当avg打开时，PNAX SMC +阶段S22不会更新

谢谢你的文章。
我能够重现这个问题，这确实是一个非常奇怪的行为。
我们将尽快调查并解决问题。
你在运行什么版本的固件？
此外，现在一个简单的解决方法是将RevOpwr跟踪添加到您的频道。
如果通道中存在RevOpwr测量值，S22似乎将在相位和平均值的情况下更新。



     以下为原文

  Thank you for your post. I was able to recreate this issue and it is indeed a very odd behavior.  we will look into and address the problem ASAP.  what version of firmware are you running?

Also, an easy workaround right now is to add a RevOpwr trace to your channel.  it appears that S22 will update with phase and averaging enabled if there is a RevOpwr measurement in the channel.

嗨Daras，谢谢你的回复！
我目前正在使用SA Beta，但我很确定这也是10.25.02的问题



     以下为原文

  HI Daras, thank you for your response! I am currently using SA Beta but I'm pretty sure it's an issue with 10.25.02 as well

你抓住了我们
为了加速SMC平均，我们只采用S22反向扫描的平均值的第一次扫描，然后对所有其他平均值使用if，这样我们就不必等待它。
Johan于10.25.02添加了此行为。
每当重新启动平均值时它都会更新。
所以，看它更新命中平均重启。
我们的想法是S22非常稳定，不需要进行平均，但Sc21阶段确实需要很多平均值，因此它不会更新S22的速度快1/3。
真正的驱动点是，如果客户正在测量下变频器并且下变频器输出处于频段的低端，而耦合器在该频段下滑，则IF需要很慢才能获得良好的结果，但扫描需要
永远，如果S22也有平均值。
告诉我你的想法;
这里有3种选择：1）保持原样。
现在工作，工作得更快，只需按平均重启重新扫描s22。
2）每次更新它，除非选中反向耦合器复选框（实际上这是原始意图，但我认为我谈到Johan即使在正常模式下也不更新它以获得更快的速度）。
修复需要几天时间。
请注意，如果反转耦合器，则不需要低IF BW，因为滚降不在路径中。
这就是我们要利用的东西。
3）添加一个单独的控件来说明是否更新它（更复杂，谁将找到它？需要几天时间才能修复）4）为Sc21测量添加一个单独的，可选择的前向平均计数，独立
S11和S21（通道）平均值。
我喜欢这一个。
修复需要花费几天时间。
欢迎大家加入新的Keysight“定义您最喜爱的模糊SMC功能”比赛。



     以下为原文

  You caught us.  In an attempt to speed up the SMC averaging, we only take the first sweep of the average for the S22 reverse sweep, and then use if for all the other averages so that we don't have to wait for it.  Johan added this behavior in 10.25.02.  It updates whenever the average is restarted.  So, to see it update hit Average Restart.

Our thought is that the S22 is very stable, and doesn't need to be averaged, but the Sc21 phase does need lots of averages, and so it makes it 1/3 faster to not update S22.  The real driving point for this was if a customer is measuring a down converter and the down converter output is in the low end of the band, where the coupler rolls off, the IF needs to be slow to get a good result but the sweep takes forever if the S22 has averages as well.

Tell me what do you think; here's 3 choices:

1) Leave it as it is.  Works now, works faster, just hit average restart to resweep s22.
2) Update it every time, except when the reverse coupler checkbox is checked (in fact that is what the orginal intent was, but I think I talked Johan into not updating it even in normal mode for more speed).  Fix takes a few days.  Note if you reverse the coupler, you don't need a low IF BW because the rolloff isn't in the path. That's what we were going to take advantage of.
3) add a separate control to say to update it or not on ever sweep (more complicated, and who will ever find it?  Takes several days to fix)
4) Add a separate, setable, forward average count for the Sc21 measurement, independent of the S11 and S21 (channel ) averaging.  I like this one.  Takes many more than a few days to fix.

Everyone welcome to join the new Keysight "define your favorite obscure SMC feature" contest.

