@pernia you can do that too. you can implement a proper binary float32 as defined by ieee 754, you can store the exponent as an 8 or 16 bit variable and the mantissa as a 32 bit one (both signed), or you can do a binary decimal thing