Triggered XTAL Oscillator

[hock]

100mhz xtal oscillator

i am designing a circuit where i need to count the time interval between a start and a stop pulse with no missing counts. The time interval is of the duration of 100nS to few mS. i need to count the pulses with resolution 10nS. The oscillator should operate at 100MHz with 50% duty cycle. i can not use a free running oscillator as the raising edge of the clock is non deterministic and unknown with respect to the trigger pulse. The time precision is very very important and hence i need to start the oscillator immediately on arrival of a trigger pulse. The oscillator should have a good stability of the order of 10 - 50 PPM. ideally i would love to use a crystal oscillator for this application, but i am also open to other circuits. The time delay between the trigger and the oscillator pulse should ideally be zero but it is acceptable if it is constant and predictable.


thanks to all in advance for helping me.
hock

 

[FANT]

triggered oscillator

You cannot think to start an oscillator and get the singal stabilized after time 0.
The crystal oscillator takes loooong time to stabilize ( say some uSec to 1 mSec ) due the fact of the high Q of the crystal.
Better is to leave the oscillator always ON and gating the signal coming from the oscillator.
Better is to have an oscillator with double frequency ( say 200 MHz ) and reduce the latency time.

Mandi

 

[hung_wai_ming@hotmail.com]

bias a xtal oscillator

If you can buy it, then a PLL based oscillator is of the best since you can use
a lower XTAL frequency which costs much less than 100MHz as there is no fundamental 100MHz frequency easily on the market.
U can use say 4MHz XTAL and multiplies it to 100MHz with very good stability.
I don't think other solution can provide your stability requirement except XTAL
or XTAL with a PLL.

 

[hock]

PLL is of no use as the trigger signal is only a long start and stop pulse.for synchronization the trigger should be a repetitive pulse.

i know one clock pulse is not a very big issue in normal design, but in this design it is.
i can even except a startup delay of say 50nS but it should be fixed and constant.

Xtal oscillators have a long startup time due to high Q. this is mainly due to the uncharged capacitors and dry inductor current in the Xtal. i am thinking of using a bias current to bias the Xtal inductor current to its nominal operating values. i am yet to do some simulation on it.


hock