当输入引脚短接到地时，PIC18F CTMU+ADC读取满量程？

更多的数据：* ADC测量并不总是满量程，但它们在几个LSB范围内。通常大于4090小数。*如果您等待一段时间-大约60秒-读数下降到接近零为一个测量。然后它弹回到满刻度。如果你在60秒之内重复测量，读数保持很高。但是，如果你等待“足够长”（这似乎是超过60秒），读数就会下降到接近零，正如你所期望的。这种“感觉”就像一个内部电容，在CTMU经历它的循环时，得到并保持电荷。等待足够长，电容放电，你得到一个有效的阅读，但此后（或如果你测量得太快）电容保持足够的电荷，给你一个完整的阅读。这个电荷（和它的电压）是不可见的针。我已经尝试了模拟和数字范围，都显示了绝对的死点在输入引脚使用。所以A/D是测量设备内部的东西。不知何故，内部的“东西”对通过引脚绑在地上很敏感。最令人不安的是它的不一致性。取决于你等待的时间，你得到（几乎）零或（几乎）满刻度。我期待一个短暂的僵局能够产生一致的结果。



      以下为原文

    More data:
 
* The ADC measurements aren't always full scale, but they're within a few LSB's. Generally above 4090 decimal.
 
* If you wait a while - around 60 seconds - the reading drops to nearly zero for one measurement. Then it pops back up to full scale. If you take repeated measurements within ~60 seconds of each other, the reading stays high. But if you wait "long enough" (which seems to be something over 60 seconds), the reading drops to near zero as you'd expect.
 
This "feels" like an internal capacitance is getting and retaining a charge each time the CTMU goes through its cycle. Wait long enough and the capacitance discharges and you get a valid reading, but thereafter (or if you measure too soon) the capacitance retains enough charge to give you a full scale reading.
 
This charge (and its voltage) are NOT visible on the pin. I've tried both analog and digital scopes; both show absolute dead ground on the input pin being used. So the A/D is measuring something INTERNAL to the device. And somehow, that internal "thing" is sensitive to being tied to ground via the pin.
 
What's most troubling about it is its inconsistency. Depending on how long you wait, you get either (almost) zero or (almost) full scale. I would expect a dead short to ground to yield a consistent result.

我要告诉你类似的事情：我只和CTMU一起工作过，但是我确实仔细研究了代码和行为。这是一个“有趣”的外围设备！IMO，接地引脚不需要产生0，明确：读出将取决于时间（严格）和一个VREF…



      以下为原文

    I was going to tell you something similar to the above:
I only worked with CTMU once, for a keyboard: but I did dig into the code and behaviour quite a lot and... it's a "funny" peripheral!
 
IMO, grounding a pin DOES NOT need to yield 0, definitely: readout will depend upon timing (Strict) and one VRef...

那是肯定的。我已经花了很多时间和它在一起，感觉我了解大部分正在发生的事情，但这一问题继续唠叨我。有趣的是，微芯片从来没有回应过关于CTMU的问题。我有两张关于主题的公开门票，现在开放了几个星期，而且从来没有对内容作出任何回应。我们把它分配给别人“我是唯一得到的。当然，但是如果你短路一个电容器，你会认为它不能获得和存储电荷，是吗？然而，将输入引脚短路到CTMU不仅没有做到这一点，而且它也在非常低和（更奇怪）的非常高的读数之间随机交替。那是我不明白的部分…并不是说它不是精确的零，而是如果输入引脚被短路，它的大部分时间几乎是满刻度的。？！？！？



      以下为原文

   
That's for sure. I've spent a lot of time with it now and feel like I understand most of what's going on, but this one issue continues to nag me.
 
It's interesting that Microchip never, EVER responds to questions about the CTMU. I have two open tickets on the topic, open for weeks now, and never a single response with content. "We're assigning it to someone" is all I ever get.
 

Certainly, but if you short a capacitor you'd expect it to not be able to acquire and store a charge, yes? Yet shorting the input pin to the CTMU not only doesn't do that - it also psuedo-randomly alternates between very low and (more weird) very high readings. THAT's the part I don't understand... not that it isn't precisely zero, but that it's nearly full scale the majority of the time if the input pin is shorted. ?!?!?

