the tool coordinate system of a 6 DOF robot

65536 is a number that any programmer will immediately recognise. It is 2^16.
In a 16-bit system you can have 65536 different values (0 to 65535).
The number 65536 is 100000000 in binary.
65535 is 11111111 in 16-bit unsigned. 0-65535 encoding.
0 is 00000000
Perhaps most suspiciously, -1 is 11111111 in 16-bit signed. (-32768 to + 32767 encoding)
The system somewhere probably uses signed representation in a 16-bit accumulator and extends that to 32 or 64 bits to keep track of multiple turns.

We need to work out where this miss-count is taking place. Whether it is in LinuxCNC (where it can be fixed) or in the hardware, where it is more difficult.

Thanks very much, andypugh
I ask this question to my teacher, yes, just as you say, he found that the -65536 was zero too and there is something wrong with the type data of actual positon. And we have a test with another driver. we find that it has the same question in 0 position, then we changed the application program, and it is ok now!! The work now I should do is to find where the question is in Linuxcnc, from the data of halscope, I put my focus on lcec, because it is lcec to get the wrong data first.
Thanks again for all the help. I will tell you my results!!!

Hello andypugh:
Sorry for disturbing you again,yes,you are right, I use the function EC_READ_S32 to read the value of 0x6064. I find that, when the value is 1 in decimal and the value from 6064 is just 0x1 in hex, the value is 0 in decimal and the value from 6064 is 0x0 in hex, when it changed to -1, the value from 6064 is FFFF while is not FFFFFFFF which FFFF in 32bit is 65535 and FFFFFFFF in 32bit is just -1, when the value is changing to -2 , the value from 6064 is just be correct and is FFFFFFFE;
I don't know this reason. I find that I have some problems to read the source code!!
Sorry for disturbing you again and again , I am really new to this.

That is very strange. It sounds like a problem at the driver end,
though perhaps it is a problem in the serial protocol.
Is the output for -1 actually 0000FFFF or is it actually a shortened sequence?

Thanks.
I will give you my hal file and I send a private mail to you and it is great that you will check it.
BTW, FFFF in 16 bit is -1..