It's not changing frequency, it's skipping cycles. Even your 40 Volt image is skipping. For the UC3843, there should be one switch pulse (yellow trace) for each oscillator cycle (blue trace.) And, if it's properly regulating, that switch pulse should end before the oscillator signal hit's it's max value.
The fact that you have switch pulses that are equal to the rise time of the oscillator shows that values in your circuit are out of balance.
You show only the schematic for the circuit that you built, which I assume that you copied from somewhere. There are a lot of other factors that are important in building a switch mode power supply. Was there any more information from the design source giving more details? Particularly about the transformer.
My first guess about the circuit behavior is that the primary inductance of the transformer is too low with respect to the operating frequency.