当然，但是如果你短路一个电容器，你会认为它不能获得和存储电荷，是吗？嗯，是的，我认为接地并不是如预期的那样，因为……之间的“其他东西”……我听起来也很疯狂，但后来我接受了……是的，确实，外围设备（CTMU）就像一个没有太多解释的魔法盒子……（在一次会议上，大约5到6年前，他们坚持你可以用它来衡量一切！）



      以下为原文

   
Certainly, but if you short a capacitor you'd expect it to not be able to acquire and store a charge, yes?




Well yes
I suppose the grounding is not acting as expected, due to "something else" in between...
It sounded crazy to me too, but then I accepted it...
 
And yes, indeed that peripheral (CTMU) is like a magic box with not much explanations... (at one meeting, some 5-6 years ago, they insisted on you can measure everything with it ;) ! )

因为关于CTMU内部电路的数据表中的细节很少，这是猜测：可能电流源和它与A/D的内部连接对输入引脚上的“短”的近零电阻没有很好的反应。如果你通过一个小电阻将引脚接地，比如说100欧姆或1K会发生什么？



      以下为原文

    Because there is very little detail in the data sheets about the CTMU internal circuits this is a guess: possibly the current source and it's internal connection to the A/D does not react well to the near zero resistance of a "Short"on the input pin. What occurs if you connect the pin to ground through a small resistor, say 100 ohms or 1K?

在PIC32系列的CTMU参考章节中，我发现：可能是2.5K串联电阻在输入接地时产生高A/D读数，因为CTMU电流必须通过这个电阻才能到达引脚的“短”。



      以下为原文

    Looking at the CTMU reference chapter for the pic32 series I find this:

Possibly it is this 2.5K series resistance that is producing the high A/D reading when the input grounded, since the CTMU current must pass through this resistance to get to the "short" at the pin.

精彩！我已经忘记了2.5K.在5.5UA充电电流，2.5K将只发展到14MV横跨它。然而，其更大的效果是浮动（否则接地）内部寄生电容远离真正的接地。这将允许它对它的另一个板块进行充电，这一板块在内部是接地的，仍然不能解释它的读数的非确定性性质。你会认为2.5K会让它变高，或者变低，但不会在这样的轨道之间波动。哦，好的，谢谢你的回复！也许我们所有人在一起，可以记住这个CTMU模块的所有细节。很明显，微芯片不打算支持它。再次感谢，非常非常！



      以下为原文

   
Brilliant! I had forgotten about that 2.5K.
 
At a 5.5uA charging current, that 2.5K will only develop ~14mV across it. However, its much larger effect is to float the (otherwise grounded) internal parasitic capacitance away from true ground. That would allow it to charge against its other plate, which is grounded internally.
 
Still doesn't explain the non-deterministic nature of its readings. You'd think that 2.5K would allow it to go high, or go low, but not fluctuate between the rails like that. Oh well.
 
Thank you for responding! Perhaps all of us, together, can remember all of the little details of this CTMU module. It's pretty clear that Microchip doesn't intend to support it.
 
Thanks again, very very much!

有道理，没错！



      以下为原文

    Makes sense, yep!

当我们共享CTMU故事{GRIN }时，有没有其他人注意到当CTMU激活时ADC的采样和保持基本上是无用的？示例：如果在转换过程中启用了当前源，ADC的输出值就会上升——大概是因为额外的电荷在累积。但是，S/H应该将ADC与信号隔离开来。S/H的整个点是在转换期间保持信号稳定。当然，CTMU并不是这样工作的。我还没有做过实验，看看S/H是否与“正常”的外部模拟信号的行为方式相同。如果不是，那么我假设当CTMU被启用到ADC的多路复用器的后端时，S/H被切换出电路。如果S/H的行为是相同的，那么S/H只是被打破了，因为它确实不“保持”很好。



      以下为原文

    While we're sharing CTMU stories {grin}, has anyone else noticed that the ADC's sample and hold is basically useless when the CTMU is active?
 
Example: If you leave the current source enabled during the conversion, the output value of the ADC goes up - presumably because additional charge is accumulating. But a S/H should isolate the ADC from the signal... the whole point of a S/H is to hold the signal steady during the conversion period. Sure doesn't work that way with the CTMU, though.
 
I have not run experiments to see if the S/H behaves the same way with "normal" external analog signals. If not, then I presume the S/H is switched out of the circuit when the CTMU is enabled onto the back of the ADC's multiplexer. If the S/H DOES behave the same, then the S/H is just broken because it sure doesn't "hold" very well.

