AT Cut and AT Strip Cut Xtals.... eh ?

[iso9001]

Hi all,

I have a project where I'm going to use an SMD crystal for the mirco clock. I've run into a problem though.

My datasheet says if its AT Strip Cut I will need a series resistor across the OSC2 line. It doesn't say what value but whatever, I'll deal with that later.

The problem is I read the definition for AT Cut (cylinder of quartz) and AT Strip (same cylinder just cut into a strip so it fits in smaller packages) but...

I have two datasheets from the same mfg and 1 says its AT Strip and one doesn't say at all. They are the same size. Am I to beleive they are both AT Strip ?

https://www.electro-tech-online.com/custompdfs/2006/07/abmm-1.pdf
and
https://www.electro-tech-online.com/custompdfs/2006/07/ABMM2.pdf

I'm going to have problems if I leave the resistor out and go with an AT Cut clock... likewise problems with no resistor and AT Strip cut.

Ideas ?

 

[hjames]

Other crystal manufacturers also have piles of documentation on the stuff, so feel free to dig through their lit. Pragmatically, I think short checkoff lists of no-no's goes something like:

1) is it an overtone crystal - most uC osc's won't work (properly) with overtone crystals.
2) serial/parallel resonance - I think pretty much everything is parallel for non RF stuff.
3) load capacitance - follow whatever the manufacturer specifies, tune for PCB layout, and pin capacitances.
4) power limits - crystals are rated for power dissipation (microWatts...). The uC datasheet will recommend the minimum power dissipation required.
5) accuracy requirements - if 200ppm isn't good enough, you'll need to read the fine print - especially with respect to temperature coefs.
6) oscillator requirements - with typical CMOS oscillators (i.e. a tuned inverter), a feedback resistor of ~1MOhm is required to bias the inverter into it's linear region. Sometimes this is built in, so read the docs on the oscillator - i.e. the uC docs.

I think the majority of the details about "cuts" primarily affect the load capacitance and the temperature coeffecients. Unless you are doing a clock/timing application, it shouldn't matter.

 

[iso9001]

Yea, I agree with most of that except that....

The datasheet for my micro (a pic 18F in this case) says a series resistor may be required on the OSC2 line for AT-Strip Cut xtals. It doesn't say WHY or of what value.

So....

The reason this is important is that I'm driving two devices that both have that note on the bottom of the osc datasheet page.

The plan is to have the xtal hooked normally to the pic (Osc1 and 2 with 18pF caps), then splice of the OSC2 line to the OSC1 of the other device. I'm planning to lower the cap value on that side slightly so they will be very similar at the xtal. (a few pF or so)

If I need a resistor, I'm not sure if the splice to the 2nd unit should be before or after that. If I knew more about why its needed in the first place, that would help.

Just seeing if anyone has dealt with this yet,

 

[hjames]

Well, the PDF for the 16F628 says in the comment on the same page that it's to prevent overdriving the crystal (i.e. the power limitation). The only other thing to be aware of is checking the voltage swing of OSC2 - you might have to bias the oscillator of the second PIC so that it works reliabily.

 

[iso9001]

Bias osc2 thats going to the other micro's osc1 ?

Not sure I'm following. They get the same VDD and VSS.

Osc2 of the other is instructed to leave open or use a pulldown to reduce system noise but increases power load, which when the clock is on I have pretty much unlimited power so I could care less there. But, I'm not sure how this will effect the osc overall. Might want to just leave it open.

 

[hjames]

The osc2 signal of the first uC might not be a "clean" CMOS logic waveform - if that is the case, then you'll need to make sure the CMOS inverter of the second uC is biased into it's linear region.

In general, there probably isn't any problem with just hooking up uC1.osc2 to uC2.osc1, but the "safe" solution, especially if you're dealing with different oscillator circuits is to AC couple uC1.osc2 to uC2.osc1, and have a feedback resistor going between uC2.osc1 and uC2.osc2 to bias uC2's inverter. The main thing is to compensate between different threshold voltages of the two different oscillators.