Alternate plan: I assume these measured control values go through an ADC at some point, so they're probably better stored as integers anyway. If it's an 8-bit ADC, the value 12 is better stored as `INT(12)` rather than as `12.0 / 256.0`, whether ASCII or FP decimal. Now you're storing data in 32-bit values, so the ratio between your present storage method and this one is 2:1 best case, not 4:1. As soon as you get to tenths of precision, it's 1:1 between the two cases: "A.B," is 4 bytes. You may want to add a "scale" constant to the `Presets` table so you can divide that into the measured integers to recover the decimal form. This will then let you work with systems with different levels of precision, such as a later 10-bit high-precision system, where the scale would be 1024.