[Question]How to solve start-up time issue in xtal osc sim-

EDAboard.com Forum Index -> RF, Microwave, Antennas and Optics -> [Question]How to solve start-up time issue in xtal osc sim-

Author

Message

GDF

Joined: 11 Aug 2005
Posts: 175
Helped: 1

18 Nov 2005 8:46 howto oscillate a crystal

How to solve start-up time issue when we do the simulation of high-Q xtal
oscillation?
For example, the oscillation frequency is 100MHz, but the Q of xstal is 1000.

Thanks,

Grig

Joined: 03 Sep 2004
Posts: 227
Helped: 26

18 Nov 2005 18:12 Re: [Question]How to solve start-up time issue in xtal osc s

Hi

What is "start up time issue"?

Thanks

Google
AdSense
Google Adsense

18 Nov 2005 18:12 Ads

crystalguy

Joined: 14 Sep 2004
Posts: 13

23 Nov 2005 4:16 Re: [Question]How to solve start-up time issue in xtal osc s

I think you have a problem with your crystal model. The Q you mention of 1000 is way off of any sort of crystal I am familiar with. The unloaded Q should be more like > 50,000. This being said, it is difficult to make recommendations about how to solve start up problems without some information on how you are simulating this circuit. I would recommend that if you are looking for something like accurate results than you take the high road and look at an analytical solution. In most numerical solutions the margin of error can be quite unpredictably large. Hope this helps,
Jim

vfone

Joined: 10 Oct 2001
Posts: 2328
Helped: 327

23 Nov 2005 8:52 Re: [Question]How to solve start-up time issue in xtal osc s

The Q value of the X-tal will decrease in inverse proportion to the frequency, if the radial dimensions are constant. The best Q value is obtained around 5 to 10 MHz. At 100MHz the Q value could be in the range of a few thousands.
To reduce the start-up time tries to increase the bias current of the oscillator, to put this one in the linear region. Also you can increase the Vcc, but probably you don’t have margin for this.

crystalguy

Joined: 14 Sep 2004
Posts: 13

23 Nov 2005 17:14 Re: [Question]How to solve start-up time issue in xtal osc s

Well, I don't know where you are getting your crystals from (or rather, your models) but the crystals my company manufacturers fall well within the handy rule of thumb for quartz crystals. Unloaded Q is approximately equal to: Q= (2 to 10) X 10^6 for AT cut devices. This is of course dependant on weather the device is fundamental, third or fifth overtone and what package is used. For At cut devices the higher Q range is for high end oscillator designs. For SC cut devices used in precision oscillators the equation is: Q=12 X 10^6. Here are some models for some devices of recent manufacture.
100MHZ AT third overtone: C3=.5fF, L3=5.066mH, R3=43ohms, C0=2.25pF Q nearly 80,000
155MHZ AT HFF fundamental: C1=5fF, L1= 2.108mH, R1=12ohms, C0=1.5pF Q near 20,000

Keep in mind that these models are taken at 100uW drive level. Since a crystal is not linear these models are gross approximations that do not include spurious responses and unwanted overtone responses. Importantly, you need to be aware that quartz crystals at very small current levels (as in the first moment of startup) and very large drive levels can have somewhat higher Resistance values, 2 to 10 times higher than at steady state. Generally, higher levels of anomalous startup resistance tend to occur over time ( 1 to 2 years after the device is made) and can come from very small amounts of contamination. Your design should be able to have enough excess negative resistance to oscillate a crystal with a resistance of 5 times the nominal resistance.

Added after 2 minutes:

WHoops that is :
Unloaded Q is approximately equal to: Q= (2 to 10) X 10^6/frequency in MHZ for AT cut devices. This is of course dependant on weather the device is fundamental, third or fifth overtone and what package is used. For At cut devices the higher Q range is for high end oscillator designs. For SC cut devices used in precision oscillators the equation is: Q=12 X 10^6/frequency in MHZ.