230V AC Wave forms in an analog ocillocope

You need an inductor in series with the circuit. To make the mains voltage collapse there is a lot of current flowing into your heater for a short time. An inductor will choke this, while allowing the 60 Hz current to flow.

An example value of 1 mH will present a 0.4 ohm reactance to the mains frequency current, so will drop a few volts with a 2 kW load. But it will present a 1000 ohm reactance to the spike in question, in the <10us time range. This might be a bit much inductance, requiring a large core and coil, but you could find something suitable in the range 100 to 1000 uH.

Edit: looking at your circuit, you could place the inductor on the output side of your circuit, so it's in series with the triac but not with the control system.

took a crack at it. this isn't it, goes from 0 to less than 90 deg shift.

principle along the lines of @hacktastical's comment.

pot makes two symmetrical control currents, and diode bridge steers, steering controlled by input clock. output of steered current bridge makes variable slope ramps out of the input clock's rising and falling edges. then comparator vs mid supply produces output.

falstad original here ... note "resistance" slider on the right that controls the pot. changes take a few cycles to settle (in case it is not ramping rail to rail, because of action of the 20k resistor to center it. but at that point the phase shift is maxed out anyway)