HX711 hal Modul

fupeama wrote:

But now I need convert u32 to float for using limit3 as input from hx711.

Yes, unfortunately that is the case. You could have your comp export two functions (like the encoder and stepgen do) with one doing the updating and the other just setting the float output, but that is just as complicated in HAL as using a conv_u32_float component.

fupeama wrote:

addf hx711.0.fast base-thread
addf hx711.0.slow servo-thread

Yes, exactly, though that is just as much extra HAL-effort as adding an int-to-float coversion.
One thing that might make it worthwhile would be to add scale and offset pins so that the output value can be converted to engineering units.

Why halcompile allowed me to compile modul with fp operation as assigning a value to float without any warning or error message?

The halcompile documentation does actually say that it has no way to check for FP operations, so it has no way to warn if they are used in the base thread.

What will happen, if modul is compiled as nofp with fp operation?

I don't know. Clearly nothing very bad, as the component compiles and works properly. I think that if you were to halscope the function time you would find that there was an increase in execution time every time the output was updated, though I don't know by how much.

Is somewhere described which operation is fp nofp?

Basically anything which uses a floating point variable or returns a floating-point result.


Do you have your work in a git repository, or are you working with an installed system and just writing the comp? The reason I ask is that it would be easier to add your component to LinuxCNC if you could submit it as a git-format patch (which would retain author and signed-off by atribution) but there are other ways to do that if you are not working inside Git.