烽火社区
标题: 晶体与振荡器—谁更适合于无线设计的成本要求 [打印本页]
作者: mqhecheng528 时间: 2006-2-16 02:44 AM
顶
作者: thgsmq 时间: 2006-2-16 02:44 AM
楼主,谢谢你了!!1
作者: fengfeiyi 时间: 2006-2-21 11:53 AM
详细,支持
作者: kevin_wjy 时间: 2006-4-5 12:20 PM
支持
作者: Catherine 时间: 2006-5-28 11:19 AM
支持
作者: duqingzhao 时间: 2006-8-3 02:05 AM
支持
作者: com10000 时间: 2006-8-17 10:44 AM
不错,支持~~
作者: thgsmq 时间: 2006-11-26 02:40 AM
Crystals vs. Oscillators - Which Are More Cost-Effective in Wireless Designs?
by Roger Burns, Fox Electronics -- ECN, 5/1/2002
Reprints/LicensingEmail This ArticlePrint this ArticleFree Subscriptions
by Roger Burns, Fox Electronics
When designing a new wireless system, one of the first decisions to make is whether to use a crystal in conjunction with an oscillator circuit built on the IC, or a pre-packaged oscillator. While the designer may want to use the crystal to reduce costs, a closer look reveals that a pre-packaged oscillator may provide the most effective solution.
There are usually three primary factors to consider when determining which frequency source to use; they vary, depending on the end application. An extremely accurate, tight tolerance frequency source is essential when designing a wireless system. In addition, for portable devices and PCMCIA cards, size is also critical, as is cost.
This article provides a comparison of the costs and design times associated with the above options, in order to provide the designer with insight to select the best, most cost-effective solution for the system being developed.
Using Crystals in Conjunction with Oscillator Circuits — Shifting Specifications
While crystals are cost-effective, they may not be the best choice for meeting the high accuracy demands of wireless applications. For example, let's look at an application that requires a ±25 ppm frequency source. This would necessitate a crystal with a tolerance of ±10 ppm, a stability of ±10 ppm, and an aging rate of ±5 ppm. While this seems fairly straightforward, we also need to consider the crystal's frequency variation due to test equipment correlation variations and, most importantly, variances in load capacitance. The frequency of a crystal will shift with a change in load — this is commonly referred to as trim sensitivity and expressed in ppm/pF. (Note: All calculations in this article will assume the crystal's trim sensitivity is 13 ppm/pF.)
When crystals are tested on a crystal impedance meter, they are measured at the desired load capacitance. However, the meter is only accurate to 2 percent of the specified load. For example, if the desired load is 20 pF, the load the crystal actually sees from the meter can vary from 19.6 pF to 20.4 pF. If the crystal has a trim sensitivity of 13 ppm/pF, the actual reading will only have an accuracy of ±5.2 ppm. Therefore, the crystal that is rated at a tolerance of ±10 ppm could, in actuality, have a tolerance of ±15.2 ppm.
To account for this, we can change the crystal specification to ±10 ppm tolerance, ±6 ppm stability, and ± ppm aging. This allows ±6 ppm of cushion for testing variations. However, we still haven't accounted for variations in load capacitance.
Variations in the load capacitance the crystal "sees" in a circuit can be due to several contributing factors, including the load capacitors themselves. For example, the desired load for two 33 pF capacitors with a 1 percent tolerance, and stray capacitance at a constant 3.5 pF would be 20 pF. (See Figure 1) The actual load capacitance due to the tolerance of the load capacitors could vary from 19.835 pF to 20.165 pF or 20 pF ±0.165 pF, while the resulting crystal frequency could vary by ±2.15 ppm. This assumes that the stray capacitance is a constant 3.5 pf; however, the stray capacitance can vary from board to board and IC to IC.
Figure 1
If, as an example, the typical variance in stray capacitance from board to board is ±1.3 pF, from IC to IC the typical variance could be ±0.7 pF, for a total of ±2 pF. The combination of the stray variations and the load capacitor variations would result in a crystal with an actual load capacitance of 17.835 pF to 22.165 pF, or 20 pF ±2.165 pF. This means the crystal frequency could vary by ±28.15 ppm.
Variations in load capacitance
Since the target is an accuracy of ±25 ppm, the variances from the board alone would prevent the designer from being able to guarantee the desired ±25 ppm accuracy. Even if the designer uses one of the tightest crystals available — for example, one rated at ±18 ppm inclusive of tolerance, stability and first year aging, while allowing for ±2 ppm testing error — the designer would have to keep total board variation down to ±0.4 pF.
Pre-Packaged Oscillators — Tighter Tolerances, Less Design Work
When manufacturing a packaged crystal oscillator, the supplier trims the crystal frequency in the circuit. This allows the production of a tight tolerance oscillator that is devoid of the correlation issues that can arise when using a packaged crystal with the designer's own oscillator circuit. It also eliminates the need for the extensive design work required when using a discrete oscillator with a packaged crystal, while maintaining a reasonable cost.
