Why does crystal oscillator run slow?

The difference is about 60ppm. Might be accounted for by differences in parasitic and device capacitances of the two circuits.

"Bill Bowden"

** You sure that crystal is 30kHz ??

Rather than 30 MHz ?

.... Phil

And the temperature, and the voltage exciting the crystal, and so on. What works at 5V and 72F, might not be quite the same at 1.5V and 65F. I guess at the end of the day, is the frequency difference that big a deal to the application. If he is doing this for educational purposes, he has been duly educated.

It could be the difference between running it series-resonant vs. parallel-resonant, but more likely is the difference in the load capacitance; it's possible to "pull" a crystal's frequency quite a bit with suitable capacitors.

Hope This Helps! Rich

Yes, it's 30KHz. I have a collection. 100K, 50K, 30K, 20K, 16K,

163.84K, 166.70K, and others. All in big cans. No idea what they were used for, just playing around to see if they work.

-Bill

Every time I've tried to "pull" a crystal, it goes slower.

-Bill

Well, yeah, with a parallel capacitor. And you _did_ say that your experimental circuit is giving you a lower frequency. So figure out how to reduce the stray capacitance, and find out what the proper load capacitance is, preferably by looking up the spec on the crystal, which apparently you don't have, or by examining the original circuit.

Good Luck! Rich

--
Can you post the circuits, please?

I would try to replace the load resistor with a potentiometer of 5kohm and slowly reduce the feedback voltage applied to the crystal. Reduce the oscillator amplitude, while start up is still reliable. That's better for the crystal(the amplitude might be too big), and the lower coupling might move the frequency closer to parallel resonance.

Try adding an inductor in series with the xtal. It usually works to counteract the effect of pulling capacitors or stray capacitances that slow down the frequency.

This page may be of some help.

m

Good idea, never thought of that.

-Bill

I tried to simulate it with LTspice but didn't know how to specify the crystal. The bench unit starts and runs at 29.982 (should be 30KHz) with a 1.3 volt supply. Frequency is better at 5 volts at 29.998. I'd like to make it run at 1 volt. The simulation runs slow with 1nF cap in place of crystal. Also duty cycle is about 30% for the bench unit, and much less for the simulation.

Version 4 SHEET 1 880 680 WIRE -176 -80 -448 -80 WIRE 128 -80 -176 -80 WIRE -176 -32 -176 -80 WIRE 128 -32 128 -80 WIRE -448 32 -448 -80 WIRE -176 64 -176 48 WIRE -176 64 -320 64 WIRE -176 80 -176 64 WIRE -96 80 -176 80 WIRE 16 80 -16 80 WIRE 128 80 128 48 WIRE 240 80 128 80 WIRE -176 112 -176 80 WIRE 128 112 128 80 WIRE -320 160 -320 144 WIRE -240 160 -320 160 WIRE 16 160 16 80 WIRE 64 160 16 160 WIRE -448 176 -448 112 WIRE 128 240 128 208 WIRE -176 256 -176 208 WIRE -320 336 -320 160 WIRE -80 336 -320 336 WIRE 240 336 240 80 WIRE 240 336 -16 336 FLAG 128 240 0 FLAG -176 256 0 FLAG -448 176 0 SYMBOL voltage -448 16 R0 WINDOW 0 -37 3 Left 0 WINDOW 3 18 4 Left 0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value 1 SYMBOL npn3 -240 112 R0 SYMATTR InstName Q1 SYMATTR Value 2N3904 SYMBOL res -192 -48 R0 SYMATTR InstName R3 SYMATTR Value 30k SYMBOL Misc\\xtal -16 320 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C1 SYMATTR Value 1n SYMBOL npn3 64 112 R0 SYMATTR InstName Q2 SYMATTR Value 2N3904 SYMBOL res 112 -48 R0 SYMATTR InstName R1 SYMATTR Value 5.6k SYMBOL res 0 64 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R2 SYMATTR Value 100k SYMBOL res -336 48 R0 SYMATTR InstName R4 SYMATTR Value 1Meg TEXT -480 408 Left 0 !.tran 10m

-Bill

On Wed, 6 Apr 2011 22:46:34 -0700 (PDT), Bill Bowden wrote:

--- Here's a great resource:

with a nice tutorial about crystals and oscillators:

Also, I substituted the crystal equivalent in your circuit, and it works, but not knowing anything about the crystal itself probably limits its usefulness.

