在保持线性的同时怎么设置最高的源端口功率

第108页的情节是一个良好的开端。
我们的分析仪将预设的光源功率设置为-5 dBm（Ecal最精确）的较低水平，该水平恒定，比最大光源功率低约5 dB。
因此，较高频率的盒子在高端具有较低的最大功率，因此预设功率较低。
在几乎所有的盒子中，测试端口功率为0 dBm且向下的动态精度几乎为零（在0 dBm时约为0.01 dB）。
该规范没有表明它的唯一原因是测量非常困难，因为没有其他工具像PNA一样准确的动态精度。
最佳情况：设置功率，使Ecal的功率介于-5和0 dBm之间（0 dBm时的Ecal压缩约为0.01dB，因此如果您的跟踪噪声超过0.01 dB，最好设置功率
更高，接受一点Ecal压缩以弥补更低的迹线噪声.PNA的一个简单经验是使用1kHz IFBW进行高质量测量。我们自动进入更稳定的模式，IFBW为1kHz及更低（更稳定）
意味着我们在设置Ecal状态之间等待更长时间，我们进入逐步扫描模式，并确保ALC更稳定。）如果需要，校准，然后根据需要将功率设置更改为不同的值。更改功率的动态误差几乎是
总是小于在非最佳功率水平校准时增加的误差。



     以下为原文

  The plot on page 108 is a good start. 

Our analyzers set the preset source power to the lower of -5 dBm (where the Ecal is most accurate) a level that is constant and about 5 dB lower than max source power. Thus, higher frequency boxes have lower max power at the high end, so the preset power is lower.

In almost all our boxes, the dynamic accuracy from a test port power of 0 dBm and down is nearly zero (it is about 0.01 dB at 0 dBm).  The only reason the spec doesn't show it is that it is terribly difficult to measure since there are no other tools as accurate in dynamic accuracy as the PNA.

Best case: set the power so that the power to the Ecal is between -5 and 0 dBm (the Ecal compression at 0 dBm is about 0.01dB, so if your trace noise is more than 0.01 dB, you are better off setting the power higher, accepting a little Ecal compression to make up for much lower trace noise.

An easy rule of thumb on PNA is use 1kHz IFBW for high-quality measurements.  We automatically go into a more stable mode for IFBW of 1kHz and lower (more stable means we wait longer between setting Ecal states, we go into a stepped sweep mode, and make sure ALC is more settled).

Calibrate, then change power settings as needed to different values if you need to. The dynamic error from changing power is almost always less than the added error from calibrating at a non-optimal power level.

感谢Joel博士的回复。
我仍然想知道我正在使用的VNA的规格中的-15dBm“标称功率”。
对我来说，我认为将源端口功率设置在0dBm到-5dBm左右的经验法则与ECAL单元相矛盾。
* _平原和简单：如果我要将源端口功率设置为0dBm到-5dBm以获得最大ECAL精度，那么我应该担心这个VNA吗？_ *此外;
从阅读你的书中，我已经通过强制步进模式消除了IF延迟的伪影，停留时间为1微秒（我们的大多数duts最大约为6米，低Dk因此时间延迟远低于此停留时间）。
有时我会将扫描延迟增加到1microsec以获得乐趣（虽然我真的不知道这有什么帮助或它真正有用的）。
幸运的是，我很容易，并测量有损的duts;）。
对于纯粹的幅度s-param扫描，我甚至可以使用非常高的IF（> 100kHz）并且仍然获得低的跟踪噪声并且满意;）。
但我对你上面提到的ALC感到好奇。
你能详细说明吗？
就像我说的，IF延迟问题已经解决，但我对ALC很好奇。
谢谢，KTBEdited：KTB于2015年7月13日上午10:36



     以下为原文

  Thank you for the reply, Dr. Joel.

I'm still wondering about the -15dBm "nominal power" in the spec for the VNA I'm using. To me, I feel it contradicts the rule of thumb of setting source port power around 0dBm to -5dBm to the ECAL unit. *_Plain and simple: Is there something I should be worried about for this VNA if I am to set the source port power to say 0dBm to -5dBm for max ECAL accuracy?_*

In addition; from reading your book, I have eliminated artifacts of IF Delay by forcing stepped mode with a dwell time of 1 microsec (most of our duts are about 6 meter max with low Dk so Time-Delay is much lower than this dwell time). Sometimes I increase the sweep delay to 1microsec for fun (though I don't really know how this is helping or what its truly useful for). Luckily I have it easy and measure lossy duts ;). For just pure magnitude s-param scans, I can even use really high IF (>100kHz) and still get low trace noise and am satisfied ;). But I'm curious about the ALC you mentioned above. Could you elaborate on that? Like I said, the IF Delay issue has been resolved but I'm curious about ALC.

