True double precision 64bit on ATMega

Can someone confirm that 64 bit double precision is basically not possible on the ATMega series (I know its supported on Due)

If so I guess my options are to use the BigNumber library and and scale all numbers so that the my smallest number e.g. 0.00000000000000001 becomes 1

i.e I'm trying to use a IMU unit MPU9150 and I appear to be having issues with loss of precision, where I'm unable to remove a sporadic offset in one accelerometer e.g. of 0.00000001 because I can't subtract subtract a long term average value of 0.000000005 because the Arduino maths lib doesnt have enough precision to do this.

Actually, I'm going to attempt to re-scale the numbers a bit e.g multiply by 1000 to see if this helps at all, but fundamentally I think I need more precision :frowning:

If you use the BigNumber library you don’t have to scale anything because you can choose as many decimal places as you want (eg. 50).


Thanks for the clarification.

I will give your BigNumber library a try