
Tonight, I have given a rough test to most of the filters I have. I used an ordinary torch bulb since they give a broad spectrum. Camera was the 450D with all filters removed. External 500 lines/mm grating on EF 35mm f/2 lens. Analysis by Rspec.
Notes on interpreting the following charts:
* The blue outline is the spectrum of the source light as detected by the camera. You can see relative to that how much of a given wavelength is passed.
* Note the amplitudes might not be perfectly matched between them.
* The focus and slit are not perfect. So take come caution when looking at passband widths. Assume the system isn't well focused in the IR region
* Some charts includes lines representing visible light pollution from Sodium and Mercury lights.
Narrowband filters
Astronomik SII CCD 12nm

Astronomik Ha CCD 12nm

Astronomik OIII CCD 12nm
Light pollution and similar
SkyWatcher light pollution filter

Astronomik UHC - intended for visual use to pass emission nebula, it does not block IR if used for imaging.

Astronomik CLS - this version does not block IR

Astronomik CLS CCD - this version does block IR
Infra-red longpass filters
Astronomik Planet IR Pro 742

Hoya R72

Pixco R72 - this does not work well on unmodified cameras as it passes too much visible red light relative to the IR.

Unbranded 850nm
I also have a similar unbranded 950nm filter, but the sensitivity was so low in this configuration there wasn't anything to measure.
Random colour filters
GSO Wratten #25 red

GSO Wratten #58A light green

GSO Wratten #80A light blue

GSO Wratten #15A yellow
Personal Solar Telescope ERF
This is the ERF from a relatively new PST scope.

This is the ERF from an old PST scope, where the ERF is visibly degraded. It appears to have wider bandwidth so would be less selective. The relative amplitude may not be comparable due to the measurement method.
I also tried measuring the blocking filter but there wasn't sufficient sensitivity in this test setup.
And finally, I also tried the Baader CaK filter. At under 400nm, there also wasn't enough sensitivity in this configuration to get a measurement. Possibly I could boost exposure to work those...