Thanks,
KTB

Edited by: KTB on Jul 13, 2015 10:36 AM

> {quote：title = KTB写道：} {quote}>感谢Joel博士的回复。
>>我仍然想知道我正在使用的VNA的规格中的-15dBm“标称功率”。
对我来说，我认为将源端口功率设置在0dBm到-5dBm左右的经验法则与ECAL单元相矛盾。
* _平原和简单：如果我要将源端口功率设置为0dBm到-5dBm以获得最大ECAL准确度，我应该为此VNA担心吗？_ *>别担心，开心。
如果VNA没有显示“source unleveled”，则没有问题。
如果您将源功率设置为0 dBm并且源不是不平坦的，那么您没问题。
此外，对于最佳实践，您应该针对有源和无源器件进行不同的校准。
对于有损DUT，您可以在0或-5 dBm下校准，然后在校准后增加功率。
如果你有一个“直通”（如f-f），你可以观察走线噪声和压缩（通过S21的变化）并增加源功率，直到压缩轨迹的变化等于走线噪声。
这是通过的最佳源功率。
对于有损耗的DUT，你可以走得更高。
通常，您不需要担心R通道压缩，因为我们将接收器相对于其他接收器降低约10 dB。
>另外;
从阅读你的书中，我已经通过强制步进模式消除了IF延迟的伪影，停留时间为1微秒（我们的大多数duts最大约为6米，低Dk因此时间延迟远低于此停留时间）。
有时我会将扫描延迟增加到1microsec以获得乐趣（虽然我真的不知道这有什么帮助或它真正有用的）。
幸运的是，我很容易，并测量有损的duts;）。
对于纯粹的幅度s-param扫描，我甚至可以使用非常高的IF（> 100kHz）并且仍然获得低的跟踪噪声并且满意;）。
但我对你上面提到的ALC感到好奇。
你能详细说明吗？
就像我说的，IF延迟问题已经解决，但我对ALC很好奇。
>>谢谢，> KTB> ALC通常只在您测量有源设备时很重要。
如果您尝试更改ALC范围内的功率，则信号源将无法使用并显示“source unleveled”消息。
所以只需稍微退一步，直到消息消失（清扫后清除消息需要2秒钟）。
>编辑：KTB于2015年7月13日上午10:36



     以下为原文

  > {quote:title=KTB wrote:}{quote}
> Thank you for the reply, Dr. Joel.
> 
> I'm still wondering about the -15dBm "nominal power" in the spec for the VNA I'm using. To me, I feel it contradicts the rule of thumb of setting source port power around 0dBm to -5dBm to the ECAL unit. *_Plain and simple: Is there something I should be worried about for this VNA if I am to set the source port power to say 0dBm to -5dBm for max ECAL accuracy?_*
> 
Don't worry, be happy.  If the VNA does not display "source unleveled" then there is no issue.   If you set the source power to 0 dBm and the source is not unleveled, then you are fine.  Also, for the very best practice, you should calibrate differently for active versus passive devices. For lossy DUTs, you can calibrate at a 0 or -5 dBm, then increase the power after calibration.  If  you have a "thru" (like f-f), you can watch the trace noise and the compression (the change in S21 of the thru) and increase the source power until the change in the trace from compression is equal to the trace noise.  This is the optimum source power for a thru. For a DUT that is lossy, you can go even higher.  Normally you don't need to worry about R channel compression as we pad that receiver down about 10 dB relative to the other receivers.

> In addition; from reading your book, I have eliminated artifacts of IF Delay by forcing stepped mode with a dwell time of 1 microsec (most of our duts are about 6 meter max with low Dk so Time-Delay is much lower than this dwell time). Sometimes I increase the sweep delay to 1microsec for fun (though I don't really know how this is helping or what its truly useful for). Luckily I have it easy and measure lossy duts ;). For just pure magnitude s-param scans, I can even use really high IF (>100kHz) and still get low trace noise and am satisfied ;). But I'm curious about the ALC you mentioned above. Could you elaborate on that? Like I said, the IF Delay issue has been resolved but I'm curious about ALC.
> 
> Thanks,
> KTB
> 
The ALC is normally only important when you are measuring active devices.  If you try to change the power out of the ALC range, the source will go unleveled and show the "source unleveled" message. So just back off a little until the message goes away (it takes 2 seconds to clear the message after a clean sweep).
> Edited by: KTB on Jul 13, 2015 10:36 AM

磕碰！
:)



     以下为原文

  Bump! :)

乔尔在下周的某个时候无法使用。



     以下为原文

  Joel is unavailable until some time next week.

Joel博士，非常感谢您提供详细而有用的回复。
我将在未来几年随身携带这些信息。
KTB



     以下为原文

  Dr. Joel,

Thank you VERY much for your detailed and useful responses. I shall carry this information with me for years to come.

KTB