我想我再次同意，但没有得到太多的细节…我应该重新连接板和它的RS232，并做一些测试…



      以下为原文

    I think I agree, again, but did not get much into details...
I should reconnect the board and its RS232 and do some tests...

你看过数据表，如第241页吗？HTTP:/WW1.MICCHIP.COM/DeLoSs/En/DeVICECD/399 77 F.PDFA.这个应用程序说明了什么？http://WW1.Microchip .com /下载/En/AppNOTEN/CTMU%2002 1375 A.PDFHTTP:/WW1.MicroChIP.COM/DeLoSs/En/AppNoSe/01317A.PDFGEALDAL-



      以下为原文

   
Have you read the datasheet, eg. page 241?
http://ww1.microchip.com/downloads/en/DeviceDoc/39977f.pdf
 
also this app-notes?
http://ww1.microchip.com/downloads/en/appnotes/ctmu%2001375a.pdf
http://ww1.microchip.com/downloads/en/AppNotes/01317A.pdf
 
Gerald
---
 

是的，我试着阅读这个模块的所有内容。我指的是PIC18F PDF文件，它经常在我的桌面上有一个快捷方式。我有一个CTMU目录，它有你引用的另外两个文件的副本，还有许多其他文件。FIY，在PIC24家庭用户指南中有一个关于CTMU的专章（虽然CTMU模块的实现在两个家庭之间有所不同）。我错过了关于采样和保持放大器的一些东西来解释我所看到的吗？谢谢！



      以下为原文

    Yes, I've tried to read everything I can on this module. I refer to that PIC18F PDF file so often it has a shortcut on my desktop. And I have a CTMU directory that has copies of the other two files you cited, as well as many others. FYI, there is a dedicated chapter on the CTMU in the PIC24 family user guide (though the CTMU module implementation differs between the two families).
 
Did I miss something about the sample and hold amplifier that would explain what I'm seeing?
 
Thanks!

顺便说一句，值得注意的是，在PIC18F文件，第18.4.1节中，它具体地说：& GT；&，跨RCAL电压的时间延迟稳定和A/D采样/保持电容器充电。在使用中。然而，现实世界的结果并不一致。



      以下为原文

    BTW, it is worth noting that in that PIC18F document, section 18.4.1, it specifically says:
 
>>Issue time delay for voltage across RCAL to stabilize and A/D sample/hold capacitor to charge.<< [emphasis added]
 
This strongly implies that they believe the S/H to be operational when the CTMU is in use. The real world results don't agree, however.
 

对不起，我不同意这个结论。您强调的语句意味着采样/保持电容器的存在，它并不意味着采样/保持开关是可操作的。一个小但可能重要的区别…



      以下为原文

   
Sorry but I disagree with that conclusion. The statement you highlight implies the sample/hold capacitor is present it DOES NOT imply the sample/hold switch is operational. A small but possibly important difference...

好吧，严格来说，你是对的。这是可以解释的。但这也是S/H如何工作的准确描述。你会等待适当的时间，让电容器充电之前，去耦输入和让A/D测量存储的电荷。关于S/H交换机是否活跃的问题，他们只是沉默了。我想说的是明确的和明确的。现在我们都在黄昏徘徊，所有的感觉大象，试图描述我们相信我们认为我们可能知道。我说“黄昏”是因为我们到处都有各式各样的细节（比如上面引用的引文），但光照不足以看到整个大象。



      以下为原文

    OK, strictly speaking you're correct. It could be interpreted that way. But that's also an accurate description of how a S/H would work... you'd wait the appropriate amount of time to let its capacitor charge before decoupling the input and letting the A/D measure the stored charge. They're simply silent on the question of whether the S/H switch is active.
 
What I'd like is something clear and explicit. Right now we are all wandering around in the dusk, all feeling the elephant, trying to describe what we believe we think we might know. I say "dusk" because we have snippets of specs here and there (such as the quote I cited above) but it's not enough "light" to actually see the whole elephant.

同意。目前我没有安装任何测试，但是我（我确信很多其他人）会对您在CTMU问题上进行的任何测试的结果非常感兴趣。过去我已经使用过几次，但由于缺乏可用的信息而没有任何真正的信心。



      以下为原文

    Agreed. I do not have the setup currently to do any testing however I (as I am sure many others) would be very interested in the results of any tests you conduct on the issues with the CTMU. I have used it a few times in the past but not with any real confidence due to the lack of information available.

