你们设计的OSC，精确吗？

behzad

   

OSC结构对精度影响也不小

有本书对结构的演进写得很详细
zhanweisu27 发表于 2009-11-9 20:30

                               
登录/注册后可看大图

能说一下书名么？

ysb0204

做trim fuse 也許可以解決

zhanweisu27

Bipolar and MOS Analog Integrated Circuit Design

第11章

zhang_huiquan

   

赞同，我们也曾经这么做过。不过不知道你有没有遇到这样一个问题，如果拿R段的电压作为参考，那么比较器的 ...
behzad 发表于 2009-11-10 11:34

                               
登录/注册后可看大图

是这样的

jinjin0412

开关电容，如此一来电路中需要两个clock，一个很快的做开关电容，不需要精确；另一个就是你要的clock，这太不现实了。。。

fhchen2002

1. Build an I_CTAT (current source, Constant to Absolute Temperature) from Bandgap Vref.  The bias current of the CTAT current source is subject to resistor's Rsh (sheet resistance) change over process corner.

2. Build an oscillator, use this I_CTAT to charge or discharge the capacitor associated with the oscillator.
V = I_CTAT * T / C
f = 1 / T = I_CTAT / (C * V)

V has to be derived from Bandgap Vref.
For instance, Bandgap Vref = 1.210 V.
V = 90/121 * Vref = 0.9 V
This V is almost constant over temperature, process and power supply.

3. Use switches to decode (i.e. select) the needed capacitance to get the desired oscillation frequency.   On-chip MiM or PiP capacitor is subject to +/-10% capacitance variation over process corner, but will have excellent (i.e., extremely small) temperature and voltage coefficient.

Once the desired capacitance (based on frequency measurement) is determined, the oscillation frequency will have low temperature, process and power supply drift.

Step 3 is called "trimming."

This design approach can achieve < +/-2% frequency variation over [0 degrees C, 85 degrees] temperature range.

fightshan

   

赞同，我们也曾经这么做过。不过不知道你有没有遇到这样一个问题，如果拿R段的电压作为参考，那么比较器的 ...
behzad 发表于 2009-11-10 11:34

                               
登录/注册后可看大图

   请问你们做的最后会有trimming吗？RC受corner的影响极为严重，温度和电源电压反而有有个办法可以搞定，关键就是RC的绝对值会跟corner巨变，如果作为PLL的参考时钟，这种变化能受得了吗？