Hello,
I also tested adapting the impedance of the osciloscope to 50 Ohm (the osciloscope has this option) and nothing changes....
Insted saw tooth wave is perfect. This signal has also armonics, not as much as the square...
Regards
As Electrofan points out, there has been a lot of previous threads on this subject - so it is worth searching.
I had a quick look at the specs and the ramp only goes up to 1MHz, so if the bandwidth is around 40MHz you could go up to the 40th harmonic without significant attenuation!
In fact the series of generators goes up to 120MHz so the bandwidth may be more than 40MHz. The datasheet gives the rise/fall times as 9nsecs which corresponds to a BW of around 40MHz.
If you want a square "square wave" then (and this is arbitrary) you might want the rise and fall times to be around 1/4 of the square which is 1/8 of the period which corresponds to a period of around 72nsecs which is a frequency of around 14MHz.
The upper limit of 25MHz is a software set thing corresponding to a period of only 40 nsecs which gives the length of the square as only 20 nsecs (for 50% duty cycle) and given that the rise time is 9 nsecs and the fall time is 9 nsecs (10% to 90% I presume) the wave hasn't time to finish rising before it has to start falling.