Help: 32K oscillator startup problem

Hi, all,
It's a 32k crystal oscillator with open drain output.

1> when 10K pullup resistor was added, if pullup VDD>3v, it cannot startup, the vpp of x1,x2 are almost 0v, output a random wave.
2> when pullup VDD<3v, it can startup.
3> when removed pullup resistor R1, osc can startup immediately. then, I add R1 back, osc still on.
4> when clock output was set to other frequency, such as 16K,8K, the osc can startup.

this oscillator was used in my products many times, gain, netive-res has no problem.
my question is,
why the opendrain output relate to oscillator startup condition?
if big switch noise of 80/0.7 nch can prevent osc startup, how it works? any threory can help me?

thanks a lot!

Any comments? any similar experience?

Try a hundreds of kOhm resistor between the amplifier output and the rest of the circuit.

The amplifier has to be in linear region for starting oscillation. Probably you are using an inverter gate. In that case, a resistor between X1 and X2 is needed (10K to 1Mohm if it is a CMOS).
When you open the drain, the buffer sees a different load, and the amplifier as well. Probably this perturbation is responsible of start up.

Regards

Z

you simply need a 10uh choke and a decoupling cap on the vdd rail to the buffer and amp stages replace the 10k resistor with a paxilon peg and wind 10 T .28mm dia ecw 1.2mm dia the rf is using the resistor as a coil and this has the effect of a dampener on the system as the rf leek away haphazardly
finaly clamp it with a zener at +5.1v before the choke and cap

a ferrite beed isnt good enought for high output p.p volts

it depends what type of resistor you used here
a 10k resistor isn't just that

i think better to place a 1MR to ground from the gate
place a drain resistor 22R

i think this design is made for analogue equipment
but again why the fet
however the cap and coil will make it start every time
instead also place a decoupling capacitor ~10nf before the choke
ahhh i see 32KHZ LOL more like 150uH choke you need
55T .38mm dia ecw on a 3 mm former
or youll get poor voltas from it
this is a timebase i feel and its a poor design by far
the fet is having trouble and draws too much input volts becouse it has no negative bias
what you really need to do most
is couple a spectrum analyzer to the output of the buffer and see what peeks your getting
my feeling is removal of the load resistor increases buffer self oscilate

youll get many harmonics at strange frequencies
as you have no choke and decoupling at all
at such low frequencies this is really important

jasmin_123 said:

Try a hundreds of kOhm resistor between the amplifier output and the rest of the circuit.

Click to expand...

Hi, jasmin,
acctually there is a 60Kohm res at the output of amplifier, I didn't draw it out.
do you means the phase delay was too small? but why it response with open drain out?

Added after 1 minutes:

zorro said:

The amplifier has to be in linear region for starting oscillation. Probably you are using an inverter gate. In that case, a resistor between X1 and X2 is needed (10K to 1Mohm if it is a CMOS).
When you open the drain, the buffer sees a different load, and the amplifier as well. Probably this perturbation is responsible of start up.

Regards

Z

Click to expand...

Hi,zorro,

Yes it's a linear amplifier.

Added after 7 minutes:

hi,VSMVDD,

Thanks for your comments. but i'm not sure about your means.....
actually between VDD&GND, I added 1U and 0.1U decouple capacitors.
do you means, add a 10uH inductor in series on VDD?

yes

you need a choke on the voltage supply to the fet final buffer amp
or the rf from the crystal will leak down this path
before the choke {voltage input end} should have a decoupling cap capacitor
if you want to keep the rf in the rf path
and not in your psu then add these parts

Could you post your details of circuit and components used? then I can try your problem. It sounds very strange, since theoretically once you use a buffer between the oscillator and the output stage, nothing from output can effect the osc.

nguyennam

ok, here is the detail img

The 60k resistor seems redundant.

Isn't the 60K to prevent the crystal from being overdriven?

Can the 0.5-pA current source overdrive the crystal?

jasmin_123 said:

Can the 0.5-pA current source overdrive the crystal?

Click to expand...

it's 0.5uA. but i dont' think 60K res can effect the start up condition.

give more infor mation