Points taken! Never really looked at quality in a signal generator. U
never stop to learn. Signal level garbage in - AMPLIFIED garbage
out.

A clean VFO or DDS + a tank circuit ahead of the PA looks like a much
better idea.

I have looked at a number of broadband amp designs, and it seems that
they share a real concern with cleanliness:
- linear amplification
- negative feedback (some even DC coupled)
- sometimes push-pull with matched devices.

As far as antenna matching is concerned, I'd first align the cascaded
tank circuits into a dummy load, and never operate without a tuner.

All this considered it looks like broadbanding requires choices that
together lead to a low power efficiency.

------

Now, re. filtering output with cascaded LC tank circuits, I have a
question stemming from this being the toroid era.

In the old days, multiple capacitor-tuned LC circuits in series could
be aligned over broad tuning ranges by means of slug adjusted coils and
small capacitive trimmers. It was a bit of an art, but satisfactory
alignment was possible.

But you can't easily trim a toroid.

When handling say up to 5 watts RF, do I already have to fear
saturation in IF-type, ferrite-tunable canned coils?

Or, does anyone ever put a small L of that kind as an indictive trimmer
in series with a toroid?