When comparing the size of a crystal and a pre-packaged crystal oscillator, the length and width are the same, while the height generally increases by 0.3 mm to 0.4 mm, depending on which product is chosen. As an example, the F535L (see Figure 2) from Fox Electronics, which is the company's smallest oscillator, measures 5.0 mm × 3.2 mm × 1.3 mm maximum, while the equivalent FX532 crystal (see Figure 3) has a height of 1.0 mm maximum. Therefore, the only disadvantage for a pre-packaged crystal oscillator is a slight increase in height.
Fox's F535L oscillator for use with the FX532 crystal (click image to enlarge)
The FX532 crystal from Fox (click image to enlarge)
Overall Cost Differences
Several factors contribute to the overall price difference between a crystal and a pre-packaged crystal oscillator. Let's compare a crystal that was designed for 802.11A with a frequency accuracy of ±18 ppm inclusive of tolerance, stability and aging with an equivalent pre-packaged crystal oscillator with an accuracy of ±25 ppm inclusive of tolerance, stability, aging, load change, and voltage change. At first glance, the prices for this particular crystal and the oscillator in 100K quantities are virtually identical, with the crystal generally only $0.01 to $0.02 less in cost. However, the cost of the necessary capacitors, as well as manufacturing and sourcing costs required for additional components, must be added to the overall cost of the crystal. Moreover, once all variances are taken into consideration, a +25 ppm accuracy may still not be attainable with a crystal.
It may seem that conventional crystals may be the best choice for most applications because they are less expensive than oscillators. However, when all of the factors are taken into consideration, it becomes very apparent that, especially to meet the high accuracy requirements for wireless applications, oscillators are the easiest, safest, and most cost effective way to go.
作者: liugang203 时间: 2015-2-15 01:08 AM
在顶 偶以前做 2.4G 都是用晶体倍上去 呵呵
作者: wuhe_2008 时间: 2015-4-29 08:50 AM
不错！顶一下。
作者: fengfeiyi 时间: 2015-4-29 10:50 AM
不错,真的不错
作者: 小白菜 时间: 2015-4-29 12:07 PM
好
作者: mqhecheng528 时间: 2015-4-29 12:32 PM
[attach]18691[/attach]http://21control.com
作者: duqingzhao 时间: 2015-4-29 02:38 PM
顶
作者: liugang203 时间: 2015-4-29 03:43 PM
谁用过MAX2607 真他妈的难用
作者: wlmlxf521 时间: 2015-4-29 05:19 PM
楼主,不错,继续努力!加油!
作者: thgsmq 时间: 2015-4-29 06:56 PM
好呀，謝謝
作者: cwr527 时间: 2015-4-29 09:07 PM
非常好，跟你又学了不少知识。
作者: com10000 时间: 2015-4-29 10:31 PM
XO：较高频率的晶振比较昂贵，供货周期较长；
PLL：可以采用便宜的晶振缩短供货周期并简化材料单；
XO：有时需要扇出缓冲器；
PLL：无需扇出缓冲器，元件数量减少，所需板面积也减小；
XO：系统通常需要多个频率；
