MCP39F521寄存器读取错误

由于您没有显示任何代码，您只需随意猜测可能出错的内容。



      以下为原文

    As you have not shown any of your code, you're just going to get random guesses about what might be wrong.
 
 

好点…我真的不相信这个问题是在我的代码，我不这么说…这样做太久了。1。核心I2C代码是我们自己的库，在许多I2C应用程序中运行了超过5年。2。运行此代码的板是一个在R2C和F2RAM上建立的设计，它在没有错误的情况下运行。这是一个产品设计，我用它来测试新的交流测量板。3。我在第一个消息中有一个打字错误，我估计我的错误率是10%，但我在第二段中键入了90%。4。我做了一个长期的测试，发现我真正的错误率是5%，不是10%。5。我没有从MCP39 F521中检测到任何响应消息，除了在第一个字节中的ACK以外，我的系统也没有检测到任何I2C总线错误。6。昨晚我运行的系统没有交流电源通过电路板，我的错误率从5%下降到0.02%（10的50K响应）。这可能与数据或电路噪声有关？我将尝试在第一个帖子中添加逻辑分析器屏幕截图，绿色箭头指示从MCP30F521正确的格式化的响应，红色箭头指向坏的响应。这些不良反应并不是在这里出现的，而是一个接着一组的3到5个。



      以下为原文

    Good point ...
I honestly don't believe the problem is in my code and I don't say that lightly ... been doing this too long.
1.  The core I2C code is our own library which has been running over 5 years in many I2C applications.
2.  The board running this code is an established design with an RTCC and an FRAM on the I2C bus which operate with no errors. This is a production design which I am using to test the new AC measurement board.
3.  I had a typo in the first message, I had estimated my error rate at 10% but I typed 90% in the second paragraph.
4.  I ran a long term test and found that my real error rate is 5%, not 10%.
5.  I have not detected any response messages from the MCP39F521 with anything other than ACK in the first byte, nor has my system detected any I2C bus errors.
6.  Last night I ran the system without the AC power running through the board, my error rate dropped from 5% to 0.02% (10 out of 50k responses). So this might be related to data or circuit noise?
 
I am going to try to add a logic analyzer screenshot to the first post, the green arrows indicate properly formatted responses from the MCP39F521, the red arrows point to the bad responses. These bad responses do not come one here then one there, they tend to come in groups of 3 to 5 one right after the other ...
 
Larry

经过大量的测试和代码调整，我们有了解决方案，下面的更改将我的错误率降低到超过350K请求的4个被损坏的响应的顺序：1。在发送消息以设置地址并读取字节之后，在尝试从芯片读取数据之前，我们有10ms的延迟。2。我们也有最小的2ms之间的数据包发送到任何设备上的总线。拉里



      以下为原文

    After a lot of testing and code tweaking we have the solution, the changes below decreased my error rate to something on the order of 4 mangled responses out of over 350k requests:
 
1. After sending the message to set the address and read the bytes, we have a 10mS delay before trying to read the data back from the chip.
2. We also have a minimum of 2mS between the packets sent to any device on the bus.
 
Larry