High quality brands of crystals?

[beaker353]

Over the years doing bench repair work I have learned there is definitely a performance, predictability, and longevity difference between component brands. Electrolytic capacitors are by far the best example. When I do the occasional small design or modification project I use this knowledge to pick top-notch components as price isn't really an issue usually. I'm not doing anything mass produced; a run of 10 would be considered a large run. I'm currently spec'ing a couple of crystals for a little project and having trouble figuring out which brands/manufactures I should be looking at for high quality options. I'm looking for a 32kHz and 24MHz crystals for a uController external clock, but I will likely get a nice random assortment for stock. Any brand suggestions from the eevblog forum collective mind? Thanks!

-EM

[MosherIV]

Hi

One brand fir crystals that has been around for a longer time and well trusted is IQD.

You can get some remarkably accurate 32.768KHz crystals, I found one with 0.03ppm inital accuracy and temp co

I would agree that the 24MHz crystal does not matter if your application is not doing any accurate timing using the micro, if it is then you may need to work out the error caused by the crystal (or the error may not be so small that it does not matter).
You can measure and trim the crystal freq by replacing one of the fixed capacitors for a variable one, do not try to measure the oscillations directly the capacitance of any measuring device will affect the circuit. Most micros have a Osc OUT on one of the GPIO pins or set up a timer or write a program to generate a freq on one of the GPIO pins and then measure that, the Osc freq will be proportional.

[kony]

I can recommend http://www.krystaly.cz/en/, especially for doing (even one off) customized crystals and filters.

[bson]

I've been very happy with Crystek, but I've only used their CCHD-575 low phase noise clock oscillators, not the plain vanilla quartz crystals.  Mouser carries some of their product line.  They do make sinewave oscillators as well; I'd give one of these a shot before rolling my own.  http://www.crystek.com

[beaker353]

For most isolated projects I agree that a highly accurate system clock isn't too critical. I have found however once you introduce external serial communications going at a fairly high rate, system clocks accuracy can become problematic. Took a while to track down some random communication buss issues to a device that relied on a ceramic resonator with terrible thermal stability (this was a commercial device, not one of mine).

Crystek looked like a good option as Mouser is my perfected distributor, but it doesn't look like Mouser stocks many options. Any feelings on ECS? Looks like plenty of stocked options in both TH and SMD form factors.

-EM

[tooki]

Any brand suggestions from the eevblog forum collective mind? Thanks!

-EM

You should probably try and get them from Walter White! (Sorry, I couldn't resist!  )

[bson]

Hey guys, the OP wanted a good quality crystal for uC, why all of you suggest high cost super stable communication grade oscillators?

Why does it matter that the crystal is for a uC?  If the uC is used to generate jitter sensitive signaling, then it will need a stable timebase.  If stability doesn't matter why use a crystal at all?  Just run off the internal oscillator and PLL.

[helius]

You should probably try and get them from Walter White! (Sorry, I couldn't resist!  )

control-f "dilithium" not found!

[Kleinstein]

At 24 MHz in other applications one sometimes uses overtone crystals - they will not run at 24 MHz, but more like 8 Mhz with an µC. So µC usually need one with 24 MHz fundamental tone. 

Even a low quality crystal is way more stable than an RC or LC oscillator. They might might be something like 1000 ppm off, but hardly more, more likely in the < 200 ppm range even for the cheap ones. This is still good enough for an UART. With RC oscillators, it takes some effort to get better than 1% - so not every RC clock inside a µC is good enough for a reliable UART.
As an intermediate the are also ceramic resonators: better than RC, not as good as a crystal, but less shock sensitive.

[Circlotron]

I found that on a uC crystals using the tall HC49 case have higher output than one using the short version of that case. The larger size of the quartz disc probably has something to do with it.

[lincoln]

do drive a non timing porject, any thing in the bin, muRata or a cti in abm7. for when it matters, Vectron, Rakon, IQD, Morion,

[calexanian]

We use tuning fork 32.768 crystals in all of our products, which are all controller timers btw. We have found Citizen to be the most stable for RTC timing applications over our temp range in the field. Abracon and Fox are also good we have found.