嗨Dr_Joel，谢谢你的见解！
Dara建议的解决方法对我们来说非常有效，因为添加跟踪它不会对我们现有的ATE数据捕获例程造成问题。
对齐技术人员已经意识到在调整s22时关闭avg或重新启动，实际上是他们向我指出跟踪没有更新。
我会给你一些想法......



     以下为原文

  Hi Dr_Joel,

Thank you for your insight! The workaround that Dara suggested will work quite well for us as adding a trace it will not create a problem with our existing ATE data capture routines. Alignment technicians have already been made aware to turn off avg or restart when tuning for s22, actually it was them that pointed out to me that the trace was not updating. I will give your options some thought....

以下是对S22问题的更深入了解。
使用RevOpwr，我们必须将源设置为端口2并以端口2频率进行测量，因此我们也可以更新S22。
但是，如果你有平均值，你可以调整或调整S22，因为平均值包含新旧数据，所以从测量完整性的角度来看，你应该随时在DUT上改变任何东西时重新开始平均（包括一个新的连接）
，调整或调整等）。
因此，严格来说，重新平均不应该是任何问题，因为无论如何你应该这样做。
没看到S22更新看起来很有趣。



     以下为原文

  Here's a little more insight into the S22 issue. With RevOpwr we must set the source to port 2 and measure at the port 2 frequency and so we might as well update S22.

But, if you have averaging on, you can be tuning or adjusting S22, as averaging contains old and new data, so from a measurement integrity point of view, you should always restart averaging anytime you change anything on the DUT (including a new connection, tuning or adjustement, etc).  Thus, strictly speaking, it should not be any issue to have to restart averaging as you should be doing that anyway.  It just looks funny to not see S22 update.

> {quote：title = Dr_joel写道：} {quote}>这里有一些关于S22问题的更多见解。
使用RevOpwr，我们必须将源设置为端口2并以端口2频率进行测量，因此我们也可以更新S22。
谢谢！
> {quote：title = Dr_joel写道：} {quote}>但是，如果你有平均值，你可以调整或调整S22，因为平均值包含新旧数据，所以从测量完整性的角度来看，你应该总是
您可以在DUT上更改任何内容时重新启动平均值（包括新连接，调整或调整等）。
因此，严格来说，重新平均不应该是任何问题，因为无论如何你应该这样做。
没看到S22更新看起来很有趣。
了解...



     以下为原文

  > {quote:title=Dr_joel wrote:}{quote}
> Here's a little more insight into the S22 issue. With RevOpwr we must set the source to port 2 and measure at the port 2 frequency and so we might as well update S22.

Thanks!

> {quote:title=Dr_joel wrote:}{quote}
> But, if you have averaging on, you can be tuning or adjusting S22, as averaging contains old and new data, so from a measurement integrity point of view, you should always restart averaging anytime you change anything on the DUT (including a new connection, tuning or adjustement, etc).  Thus, strictly speaking, it should not be any issue to have to restart averaging as you should be doing that anyway.  It just looks funny to not see S22 update.

Understood...

我们希望对S21和S11进行平均。
输入功率低，输入回波损耗低，S11走线噪声非常大。
每个平均数更新一次S22会有意义吗？
那么，如果将通道平均设置为10，那么在每10次正向扫描之后会有一次反向扫描？
无论如何，我个人并不认为这是一个大问题，但这次讨论让我想起了一些事情......现在我想到了，我似乎记得有人建议反转输入耦合器以帮助低输入
功率测量。
暂时没试过，但当时，我们使用的固件（不管它是什么）。
我们遇到了问题并放弃了，我认为这是源功率设置/校正？
如果内存服务，我们测量CW功率扫描的相位，试图获得一个完整的30dB扫描范围，比如-80dBm到-50dBm。
我们决定将源端口耦合器保持在正常位置并关闭相位跟踪校正，以便不引入校正噪声。
这是我们希望状态召回记住保存校准状态的一个原因，如前面的帖子所述。
我愿意再次尝试反向输入耦合器方法（看起来很有意义），这样我可以更好地定义我们遇到的问题。
AFAIK可以修复当前的固件最佳问候



     以下为原文

  We do like to have averaging for S21 and S11. With low input power and low input return loss the S11 trace can be very noisy. Would it make sense to just update S22 once per number of averages? So, if channel averaging is set to say 10, after every 10 forward sweeps there is a single reverse sweep? Regardless, I, personally, do not see this as huge deal but this discussion has reminded me of something...