PLL： PLL合成器产生频率然后分频，不需要多个晶体振荡器模块。节约成本，减小所需板面积；
XO：
PLL： 可包含频谱扩展电路，EMI较小。
作者: kevin_wjy 时间: 2015-4-30 01:08 AM
好文章。1
作者: thgsmq 时间: 2015-4-30 01:41 AM
辛苦了,支持
作者: 朱宁文yangytao 时间: 2015-4-30 03:09 AM
显然，这是出自专家这手
作者: liugang203 时间: 2015-4-30 04:04 AM
一片很不错的文章。
作者: xuyu 时间: 2015-7-19 08:13 AM
标题: 晶体与振荡器—谁更适合于无线设计的成本要求
当设计新的无线系统时，首先要做的决策之一就是到底采用晶体加独立振荡器电路的方式，还是直接选用预封装的成品振荡器。尽管设计人员希望利用便宜的晶体再配合自己设计的振荡电路来降低成本，但研究表明，成品振荡器为无线系统设计提供了最经济的解决方案。
根据最终应用不同，在决定采用什么样的频率源时，通常有三个主要因素需要考虑：对于无线系统设计来说，超高精度的频率源是非常关键的；此外，对于便携式设备和PCMCIA卡来说，尺寸和成本也非常重要。
在这我们比较了成品振荡器与晶体加独立设计的振荡电路两种方案在成本和设计时间方面的差别，从而帮助设计人员在深入了解的基础上为所开发的系统选择性能最好、成本最经济的解决方案。
晶体加独立振荡器电路--技术指标的变化
尽管晶体本身成本低，但对于满足无线应用的高精确度要求来说，晶体加独立振荡器电路却并不一定是最好的选择。让我们以一个需要±25 ppm频率源的应用为例来讨论这一问题。这样的频率精度要求使用的晶体偏差要达到±10 ppm，稳定度达到±10 ppm，老化率达到±5 ppm。尽管这看起来相当简单，但我们仍然需要考虑由于测试设备相关的变化以及负载电容量所导致的晶体频率偏差。晶体频率会随着负载的变化而变化，这通常称为微调灵敏度(trim sensitivity)，并以ppm/pF来表示。(请注意：本文中的所有计算都假设晶体的微调灵敏度为13 ppm/pF。)
当利用晶体阻抗计对晶体进行测试时，一般是在与实际设计一样的负载电容条件下测试。然而，阻抗计的精度仅为特定负载的2%。例如，如果需要的负载为20pF，那么晶体从阻抗计实际“看”到的负载可能范围为19.6pF 至20.4pF。如果晶体的微调灵敏度为13ppm/pF，那么实际的读数精度将只有±5.2ppm。因此，对于测试偏差为 ±10ppm的晶体，其实际偏差可能为±15.2ppm。
考虑到这一点，我们可以将晶体的技术指标要求修改成偏差为±10pm，稳定度为±6ppm。这为测试偏差留出了±6ppm的缓冲。然而，我们仍然必须考虑到负载电容数值的变化。
电路中晶体实际的负载电容变化源于几个因素，包括负载电容器件本身的变化。例如，设计负载为两个精度为1%的33pF电容，而 20pF固定电容的寄生电容可达3.5pF。因此由于负载电容的偏差，实际的负载电容可能从19.835pF至20.165pF或20pF ±0.165pF，这样，最终的晶体频率会变化±2.15ppm。这里假设寄生电容为恒定的3.5pF，而实际上寄生电容对于不同的电路板和不同的IC都是不同的。
假设不同电路板之间的寄生电容变化为 ±1.3pF，不同IC间的变化为±0.7pF，那么总的变化为±2pF。寄生电容以及负载电容的变化总值将使晶体的实际负载电容为17.835pF 至 22.165pF，或20pF ±2.165pF。这意味着晶体的频率变化可达到±28.15ppm。
由于设计目标精度为±25ppm，单块电路板本身的变化就使设计人员很难保证±25ppm的设计精度了，即使设计人员使用精度最高的晶体，例如，使用偏差、稳定性和第一年老化率在内的总偏差为±18ppm的晶体，同时假设测试误差仅±2ppm，那么要想满足设计目标，设计人员仍然必须将电路板引起的总电容变化控制在±0.4pF。
预封装振荡器—精度更高，设计更容易
在生产预封装成品振荡器时，供应商在电路中对晶体频率进行了精细调整，这样可以生产出与相关因素无关的精度更高的振荡器，而设计人员采用成品晶体和自己设计的振荡器电路组合时就无法避免这一问题，这样也避免了采用成品晶体和分立振荡器时所需要的大量设计工作，并且可保持合理的成本。
在体积方面，晶体加分立振荡器组合和成品振荡器两种方案的长度和宽度都一样，而根据所选择的产品不同，后者的高度通常增高0.3mm到0.4mm。举例来说，Fox Electronics公司的F535L(参看图2)是该公司最小的振荡器，尺寸最大为5.0 mm × 3.2 mm × 1.3 mm，而同样长宽尺寸的FX532晶体(参看图3)高度最大仅为1.0mm。因此预封装晶体振荡器唯一的缺点是高度稍微增加了。
总体成本差别
主要有几个因素影响晶体加分立振荡器组合和预封装振荡器两种方案总成本差别。让我们比较一下针对802.11a设计的晶体，其频率总精度(包括公差、稳定度和老化率)为±18 ppm，而用于同样目的的预封装晶体振荡器总精度(包括偏差、稳定度、老化率、负载变化和电压变化)为 ±25 ppm。粗看上去，订货量为10万件时晶体和预封装晶体振荡器的价格几乎一样，晶体成本通常低0.01~0.02美元。
然而，对于晶体加独立振荡器来说，还需要考虑到所需要的电容器成本，以及额外部件的制造和外包成本，而且，如果考虑到所有的变化因素，那么利用晶体加独立振荡器的方式可能无法保证+25 ppm的精度。
传统的晶体加独立振荡器的方式看起来似乎是大多数应用的最佳选择，因为表面上看其成本比预封装晶体振荡器要低。然而，如果考虑到所有的因素，尤其对于无线应用的高精度要求来说，预封装晶体振荡器是最容易、最安全、成本最经济的解决方案。
作者: mysheep 时间: 2015-12-14 10:30 AM
不错！顶一下。
作者: WANGAIMIN 时间: 2015-12-28 09:59 AM
好文，顶一下！
| 欢迎光临 烽火社区 (http://bbs.cnecport.com/) |
Powered by Discuz! X3.2 |