[求助]菜鸟问一下sensitivity问题

jash · 发表于 2005-9-26 14:33:00

手上直板机连线测试512-700显示-107 ~-105趋势，然后700-800就在-104附近，后面又慢慢变回-107左右，中间一段一直这样，耦合测试中间一段更菜-100左右，都是逐渐变坏的，各位有什么建议麻烦提出来大家一起研究

tongange · 发表于 2005-9-26 16:53:00

手机受到干扰，插损（s21）也需要调好一点。先调有线的[br]+3 RD币

ciscospeed · 发表于 2005-9-26 17:21:00

The most of the problem are definite from mathcing,pls check your receiver path.Else even the highest value you got is too bad(-107dBm),nowadays it usually reach at least -109dBm-110dBm with good PCB artwork.[br]+5 RD币

cubain · 发表于 2005-9-27 09:20:00

-109dBm-110dBm 做到是不难但也不能说-107dBm too bad吧[br]+3 RD币

jash · 发表于 2005-9-28 11:37:00

自己顶一下，有人么？匹配我换了，可是改变不大，还有什么建议么？

ciscospeed · 发表于 2005-9-28 13:46:00

There must be some noise coupled to your RF part, check RF trace firstly, If still have problem then I will give you more suggestions about BB incl.co-processor staff part Critical Trace ü The PA output à Antenna will be high power transmit, to minimize the return loss the PA output trace should be strictly Micro stripline; as straight as possible; as short as possibleü The PA input trace is very sensitive and it is better to be separated by Groundü Antenna à LNA input should be minimized the noise and trace route loss. It is better to be Micro stripline and as short as possibleü The loop filter of LOVCO & TXVCO is taken the LDO output as reference level, so it is better to separate these circuits with Ground by LDO output copperü The Transceiver 26MHz ref clock output may affect the LOVCO phase noise, so it is better to separate with loop filters of LOVCOü The I/Q signal trace should be differential (parallel and symmetry) routing, and better to be separated by Groundü The TX_RAMP signal is very sensitive. It should be separated by Ground and better to make the layer bellow the TX_RAMP signal a completed GND area.ü The Audio input & output trace should be separated by Ground and it is better to trace differential.ü ABB will measure the voltage difference of the high accurate Resistor , so the two traces of R2703 à ABB should be same length and same width.ü Components of Microphone input circuit should be put as symmetry as possible.ü SIM test is very sensitive. While PCB layout, the SIM trace, such as V_SIM, SIM_DATA, SIM_RST, SIM_CLK should be kept clean. And reserve a R-C filter location at the side of SIM socket for above lines respectively. [br]+5 RD币

yellowtoot · 发表于 2005-9-28 20:16:00

<DIV class=quote>以下是引用ciscospeed在2005-9-28 13:46:44的发言：

There must be some noise coupled to your RF part, check RF trace firstly, If still have problem then I will give you more suggestions about BB incl.co-processor staff part

Critical Trace






ü The PA output à Antenna will be high power transmit, to minimize the return loss the PA output trace should be strictly Micro stripline; as straight as possible; as short as possible


ü The PA input trace is very sensitive and it is better to be separated by Ground


ü Antenna à LNA input should be minimized the noise and trace route loss. It is better to be Micro stripline and as short as possible


ü The loop filter of LOVCO & TXVCO is taken the LDO output as reference level, so it is better to separate these circuits with Ground by LDO output copper


ü The Transceiver 26MHz ref clock output may affect the LOVCO phase noise, so it is better to separate with loop filters of LOVCO


ü The I/Q signal trace should be differential (parallel and symmetry) routing, and better to be separated by Ground


ü The TX_RAMP signal is very sensitive. It should be separated by Ground and better to make the layer bellow the TX_RAMP signal a completed GND area.


ü The Audio input & output trace should be separated by Ground and it is better to trace differential.


ü ABB will measure the voltage difference of the high accurate Resistor , so the two traces of R2703 à ABB should be same length and same width.


ü Components of Microphone input circuit should be put as symmetry as possible.


ü SIM test is very sensitive. While PCB layout, the SIM trace, such as V_SIM, SIM_DATA, SIM_RST, SIM_CLK should be kept clean. And reserve a R-C filter location at the side of SIM socket for above lines respectively.




</DIV>


挺全的,但觉得不是都与接收灵敏度有关啊
如:The TX_RAMP ,The PA input trace ,The PA output à Antenna

是从哪抄来的吧

nangyang · 发表于 2005-10-6 18:12:00

你提供的信息太少，你先看看你射频器件下边的那一层有没有问题！

jash · 发表于 2005-10-7 13:33:00

感谢各位大爷啊问题已经解决了更换了saw to tranceiver之间的匹配，大家参考一下顺便问一下，那些值都是试出来的么？有没有什么理论依据，可否告知[br]+3 RD币

yellowtoot · 发表于 2005-10-7 21:25:00

好好玩呀[em02]前面说换过匹配,不行;现又说换了匹配,好啦[em06]偶不能理解的就是:如果是匹配不好,那为何在1800MHz上,两头好(500-700,800-880),中间差(700-800)呀[em13]你改完匹配解决后,现在的灵敏度都是多少呀?[br]+3 RD币

jash · 发表于 2005-10-11 15:53:00

开始换匹配不对，所以灵敏度不行修改过后-107[br]+3 RD币

yellowtoot · 发表于 2005-10-12 09:56:00

<DIV class=quote>以下是引用jash在2005-10-11 15:53:07的发言：

开始换匹配不对，所以灵敏度不行
修改过后-107</DIV>

请问你换匹配, 是一点点试, 还是通过计算的呀?
如果是计算 ,请问如何计算的呀?