...Now that I think of it, I seem to remember somebody suggesting to reverse input coupler to help with the low input power measurements. Haven't tried that for a while but at the time, with the firmware we were using(whatever it was). We were having issues and abandoned that, I think it was the source power setting/correction? If memory serves, we were measuring phase in CW power sweep trying to get a full 30dB sweep range, say -80dBm to -50dBm. We instead decided to leave source port coupler in normal position and turn phase trace correction off as to not introduce noise from correction. This is one reason we want state recalls to remember calibration state as saved, as mentioned in an earlier post. I will be willing to try the reversed input coupler method( it seems to make good sense) again so I might better define the problem we had with that. AFAIK it could be fixed in current firmware


Best Regards

> {quote：title = rfelectek写道：} {quote}> ...现在我想到了，我似乎记得有人建议反转输入耦合器以帮助进行低输入功率测量。
暂时没试过，但当时，我们使用的固件（不管它是什么）。
我们遇到了问题并放弃了，我认为这是源功率设置/校正？
如果内存服务，我们测量CW功率扫描的相位，试图获得一个完整的30dB扫描范围，比如-80dBm到-50dBm。
我们决定将源端口耦合器保持在正常位置并关闭相位跟踪校正，以便不引入校正噪声。
这是我们希望状态召回记住保存校准状态的一个原因，如前面的帖子所述。
我愿意再次尝试反向输入耦合器方法（看起来很有意义），这样我可以更好地定义我们遇到的问题。
AFAIK可以在当前固件中修复>>>最好的问候现在还可以做一些其他的事情：首先你可以开环，我们现在支持开环模式下大约70 dB的功率范围（向上）
从正常的约40 dB）。
对于给定的注意设置，信号源设置比注意设置低-30 dB，现在为-60 dB，因此对于30 dB衰减器设置，您可以从-90扫描到-20！。
并且，如果您正在执行电源扫描，您可以以更高的功率执行校准（设置启动=停止），然后更改启动并且校准仍然有效。
如果要提高源线性度，然后将其保存为新的源功率校准，可以使用Rx调平来获得良好的干净源校准，因此有很多新方法可以解决这个问题。



     以下为原文

  > {quote:title=rfelectek wrote:}{quote}

> ...Now that I think of it, I seem to remember somebody suggesting to reverse input coupler to help with the low input power measurements. Haven't tried that for a while but at the time, with the firmware we were using(whatever it was). We were having issues and abandoned that, I think it was the source power setting/correction? If memory serves, we were measuring phase in CW power sweep trying to get a full 30dB sweep range, say -80dBm to -50dBm. We instead decided to leave source port coupler in normal position and turn phase trace correction off as to not introduce noise from correction. This is one reason we want state recalls to remember calibration state as saved, as mentioned in an earlier post. I will be willing to try the reversed input coupler method( it seems to make good sense) again so I might better define the problem we had with that. AFAIK it could be fixed in current firmware
> 
> 
> Best Regards

There are some other things that can now be done as well: for one thing you can go to open loop and we now support about  70 dB power range in open loop mode (up from about 40 dB in normal).  For a given atten setting, the source setting was -30 dB below the atten setting, now it is -60 dB, so for a 30 dB attenuator setting you can sweep from -90 to -20!.

And, if you are doing power sweeps, you can do the calibraiton in higher power (set start = stop) and then change start and the cal is still valid. You can use Rx leveling to get a good clean source cal if you want to improve the source linearity, then save it as a new source power cal, so lots of new ways around that issue.

谢谢，所有好东西！
我会尝试这个。
我们已经在更高的功率水平进行校准，包括启动=停止功率。
再次感谢，谢谢！



     以下为原文

  Thank you, all good stuff! I will be trying this out. We do already calibrate at higher power level, including start=stop power.

Thanks again, appreciated!