Floating Point variable debug - limit decimal places ?

[tensaisakuragi06]

To sum up my project, i get temperature value from NTC. I made ADC reading. There is no problem about getting values. I define temperature value as floating point variable. When observing it on debug screen, because it was floating, numbers after . is too much like seven digit. As you see in picture, is there any way to reduce the number of digits on debug window?

temperatureDisplay

[Andrew Neil]

---

@tensaisakuragi06 wrote:

I define temperature value as floating point variable.

---

Why ?

[tensaisakuragi06]

Its a little disturbing. It feels like value changes frequently. I do not need precise measurement. Just two digits is enough for me. This is actually not a problem. I seek for if it was possible.

[waclawek.jan]

Instead of the variable itself, you can perhaps display an expression which contains rounded/truncated value.

Just an idea to be experimented with, maybe not viable - I don't use Cube.

JW

[Ozone]

You can do the calculation in integer. Or in integer fractions ("scaled math", e.g. in 1/10th °C).
Not using float at all will save you several kB of code space.

I can't really comment on the CubeIDE debugger, I don't use it.
But I would check if the context menu has such a precision option, or check the manual.

[tensaisakuragi06]

In excell, it gives option show two digits after , :) . Its not a major problem(it is actually not a problem). Maybe later this option can be added. I am not sure this kind of thing is logical or not. By the way, when i display value on LCD or OLED(any kind of screen), i can display it as two digits after . as library functions allows me to do. So, if it is not possible i can left it as it is.

[Andrew Neil · Accepted solution]

---

@Ozone wrote:

Not using float at all will save you several kB of code space.

---

Indeed.

Even though the F4 does have an FPU, it is only single-precision - which can cause issues with standard C libraries:

Be aware: Floating Point Operations on ARM Cortex-M4F

And that's apart from all the other issues inherent in floating-point on binary computers:

Pitfalls of Floating-Point Numbers (And How To Avoid Them)

What every computer scientist should know about floating-point arithmetic

See also: https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html 

[tensaisakuragi06]

Be aware: Floating Point Operations on ARM Cortex-M4F this article give idea and helpful for me. I can accept it as solution. TY.

[unsigned_char_array]

You may want to scale to an integer. If you want 2 digits multiply by 100, if you want 3 multiply by 1000. etc. Add 0.5 to round.

You can also scale back to float/double, but the downside is that you cannot represent all decimal fractions perfectly with powers of 2. Such as 0.1:

If you want more options you can use snprintf to print the value to a string and then add the string to live expression.

[Andrew Neil]

---

@tensaisakuragi06 wrote:

 I can accept it as solution. TY.

---

It is a side-track from the original question of formatting in the debugger.

The thing with formatting in the debugger is - how much would it actually get used?

Remember that it's just a debug view - it's not an end-user display

As noted, a lot of embedded system just avoid floating-point altogether anyhow.