Why is it recommended to use Split in FT8?
The answer is simple, the radio has a bandpass filter, which produces the SSB signal, whose bandwidth is 2.4 Khz, that is, as the frequency response is designed for human voice, it only lets through from 300 Hz to 2400 Hz.
In modern equipment this is sometimes adjustable and can be opened up to 3000 Hz, but in the equipment of the 80's, 90's and the first decade of the 2000's, it is not. This causes the frequency response when transmitting to be low at the extremes, i.e. below 300 and above 2500 Hz, for this reason you may have noticed that the power is lower when transmitting at the extremes.
For this, the following compensation is made, the program generates the 1500~2000 Hz audio signal at all times and the difference is made by producing a change in the radio: the split, either on the same VFO or on another one. Of course, the radio must be controlled by CAT Control.
If the transmit cursor is in the 1500~2000 Hz range, there will be no frequency difference in the radio, but if the transmit cursor is for example between 1000~1500 Hz, then the radio will have 500 Hz less, likewise if the transmit cursor is between 2500~3000 Hz then the radio will have 1000 Hz above.
Since FT8 normally uses the full spectrum from 100 to 3000 Hz, by using split you can transmit over the entire spectrum while maintaining the same output level and keeping the ALC within the limit.
It is advisable whenever possible to use splitting when making FT8, FT4, etc.