The MCP3208 related bug reminds me something similar when playing with MCP3201. In developing the function TestMcp320101, I kept on patching and patching and forgot where and why I did the patching for. And in the statement to get adcDecimalValue, I multiply adcOutPutDoubleByteList by 2 ** 7. I vaguely remember I first try 2 ** 8, but found one order too high, so I changed it to 2 ** 7 and found it OK. But I did not go deep if indeed I should use 2 ** 7.
The other big mistake is that I casually assume that I assume the most significant 4 bits sit in the least significant part of the first read in byte. I also assume that the most significant part of the byte is zero. But there is no reason to assume that! I think that must be the reason that from time to time the function does not work and give a big and wrong analog voltage.
So I should do the same thing to make sure the indeterministic part should be cropped away.
.END
No comments:
Post a Comment