▼
Posts: 117
Threads: 31
Joined: Sep 2008
This thread says that you need 15pF capacitors, but the manual says 18pF.
So, which is correct value (correctest, in case both are)? And could you please explain the difference, what it will affect? Rate? Accuracy?
Also, there are 2 kind of 32768Hz crystals on digikey, one with 30kOhm internal resistance and another with 35kOhm. What difference does that make?
▼
Posts: 250
Threads: 14
Joined: May 2007
The exact value doesn't really matter as long as you are in the right range. I use 18 pF.
I'm not familiar with the Digikey parts on the crystal, but I use the Citizen CMR200T-32.768KZDF-UT from Mouser.
Eric
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
Actually the capacitors make a big difference to the 32Khz oscillator accuracy. You should use the exact capacitor value that is specified for the particular crystal you use in order to be within the specified crystal accuracy, typically 20 to 50ppm. If you want better accuracy you need to use a trimmer capacitor in place of one of the fixed capacitors; take a look here.
▼
Posts: 151
Threads: 8
Joined: Oct 2009
Very true, Katie. However, the crystal's specified capacitance would include the circuit board trace capacitance, the capacitance of the processor's oscillator pins and so on as well as the actual capacitors that are added. Since the processor pin capacitance is not tightly constrained, accounting for it is always an approximation. Thus the desire to add a variable cap to tweak to perfection.
All that said, many inexpensive watches gave up having a variable decades ago and still keep time sufficiently well. My '34S definitely drifts without having one but not enough to be a problem - my Casio wristwatch is there for accuracy.
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
Yeah, they got rid of trimmer caps and went to laser trimmed resistors in almost all watch oscillator modules and are able to get within a few ppm even in watches that sell for $10. Maxim sells 0.5ppm TCXO 32Khz oscillators for just a few dollars (in quantity). But why bother with that, I'll bet that you'll see GPS synced watches for under $20 in the not too distant future
Posts: 117
Threads: 31
Joined: Sep 2008
Now I am confused.
The datasheet for crystal says 12.5pF, but the design calls for 2 x 18pF (or 2 x 15 pF, depending where I look). Is it "close enough"? Or is it accounting for all those stray capacitances that should be taken care of?
I have another question aboout capacitors. When I ordered from digikey, I typed in "0603 capacitors", it gave off million pages of results, I clicked on first that seem to fit,
445-1777-1-ND
I assume that 0603 is the size factor that is used in calculator in other places. However, when the package arrived, the capacitors were like 10 times smaller. They are so small, one actually fits in the gap between 2 pads. so I, probably got the wring size.
Now, I wonder, if 0603 is the right size? Or did they sent me wrong size capacitors? What is the example part number for the right form factor capacitors?
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
18 pf will probably work but give you a slowish clock speed. If you're going to order the correct size capacitors get ones close to the value specified by the crystal manufacturer, 12.5pf.
As far as the size goes, you need 0603 (imperial), that's 1.6mm x 0.8mm. What you got was 0603 (metric) which is the same as 0201 (imperial) -- really tiny! It's confusing that there are two systems for this, but Digikey usually specifies both imperial and metric.
I don't see any 12.5pf ceramic 0603 (1.6mm x 0.8mm) caps at Digigkey but here's the part number for some 13pf ones: 490-1406-1-ND.
▼
Posts: 117
Threads: 31
Joined: Sep 2008
Katie, thank you for explanation. It seems that i did order 0603 metric. Live and learn. I guess.
About the capacitor: since there are 2 of them in series, shouldn't I be getting 2 x 25 pF caps to get resulting 12.5 pF?
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
The capacitors on the board are not in series, the are the legs of a pi-network with the crystal. While you could try to fit two 25pf caps in series to get 12.5pf it's useless to do so. These particular caps have a 5% tolerance so a nominal value of 13pf might in fact measure 12.5pf. On top of that the stray capacitance from the board will likely be several pf. So go with any value form 10pf to 15 pf and you'll be just fine.
If you really need accuracy on the WP-34s clock you'll need to use a trimmer capacitor and spend some time adjusting it to the ideal value.
I suppose that you could instead use an external, calibrated 32khz oscillator chip like this one. I haven't tried that and the processor in the calculator might not like an external 32khz clock, but I think it will work. However it's fairly low power (there might be lower power ones available), temperature compensated and extremely accurate (1 minute per year) without having to do any adjustment.
▼
Posts: 2
Threads: 0
Joined: Aug 2013
The goal is to provide the capacitance that the crystal wants to "see" across it. This is the total of the 2 capacitors in series plus the stray capacitance of the holder/package.
Two 18pF capacitors in series equal 9pF. Add 3 to 4 pF holder/stray and you end up with 12 to 13pF which is suitable for a crystal spec'ed at 12.5pF.
That's why the 18pF are specified in the manual. You could use 20pF with little difference, the frequency is not that critical.
Randy Ott
▼
Posts: 250
Threads: 14
Joined: May 2007
According to the schematics, they are installed in parallel (one on either side of the crystal), not in series. But then again, the schematics also say they are supposed to be 15 nF, which is obviously not right.
Eric
▼
Posts: 2
Threads: 0
Joined: Aug 2013
The effective capacitance in parallel with the crystal is the total of the two capacitors in series plus the Co of the crystal plus holder and stray capacitance. This is how the load capacitance is calculated. I didn't just make this up. It is how it is specified.
See this link
Randy Ott
Edited: 3 Aug 2013, 1:06 p.m.
Posts: 80
Threads: 14
Joined: Sep 2010
I'm just a hacker but whenever I build time critical circuits (that don't require a trimmer) I always use the best tantalum caps I can get.
|