# Crystal Selection

Discussion in 'General Electronics Discussion' started by EasyGoing1, Jul 17, 2018.

1. ### EasyGoing1

16
2
Nov 27, 2014
Hello,

I need to replace a crystal that I busted the leads on ... the specs call for a 32.768kHz at 6pf with 45k series resistance

My question is: How relevant is the load capacitance and series resistance in terms of accuracy?

The application is a real time clock for an Arduino project using a DS1302 rtc chip.

2. ### Harald KappModeratorModerator

11,974
2,803
Nov 17, 2011
It seems the DS1302 uses the crystal in parallel resonant mode. Therefore the load capacitance is relevant and enters the frequency in a term proportional to 1/sqrt(Cload). Any deviation in load capacitance will lead to a deviation in frequency proportional to this term.

3. ### EasyGoing1

16
2
Nov 27, 2014
OK, I can appreciate that ... what about the series resistance? It calls for 45k, but I'm able to source 50k fairly easily ... would a 5k difference be very relevant?

Last edited by a moderator: Jul 17, 2018
4. ### Harald KappModeratorModerator

11,974
2,803
Nov 17, 2011
Why would you use a separate resistor? Series resistance is a property of the quartz crystal.

5. ### davennModerator

13,930
1,981
Sep 5, 2009

50k Ohm is not a standard value

47k Ohm is a standard value and from a tolerance perspective would be well within range of the 45k

6. ### davennModerator

13,930
1,981
Sep 5, 2009

yeah, that is most probably correct .... I didn't twig to the OP's comment

7. ### EasyGoing1

16
2
Nov 27, 2014
looking at datasheets from crystals on mouser, the reported spec for series resistance is usually a range that tops out at 50k so im wondering how relevant this spec is to the accuracy of the crystal ... my knowledge of crystals is pretty much nill.

8. ### EasyGoing1

16
2
Nov 27, 2014
It is the property of the quartz crystal that I am referring to. Not sure why you thought I was intending to include a separate resistor???

9. ### EasyGoing1

16
2
Nov 27, 2014
See the details on this crystal which clearly has a value of 50k. I have not been able to find one yet with all three of the required specs (32.768 kHz; 6pf; 45 kOhm), which is why I started this thread ... if 50k is "close enough" then ill buy it ... but if its going to cause a drift in time accuracy by even a few minutes every year, then thats relevant to me.

10. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,501
2,841
Jan 21, 2010
Those things aren't that accurate to begin with, good luck if you expect to get any sort of accuracy like that. The frequency will drift with temperature way more that that.

Just pick an approximately correct crystal and see how you go. If that device was any good there would be a means to adjust the frequency. If there is, you can cancel out any small difference between crystals. If not, then different crystals could vary from the mean by up to 4 seconds per day and still be in spec (and that's assuming a reasonably high quality crystal that matches the circuit correctly).

11. ### EasyGoing1

16
2
Nov 27, 2014
That's interesting... so that makes me wonder ... lets assume 4 seconds per day as an expected drift ... is it reasonable to assume that if it were 4 seconds slower one day, then it would always drift 4 seconds slower every day? Or could that change depending on whether or not its summer time or winter time (temperature variance being a factor) and if THAT is the case, could it then go a few seconds per day FASTER under some temperatures than others effectively canceling out the slow drift in the other temperature range... ???

4 seconds per day is about 20 minutes PER YEAR ... I have yet to see any timing device sold commercially that has that much drift in it.

12. ### davennModerator

13,930
1,981
Sep 5, 2009
yeah, you misunderstand what you are reading in the datasheet

As Harald said … it IS NOT an external resistance,
It is the ESR = Effective Series Resistance of the crystal

Capacitors also have an ESR value

You DONT NEED an external resistor

13. ### EasyGoing1

16
2
Nov 27, 2014
I never once thought I needed an external resistor. I am merely trying to spec a crystal that meets the requirements of my project. And I have been unable to find a 32.768 kHz crystal that also has a load capacitance of 6 pico farads and series resistance rating of 45k Ohms and therefore my question is and has been ... how RELEVANT are these values in a crystal should I be forced to purchase one that deviates from the stated requirements. So if I purchase, for example, a crystal that has a Series Resistance rating of 50k while my requirement is 45k ... will that introduce a significant problem for me?

I am still puzzled as to why people think I was insinuating that I needed an external resistor when I never made such a statement nor did I ever imply it.

14. ### davennModerator

13,930
1,981
Sep 5, 2009
you implied it a couple of times
hence why Harald and I cottoned on to it

But am pleased it finally got sorted out

I seriously doubt it
I have been using crystals, mainly in radio gear to 45 years ….. I have never had to worry about the crystal ESR

EasyGoing1 likes this.
15. ### EasyGoing1

16
2
Nov 27, 2014
Since I never once thought I needed an external resistor, can you enlighten me by citing the exact verbiage that I used that led to your conclusion that I thought I needed an external resistor? Showing me where I went wrong will help me in the future.

Thank you.

16. ### davennModerator

13,930
1,981
Sep 5, 2009

we all finally got onto the same page

EasyGoing1 likes this.
17. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,501
2,841
Jan 21, 2010
1 second per year is about 30 parts per billion.

A crystal might have a frequency stability of 0.005%, that's 50 parts per million.

So roughly speaking, that's 1000 times worse than you need.

The frequency stability is typically specified over a large temperature range (say -45degC to 85degC). In a more limited temperature range the variation will be less.

However you also need to consider the frequency tolerance (which may be of similar magnitude) that indicates the variation from the nominal frequency when load correctly at some nominal temperature (frequently 25degC). If you load it incorrectly, the variation will be larger.

If you want a guarantee near 1 second per year, you'll need to look at an oven controlled crystal oscillator (ocxo)

You might say "hey my watch is that good". Your watch uses your body to maintain the crystal at a close to constant temperature. Take it off and it will drift a lot more.

18. ### EasyGoing1

16
2
Nov 27, 2014
I had no idea that body temp was a factor in watch engineering ... fascinating!

Thank you.

19. ### EasyGoing1

16
2
Nov 27, 2014
This is the crystal I decided to purchase.

I'm sure it will work just fine.

Thank you everyone who took the time to reply, your time is much appreciated.

Mike