I guess I need to make the loop larger. I'll have to experiment and see what I can come up with.
A larger loop generally would improve efficiency, but that alone doesn't solve your current
problem. If you want to use the loop on the higher bands, you can't make it too large.
But for just 40m, making it larger is a good starting point.
The variable capacitor needs to be sized for the appropriate tuning range. You might look
at the setting of the current variable and estimate how much capacitance you really need,
then use perhaps a 50 or 100pf with some fixed capacitance across it (or even just a small
variable for fine tuning across the current one.) The larger the capacitor the sharper the
tuning because the ratio of pf / degree of revolution is higher. Ideally you would restrict
the tuning range to, say, 6 to 8 MHz. (The larger capacitor will also limit the high
frequency coverage because of the higher minimum capacitance.)
The SWR at resonance is determined by the shape, size and placement of the coupling loop.
This is what transforms the low impedance of the loop to something that can be matched
to the feedline. The common rule of thumb is 1/5 the diameter of the main loop, which is
a good starting point, but needs some experimentation. Try varying the length of the wire
and making the loop more oblong in one dimension or the other to see what effect that
has. The exact requirements depend somewhat on the local environment and losses, but
if you made the loop itself longer than the original you may need to make the coupling loop
larger as well.