystals.

Some manufacturers have adapted them by bending the leads and putting them into reel packaging appropriate for surface mounting use, suggesting a pad and soldering layout that allows the can to be affixed using solder. Since the boards are already going through a solder paste, placement, and oven process, this requires no extra steps, such as gluing, hand soldering, wave soldering, etc and reduces the labor required to use these devices.

So this all comes back to using a part intended for another big industry in your design, and having to deal with the tradeoffs that suit the other industry, but might have to be accounted for in your design.


I think you'll find more information in this old answer Is case grounding compulsory in typical 32.768kHz crystal for Real Time Clock?

I would have said it was more for mechanical reasons and that those packages they typically use are the cheapest possible. They're basically old through hole packages converted to surface mount that wouldn't have much mechanical stability without soldering the case down.

However another person in that answer notes that it may help with parasitic capacitance.

  • \$\begingroup\$ The mechanical part makes sense. Is there a reason then why higher frequency 2 pin crystals (8Mhz, 12Mhz etc..) don't ground its case for similar parasitic capacitance reasons ? \$\endgroup\$ – efox29 Mar 6 '15 at 4:47

It's done because it's cheaper than providing a socket.

32.768 KHz is used because it can be divided down to 1 Hz. Systems which do it this way typically don't require an enormous amount of accuracy, so an external crystal is used in conjunction with an on-chip oscillator, rather than a complete oscillator which can be trimmed to final value as part of the assembly process.

ETA: (after the new picture and a bit of clarification in the OP) Ah. There are at least 2 reasons. The third tab is the case - this provides physical attachment for the crystal and keeps the legs from failing as a result of vibration. It also provides a fixed capacitance to ground for the internal leads, which helps keep the frequency constant.

While it does (also) provide shielding, this is not a particular concern. The crystal is working at fairly high voltages (a volt or more counts as high in this case), and has an extremely high Q, so it's hard for radiated noise to affect it significantly. Plus, of course, there really isn't a lot of radiation going on at 32 kHz. Everything else on the board will be running in the MHz range.

  • \$\begingroup\$ I don't think I explained my question properly at all. Give me a few mins and I'll add a picture. I know what the xtal is used for lol \$\endgroup\$ – efox29 Mar 6 '15 at 4:22
  • 1
    \$\begingroup\$ This answers why you'd provide a watch crystal, not why you'd solder its can down. \$\endgroup\$ – Ignacio Vazquez-Abrams Mar 6 '15 at 4:23

One reason to solder the case to the board is purely for mechanical stability. If you want to lay the case down to minimize height, soldering it is more stable than just leaving it loose to move around. This is probably the main reason you see it done.

There are also electrical reasons to solder the case to the board.
One reason is to ground the case to prevent damage from ESD if you plan on touching the board near the crystal frequently. Usually in production this is not needed, but if you're testing/programming a board and plan on handling it a lot this might be a good idea.
There is also a potential coupling path from the case of the crystal to the crystal itself. In most cases this doesn't matter because it is very small, but in the rare occasion it can be significant.
SHG's mention of parasitic capacitance as well is not going to be an issue most of the time.

The electrical reasons for soldering the can are just not very valid in most projects/situations. If I had to bet I'd say the boards you've seen with the cans soldered to the board were done purely for mechanical reasons, and there wasn't an electrical reason behind the decision.


Gluing, which is the other common option, is used when the case is smaller and the required temperature range is very narrow. It is difficult to get an adhesive which is both a "good glue" and also very temperature and water stable. Even more important, it adds an extra step -- even if you are already gluing other components, but particularly if you are NOT already gluing other components.


Not the answer you're looking for? Browse other questions tagged or ask your own question.