so i've been building this 40m CW xmtr from scratch for the past couple months, loosely based on the old ARRL handbook crystal oscillator designs but with some mods i thought made sense at the time. finally got it oscillating and driving the final stage last night and it's putting out maybe 3-4 watts into a dummy load which is more than i expected honestly.

problem is when i look at the output on my scope the waveform has this weird shelf or shoulder on the positive peaks. not like a clean sine at all. i thought maybe it was my IRF510 being driven too hard so i backed off the drive but it didnt really help. could also be my output filter isnt doing what i think its doing, i wound the toroid myself and im not 100% sure i counted the turns right on the secondary. the key clicks on receive sound not great either which is a separate annoyance. anyone dealt with something like this before

the shoulder thing on the peaks is almost certainly flat-topping, means the IRF510 is being driven into saturation even if you think you backed off the drive. those fets are real sensitive to gate voltage and the sweet spot is narrow. also the IRF510 has kind of a rough reputation for RF work, it can work but the capacitances are all over the place and if your bias isnt set right you get that exact symptom. try dropping your drain voltage a bit and see if the waveform cleans up before you chase the filter. and yeah keyclicks are usually a shaping issue at the keying circuit, if you're hard keying the oscillator or the driver stage directly that'll do it every time. a simple RC network on the key line smooths things out a lot.

recount those toroid turns, seriously. i built a 7mhz lpf once and was off by two turns on one winding and the thing was passing harmonics like it wasnt even there. grabbed my antenna analyzer and the rolloff was way off from what the design said. fixed the turns and suddenly everything looked right. probably not your whole problem but if your filter isnt right the scope is going to show you a mess regardless of what the amp is doing. whats your cutoff frequency supposed to be and do you know what core you used