我们得到了很好的结果，从某种意义上说，它们是合理可重复的。但是我们的应用是相对的而不是绝对的，也就是说我们没有建造实验室电容器！我们本质上把它当作一个恒定电流源，可以由定时器触发。Microchip的附注暗示了各种其他可能性，提醒我NE555计时器芯片何时出现。这是一个混合了几个建筑砌块，可以以各种方式连接，以建立“几乎任何东西”。它确实是灵活的，但最终使用的是几个核心选项，更深奥的一部分（大部分是由于NE555不是特别精确的事实）而死亡。如果微芯片对CTMU的实际情况更加开放，它可能会刺激更多的使用。外围设备，因此包括它的MCU。基本上，这个行业需要一个完整的规格表。我希望他们在倾听，因为CTMU确实是一个聪明的设备，它有可能作为MCU的一部分而令人惊讶地有用。如果工程师知道如何正确地使用它。IMHO文档中最大的空缺是CTMU是如何连接到另一个P上的，并与其他P交互的。小精灵。CTMU本身是相当有记录的，但它基本上是一个猜测和实验游戏，以找出它如何与其他事情一起玩。我们现在使用计时器和ADC几乎完全自动化了，但是到达那里真的很痛苦。



      以下为原文

    We're getting good results with it, in the sense that they are reasonably repeatable. But our applications are relative rather than absolute, i.e. we're not building a laboratory capacitance meter!
 
We're essentially treating it as a constant current source that can be triggered by timers. Microchip's appnotes suggest all sorts of other possibilities, reminding me of when the NE555 timer chip came out. It was a mixup of several architectural building blocks that could be connected in various ways to build "almost anything". It was indeed flexible but eventually the uses settled down to a few core options and the more esoteric ones died off (due in large part to the fact that the NE555 wasn't especially accurate).
 
If Microchip were more open about what's really going on with the CTMU, it might stimulate more use of the peripheral and thus the MCU's that include it. Basically the industry needs a *complete* spec sheet. I hope they're listening, as the CTMU really is a clever device that has the potential to be surprisingly useful as part of an MCU - if Engineers knew more about how it can (properly) be used.
 
IMHO the biggest void in the documentation is how the CTMU is connected to, and interacts with, the other peripherals. The CTMU itself is reasonably well documented, but it's basically a guessing and experimental game to figure out how it plays with other things. We have it almost entirely automated now using a timer and the ADC, but getting there was really painful.

最后，我得到了一个实验室级电容器，一个使用恒定电流源来测量电容就像CTMU一样（与一些交流频率相反，这可能由于电抗而产生不同的结果）。我想用CTMU之外的CTMU来表征我们正在测量的电容。有趣的发现：CTMU报告了在特定应用中的电容的~3:1动态范围。但是，在被测试的电路中，同样的精确的测试装置，单独测量一个6:1的动态范围。结论：CTMU（或A/D，或MUX，或端口PIN接口电路，或某物）影响测量。我的猜测是MCU的寄生电容增加了器件的电容，使得被测器件的动态范围只将电容量的一半减去。例如：由仪器测量的固有器件电容是26-173PF（～6.6:1）。如果MCU增加（比如说）26PF，那么摆动就变为52-19PF（~3.6∶1）。寄生电容是生命的事实，所以它不会消失。但至少现在我能把它描述一下，更好地了解我们所看到的。



      以下为原文

    Finally got my hands on a lab grade capacitance meter, one that uses a constant current source to measure the capacitance just like the CTMU does (as opposed to some AC frequency, which could yield different results due to reactance). I wanted to characterize the capacitances we are measuring with the CTMU outside of the CTMU itself.
 
Interesting discovery: The CTMU reports a ~3:1 dynamic range of capacitance in a specific application. But that same exact device under test, with the circuitry removed, measures a 6:1 dynamic range sitting alone by itself.
 
Conclusion: The CTMU (or A/D, or mux, or port pin interface circuitry, or something) is affecting the measurement. My guess is that the MCU's parasitic capacitance adds to the device capacitance, such that the dynamic range of the device under test only swings the capacitance by half the total.
 
Example: The inherent device capacitance, as measured by the meter, is 26-173pF (~6.6:1). If the MCU adds (say) 26pF to that, then the swing changes to 52-199pF (~3.6:1).
 
Parasitic capacitance is a fact of life, so it's not going away. But now at least I'm able to characterize it a bit and better understand what we're seeing.
 
Research continues....

谢谢



      以下为原文

    Thanks