Modbus: Unsigned float

### Use Case

I am using power meters that provide measurement values using a Non-IEEE floating point formats:

Unsigned Measurement (32 bit):
- Decade Exponent(Signed 8 bit)
- Binary Unsigned Value (24 bit)
Example: 123456*10–3 stored as FD01 E240(16)

Signed Measurement (32 bit)
- Decade Exponent (Signed 8 bit)
- Binary Signed value (24 bit)
Example: - 123456*10–3 stored as FDFE 1DC0(16)

I am looking for a way to configure/extend the modbus input plugin to handle these values.

Am I right that currently, only IEEE (signed) float values are supported? Can I just come up with new data_types like UFLOAT32 and MSIGN_FLOAT32 (as the mantissa holds the sign)?

I'd be able to implement it on my own, just checking what a sane convention would be.

Kind regards

nischu

### Expected behavior

I would like the Modbus input plugin to generate correct floating point values

### Actual behavior

I cannot really specify this custom data type

### Additional info

https://cdn.findernet.com/app/uploads/Benutzerhandbuch_Typ_7M38_DE.pdf page 52

Start with the Modbus input plugin's data type configuration and existing IEEE float handling. Determine how unsigned and signed 32-bit decade-exponent values should be represented and converted, then verify that the plugin can configure them and emits the expected floating-point measurements from the examples.

Written by the indexing model from the issue text.