另外,我还是不明白,"如果是匹配不好,那为何在1800MHz上,两头好(500-700,800-880),中间差(700-800)呀"
请问你能回答吗?[br]+3 RD币

jash · 发表于 2005-10-12 10:23:00

是试出来的我本人也感到纳闷的，按照道理说匹配部队，应该整个频段都有问题啊，或者偏移出去，为什么唯独中间不好呢，还是高人来回答吧！[br]+3 RD币

meijm2000 · 发表于 2005-10-12 11:50:00

这里是个高阻抗区域，在不同频段的响应的差别大很正常啊[br]+3 RD币

yellowtoot · 发表于 2005-10-12 16:41:00

<DIV class=quote>以下是引用meijm2000在2005-10-12 11:50:12的发言：
这里是个高阻抗区域，在不同频段的响应的差别大很正常啊</DIV>

我觉得,通常的匹配都是对中心频率进行的, 两边的频率上可能匹配不好.
但楼主的情况洽洽相反.
难道这是正常的吗??
请问能说说"高阻抗区域"的概念吗?
谢谢啦
[em05][br]+3 RD币

yellowtoot · 发表于 2005-10-12 16:43:00

<DIV class=quote>以下是引用jash在2005-10-12 10:23:24的发言：

是试出来的
我本人也感到纳闷的，按照道理说匹配部队，应该整个频段都有问题啊，或者偏移出去，为什么唯独
中间不好呢，还是高人来回答吧！</DIV>

有traceiver的输入阻抗和FEM的输出阻抗吗? (一般Datasheet上有的)
如果有, 请将几次匹配的电路形式及元件值告诉大家, 一起算算看吧.
相信,人多力量大啊.
自已一个人不一定能想明白啊.[br]+3 RD币

r-snake · 发表于 2005-10-13 14:26:00

关于Rx Receive中FEM－Tranceiver的高阻抗（150～200 Ohm）匹配，个人感觉有一个办法可以验证：1、清楚或取消掉手机校准数据中，关于Rx Band－Ripple的补偿数据；2、通过传导测试，测量一个频带内各个信道的Sensitive与此时Rx Level之间的差；3、阅读上面测出来的数据，“感觉”出这个Ripple来；PS：如果你明白Rx Level、Rx Band－Ripple的意义，那你就懂的这一项测试的意义了。[br]+5 RD币

yellowtoot · 发表于 2005-10-14 12:33:00

<DIV class=quote>以下是引用r-snake在2005-10-13 14:26:15的发言：

关于Rx Receive中FEM－Tranceiver的高阻抗（150～200 Ohm）匹配，个人感觉有一个办法可以验证：
1、清楚或取消掉手机校准数据中，关于Rx Band－Ripple的补偿数据；
2、通过传导测试，测量一个频带内各个信道的Sensitive与此时Rx Level之间的差；
3、阅读上面测出来的数据，“感觉”出这个Ripple来；
PS：如果你明白Rx Level、Rx Band－Ripple的意义，那你就懂的这一项测试的意义了。

+5 RD币</DIV>

不好意思啊, 你说得太高深啦! 没明白你在说什么啊.
你的意思是不是说, 楼主的问题是由这个ripple造成的呀,即transceiver本身的纹波, 而与匹配无关?
我不明白的有两点:
1,"高阻区"是指你所说的100欧吗?
2,匹配时,应以中心频率进行计算,然而为何两头匹配好,而中间差呀?
谢谢啊!
[em05][br]+3 RD币

meijm2000 · 发表于 2005-10-14 13:52:00

<DIV class=quote>以下是引用yellowtoot在2005-10-14 12:33:02的发言：

不好意思啊, 你说得太高深啦! 没明白你在说什么啊.
你的意思是不是说, 楼主的问题是由这个ripple造成的呀,即transceiver本身的纹波, 而与匹配无关?
我不明白的有两点:
1,"高阻区"是指你所说的100欧吗?
2,匹配时,应以中心频率进行计算,然而为何两头匹配好,而中间差呀?
谢谢啊!
[em05]</DIV>

看看FEM的datasheet里面都有的阻抗是150-200的高阻抗。调试的时候能打到最大效率传输就好了，而且这些不是简单的计算就可以了。ripple感觉与校准的方式有关啊，如果每个信道就能单独校准一下，也就不存在了吧[br]+5 RD币

yellowtoot · 发表于 2005-10-14 21:24:00

<DIV class=quote>以下是引用meijm2000在2005-10-14 13:52:31的发言：

看看FEM的datasheet里面都有的阻抗是150-200的高阻抗。调试的时候能打到最大效率传输就好了，而且这些不是简单的计算就可以了。ripple感觉与校准的方式有关啊，如果每个信道就能单独校准一下，也就不存在了吧</DIV>

FEM的150是差分的, 转换成单端也就75, 与50很接近呀, 为何称为高阻啊???
另外,衡量最大传输效率的参数是什么呀? 如何测量啊? 调试时,是根据什么参数调试呀?
估计还是归结为驻波吧.
调试时,也只能根据接收灵敏度进行调试吧.
而楼主的结果是中间差,两头好, 遇到这种情况,还如何调呀?[br]+5 RD币

		自动登录	找回密码
密码			注册

[讨论] [求助]菜鸟问一下sensitivity问题