[SOLVED] [help!]How to avoid 32.768K xtal oscillator working in 194K frequency

[frederick.im]

I designed a 32.768K xtal oscillator
i didn't add RD because RD seems it is unnecessary

but there was one sample, in more than 50 samples, working in 194K frequency
but when i put probe to its XIN/XOUT pin, working frequency changed from 194K to 32.768K (we can actual see the transformer)

and when i press the shell of xtal, the osc freq change to 32K too

is there any guy know the reason of this condition? the osc circuit or the xtal?

how can i suppress harm freq osc beside RD

 

[matthai]

i think the crystal working in sixth harmonics.

if u r using this in a digital way,
ground the body of crystal and try.
otherwise add 22pf on both sides to ground.

matthai

 

[jiripolivka]

What test instrument did you use to detect the 6th harmonic? If you used a digital counter, such counters have input pulse shapers that easily generate such harmonic. If you use a spectrum analyzer, take care not to ovedrdrive the input as also there harmonics are generated.

Use a minimum voltage to operate the oscillator, and use the above mentioned 22 pF capacitors to prevent overtone oscillations. THe output may be a distorted sine wave which does contain hArmonics.

 

[frederick.im]

thx for ur reply
u r right
when i ground the body of xtal
the freq back to 32.768K
but what can i do when i used a smd xtal

why the crysital working in 6th harmonics?
is it because of the overdrive cause unstabilization @ 6th harmonics

- - - Updated - - -

thx for ur reply

i use a normal oscillography
the probe is 3.9pF 10Mohms

i think u r right
this is caused by the overdrive

 

[neddie]

Let's see the schematic. Have you got load capacitors on either side of the xtal to ground?

 

[FvM]

Presumed you want a certain clocking accuracy from the watch crystal, additional parallel capacitors are hardly an option. Load capacitors are usually provided inside the oscillator circuit, in some cases they are prescripted by the data sheet, e.g. with some microcontrollers.

Without a schematic or detail circuit description, we can't decide between two likely causes:
- defective crystal production lot
- unsuitable oscillators circuit

 

[RCinFLA]

Need to see the circuit. You are probably driving the crystal too hard. 32 kHz crystal can break due to overdriving. Most are designed for 10 pf load but can be 6 pF to 32 pF. For 10 pF load, two 18 pF's on either side of crystal, along with PCB stray and IC input capacitance comes close to 10 pF across crystal. If driving from CMOS rail to rail logic output, there should be a series resistor from output to crystal/cap node to drop the drive level down to a safe level.

A 32 kHz crystal oscillator can take 2-3 seconds to reach steady state run condition.