CRYSTAL FAILURE

[tomd3583]

Hello, I currently have a problem with a crystal used on one of our PCBs. When the board is first mounted in the unit a faulted circuit indicator)it is tested. Once it passes the tests an LED is glued to the enclosure that the PCB is mounted in using an ultraviolet flood lamp. The enclosure is then potted with an epoxy resin and the enclosure is sealed using an ultrasonic welding process. Somewhere between the initial test and the final test the crystal that is used for the microcontroller fails. Does anyone have any suggestions? Would one company make a stronger crystal than the other? The failure rate is about 5%.

 

[itsmoked]

The epoxy resin is probably the culprit. Does it get hot? It often also runs thru an expansion cycle which can break things.

Often crystals are precoated with some silicon (electrical type!) and then potting occurs giving the crystal some "space".

You never said - SMD or thruhole?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Skogsgurra]

Ultrasonic welding? That's alotta shakin'n'rattlin' and crystals are mechanical in nature. I think the wires connecting crystal surface to pins can't take it. Vibrate loose. Probably. Try running a test lot (ten or so) crystals through the welding process alone to see if that kills them.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Comcokid]

I go with skogsgurra - the ultrasonic welding process may be damaging the crystal. I find that during our design approval process, a simple drop test can damage a crystal as well. I've even seen snipping the leads of a HC-18 type crystal with diagonal cutters do damage - the shock of the cut "snap" travels up the lead into the crystal.

If it is a leaded crystal, you can try changing it's orientation so it is subjected to the ultrasonic welders vibration through another axis. If that doesn't work, you may have to go to a ceramic resonator - that is if it will have enough accuracy for your application.

 

[tomd3583]

Thanks for tips. As for itsmoked's question it is a thru-hole crystal.

Would there be a way to protect the crystal without changing the welding process?

 

[Comcokid]

You will have to experiment some to see what might work.

A) If the crystal is a full height HC49 package, try the low profile version.

B) Try standing the crystal slightly off the board.

C) Bend the crystal over (change it's orientation).

D) Secure the crystal with RTV or wrap it in foam tape before potting so it is not as rigidly mechanically coupled to the potting.

E) Reduce the ultrasonic welder amplitude and duration cycle time.

F) Combination of the above.

G) Check out the ultrasonic web sites (Dukane, etc) for app notes. Give them a call. They have probably encountered the problem before.

 

[BobbyNewmark]

I've seen ultrasonic welders destroy crystals before. I will personally never ultrasonically weld a crystal product in a low-volume product. You can only justify the welder when volume is high enough to pay for qualifying the post-weld crystals.

What frequency is the crystal? Welders typically run at 20-50 kHz, which is a really unfriendly frequency for 32.768 kHz tuning fork crystals. If that's what you are using, you might have better luck with a 1+ MHz non-tuning-fork part.

"The failure rate is about 5%."

You mean the immediate catastrophic failure rate is about 5%. Many of the remaining units are probably running with greatly reduced margin and will fail in service.

 

[tomd3583]

When I checked the old schematic for the circuit I saw that it used to use a ceramic resonator, but when a redesign was done it was switch to a crystal at 8 MHz. Would the ceramic resonator hold up better?

The 5% failure rate is an immediate failure.

 

[BrianG]

tomd3583: you would do well to appreciate the last sentence in BobbyNewmark's post. It's not just the obvious failures you need to consider: potential latent problems like this (or ESD damage)in your products can kill your company's reputation. Can you afford that?

 

[Comcokid]

A cermic resonator would hold up better. The crystal consist of a quartz disc with silver plated contacts suspended at the end of two contact wires. This structure can easily be mechanically damages with vibration or shock. A ceramic resonator is built more like a ceramic capacitor- i.e. a solid monolithic structure.

Ceramic resonators don't have as much inital tolerance, and have more temperature drift than crystals. However, for a microprocessor clock they are fine as long as there are no critically-timed loops or measurements being made by the processor. You will need to check with the design engineers of the module.

 

[wwward]

Can you put the crystal on the outside of the finished package?

Warren