Here's a link on how to measure the crystal parameters:

and the modified netlist:

Version 4 SHEET 1 880 680 WIRE -176 -80 -448 -80 WIRE 128 -80 -176 -80 WIRE -176 -32 -176 -80 WIRE 128 -32 128 -80 WIRE -448 32 -448 -80 WIRE -176 64 -176 48 WIRE -176 64 -320 64 WIRE -176 80 -176 64 WIRE -96 80 -176 80 WIRE 16 80 -16 80 WIRE 128 80 128 48 WIRE 240 80 128 80 WIRE -176 112 -176 80 WIRE 128 112 128 80 WIRE -320 160 -320 144 WIRE -240 160 -320 160 WIRE 16 160 16 80 WIRE 64 160 16 160 WIRE -448 176 -448 112 WIRE 128 240 128 208 WIRE -176 256 -176 208 WIRE -320 336 -320 160 WIRE -208 336 -320 336 WIRE -192 336 -208 336 WIRE -80 336 -112 336 WIRE 16 336 -16 336 WIRE 112 336 96 336 WIRE 240 336 240 80 WIRE 240 336 112 336 WIRE -208 448 -208 336 WIRE -80 448 -208 448 WIRE 112 448 112 336 WIRE 112 448 -16 448 FLAG 128 240 0 FLAG -176 256 0 FLAG -448 176 0 SYMBOL voltage -448 16 R0 WINDOW 0 -37 3 Left 0 WINDOW 3 18 4 Left 0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value 1 SYMBOL npn3 -240 112 R0 SYMATTR InstName Q1 SYMATTR Value 2N3904 SYMBOL res -192 -48 R0 SYMATTR InstName R3 SYMATTR Value 30k SYMBOL npn3 64 112 R0 SYMATTR InstName Q2 SYMATTR Value 2N3904 SYMBOL res 112 -48 R0 SYMATTR InstName R1 SYMATTR Value 10k SYMBOL res 0 64 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R2 SYMATTR Value 10k SYMBOL res -336 48 R0 SYMATTR InstName R4 SYMATTR Value 1Meg SYMBOL ind -208 352 R270 WINDOW 0 32 56 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName L2 SYMATTR Value 1e-7 SYMBOL cap -16 320 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C1 SYMATTR Value 2e-12 SYMBOL res 112 320 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R5 SYMATTR Value 10 SYMBOL cap -16 432 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C2 SYMATTR Value 1e-12 TEXT -480 408 Left 0 !.tran 1e-3

-- JF

A 100nH inductor and a 2pF cap? That resonates at 356MHz, and hardly has the characteristic crystal motational inductance. You didn't even get the resonant frequency right.

At a guess I'd give the thing a motational inductance of 10.0mH, a motational capacitance of 2.815nF. That's probably wrong even so -- the characteristic impedance is still way low for a crystal, but I don't know what the right neighborhood is for a crystal at that low of a frequency.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

--
Nor did I try to, since all I did was substitute the crystal
equivalent circuit for the cap he had in there and slap some values in
there.

Bill knows how to run LTspice, and I posted some references he can use
to figure out what the right values need to be.

I wish the watch people of the world wrote detailed app notes about what they know. I suspect they have much better crystal models than anyone else and thoroughly vet them in order to build watches that run so well. Of course, there is the argument that they can depend on temperature stability, because people wear watches. But that's not entirely true. I have a watch I rarely wear, only when I go out and don't want to use a cell phone for a clock, and it keeps near flawless time. And when I did wear it rigorously, it still did so. I'd guess they knew a lot more than we can imagine about the crytals they use. I'd be interested in seeing a disclosure from such folks.

Jon

You might be right, but I would have looked to the guys at NBS and Bell Labs from the 1930s and 1940s. I thought the modern watch guys used the tuning fork shape to reduce the amount of quartz needed to resonate at low frequencies and lower the cost. The 30kcs crystal in an HC6/U is probably a rectangular prism, not a tuning fork. I remember the old books on radio design had descriptions about the stability and temperature coefficients for different crystal cuts, and how you would choose them depending on your application. Whatever the modern watch guys know probably is not all that relevant to AT-cut crystals.

At 38kHz, which is the OP's crystal, I imagined they might have something pertinent to say. However, they probably use highly custom designs, so I've no idea. Regardless, I'd love to see their models.

Jon

: h

nd

s e

I tried John's website for measuring crystal parameters and came up with a motional inductance of 2200 Henries. Seems a bit high in inductance, but I found another reference that indicated inductance could be in the thousands of Henries range. This is a fairly large crystal can that measures 1.5 inch high by 5/8 wide and maybe 1/4 thick. Seems to simulate ok, but I don't get the same duty cycle as the working unit.

-Bill