晶体振荡器启振

wwm101

你们说模型有问题
请指教啊
至于C1C2接高压还是接地
其实在小信号模型下并没有实质上的不同啊
而且实际操作中我也基本上是把这两个接地！

“这个电路的基本原理是负电阻”

如果是其它问题还请指教
谢谢！
ps：文献我已经看了一大把了

zouwanghui

图片看不到！

如果电路没有问题，那可以在晶体模型的电容或电感上加初始值

wwm101

加入初始值，是可以迅速启振
但是我碰到了是
振荡一段时间后
振荡波形迅速衰减。

lianggu

The problem is your feedback resistor. That resistance can neither be large nor be small. If it is too large, the operating point of inverter-amplifier will shift to GND. If it is too small, the resistor feedback loop will dominate. Thus, fine tune that resistor is a must.

In your case, the resistance is definitely small. You may use small W/L ratio for the feedback transistors. Good luck!

wwm101

原帖由 lianggu 于 2007-6-15 04:53 发表
The problem is your feedback resistor. That resistance can neither be large nor be small. If it is too large, the operating point of inverter-amplifier will shift to GND. If it is too small, the  ...

Thank you very much
can you tell how to confirm the min and max?

wwm101

原帖由 lianggu 于 2007-6-15 04:53 发表
The problem is your feedback resistor. That resistance can neither be large nor be small. If it is too large, the operating point of inverter-amplifier will shift to GND. If it is too small, the  ...

thank you very much
can you tell how to confirm the min and mix feedback resistor?
thank you!

lianggu

First, make sure the inverting amplifier will NOT be outside amplifying region (become latch) as resistor goes high, R-max is found. Second, make sure the oscillation has no attenuation as resistor goes low, R-min is found.
You may pick up R-min<R<R-max. For CMOS, the first step may be ignored since input impedence of inverter is close to infinity.

By the way, if you have Cadence Spectre, I suggest to use it instead of Hspice. Spectre can make crystal oscillator to start up more easy and quickly.

wwm101

原帖由 lianggu 于 2007-6-15 13:34 发表
First, make sure the inverting amplifier will NOT be outside amplifying region (become latch) as resistor goes high, R-max is found. Second, make sure the oscillation has no attenuation as resist ...

can you offer expample about how to confirm!
because i simulation a crystal osc as follow in pspice,but when i change the feedback res from 5.55meg to 5.7meg ,i find to have great effect .it can oscillate at 5.55meg ,but cann't oscillator when i change the feedback res.
that's why?
thank you

netlist
*******************
* source 100K
C_C4         N00619 N00563  1.72169576e-012  
M_M1         N00563 N00619 N01138 0 nmos  
+ L=16u  
+ W=0.8u         
+ M=10
M_M2         N00563 N00619 N01172 N01467 pmos  
+ L=16u  
+ W=0.8u         
+ M=22
V_V1         N01467 0  
+PWL 0 0 20u 3v 30u 0 40u 3v
R_R1         N01172 N01467  1000k  
R_R3         N00619 N11450  18.5k  
R_R2         0 N01138  1000k  
C_C1         0 N00619  5p  
C_C2         0 N00563  5p  
L_L1         N11450 N03952  588.873289 IC=1uA
R_Rf         N00619 N00563  {r}  
C_C3         N03952 N00563  4.30423941e-015  
.PARAM  r=5.55meg

.model pmos pmos vto=-0.70 kp=50u gamma=0.57 lambda=0.05 phi=
+0.8 mj=0.5 mjsw=0.35 cgbo=700p cgso=220p cgdo=200p cj=
+560u cjsw=350p ld=0.014u tox=14n


.model nmos nmos vto=0.70 kp=110u gamma=0.4 lambda=0.04 phi=
+0.7 mj=0.5 mjsw=0.38 cgbo=700p cgso=220p cgdo=200p cj=
+770u cjsw=380p ld=0.016u tox=14n

******************

lianggu

Why is there a 1Mohm resistor between sub and power? Why is the power stimulus applied through PSUB? Strange!

The simulation results match the analysis under different Rf condition, respectively, after removing those 2 resistor. (Refer to attachments)

Keep in mind that horn-shape waveform means correct results, just like waveform under 500M Rf condition.

lianggu

Actually, you use 100kHz crystal, right?