Crystal load capacitance and using the wrong caps

[PeteAU]

Hi Guys,

I reordered some crystals for an existing design but this time they're a different brand/make and I just realised a totally different load capacitance of 30pF. (Element14's product page says so, but it's not in the datasheet per se, depends on manufacturer order code?).

I'm using these crystals with a microcontroller and 27pF caps, which I realise are now the wrong value.

The crystal wants to see 30pF and I'm giving it ((27*27)/27+27)+5 = ~ 19pF.

Can I get away with it? How bad will it be? If I build 100 units, could some of them refuse to fire up?

[mrkev]

Differences of few % in this case doesn't matter. Those capacitors don't need to be very precise, because they are used just as "kind of load", lowering the amplitude on xtal. So 27pF vs 30pF is completely OK. You could check on oscilloscope what would f.e. 10pF vs 47pF do.

I'm using these crystals with a microcontroller and 27pF caps, which I realise are now the wrong value.

The crystal wants to see 30pF and I'm giving it ((27*27)/27+27)+5 = ~ 19pF.

I don't really get where you came up with this value. If it's standard connection on each side of xtal, you can't calculate it like this...

[alm]

Differences of few % in this case doesn't matter. Those capacitors don't need to be very precise, because they are used just as "kind of load", lowering the amplitude on xtal. So 27pF vs 30pF is completely OK. You could check on oscilloscope what would f.e. 10pF vs 47pF do.

Note that even a good 10x scope probe will have a capacitance of approximately 10 pF.

I don't really get where you came up with this value. If it's standard connection on each side of xtal, you can't calculate it like this...

Seems pretty standard to me. Compare this appnote, page 2. Note that Fig. 6 gives a typical curve of load capacitance vs (relative) frequency.

[Hatimoooo]

Hello.
You can check your microcontroller's datasheet. I've seen some before and they offer an interval for crystal capacitors, check if yours are in the interval

[mrkev]

I don't really get where you came up with this value. If it's standard connection on each side of xtal, you can't calculate it like this...

Seems pretty standard to me. Compare this appnote, page 2. Note that Fig. 6 gives a typical curve of load capacitance vs (relative) frequency.

Well ok, my fault . I didn't get the jump from parallel capacitors to overall load capacity (as it's not really clear if those 30pF are ment to be load capacitance or two capacitors that this manufacturer recommend to use).

[PeteAU]

I think my post wasn't very clear Here is the excerpt from farnell.

So from that, I assume I _should_ be using two 48pF caps, but that seems really high, you don't see that every day. Am I on the wrong track?

[mariush]

Where do you see 54pF, how do you figure that out?  It says right there, 30 pF load capacitance...

[sleemanj]

Where do you see 54pF, how do you figure that out?  It says right there, 30 pF load capacitance...

Indeed, according to the formula in the appnote alm linked for calculating load capacitance,

this appnote, page 2.

in order to get a 30 pF load capacitance, one should have 50pF capacitors.

[mrkev]

Well, I would do this. Take one board with 27pF caps, check the frequency after the oscilator (if you have uC, just use Timer/counter output or something) with good log. counter. Then mout another 27pF caps at the top of caps that are already there (so they are parallel). Check frequency one more time. If you see big improvement, change the caps; if you don't see any improvement, don't bother...

[mikeselectricstuff]

A good guideline is that a "happy" xtal oscillator will have a similar amplitude on each pin. Remember you need to use a low capacitance scope probe to test thos - x100 probe, or put a 10M resistor on the end of a x10 probe