[Wavytone]

Alex, all digital cameras seem to have a sweet spot with respect to ISO, as you know the fundamental issue is noise vs exposure duration. What I've done with mine is this:

- go to a really dark site where you really will hit the sensor limits regarding noise; suburbia isn't dark enough for this test.

- ideally have the camera on a tripod or mount so you can repeatedly photograph the same region of sky, this helps the subsequent comparisons.

- take a series of shots that should have the same exposure, using the same lens aperture for all (f/2 is good for this - helps if you can set the camera to manual everything) and varying both the ISO and exposure duration by factors of 2. For example:

ISO 100 for 4 minutes,
ISO 200 for 2 minutes
ISO 400 for 1 minute
ISO 800 for 30 sec
ISO 1600 for 15 sec
ISO 3200 for 8 sec

Even though star trails will appear in the longer shots, what you're really interested in is the sensor noise - which appears as a speckle with no trails. Anything that trails is a star.

Examine the dark areas of the resulting images and you may find the optimum shot (lowest noise) is not necessarily the base ISO of the camera. For example with both my Panasonic LX5 (base ISO 80) and GX85 (base ISO 200) the best results are at ISO 400.

From semiconductor theory it's possibly also temperature dependant to some extent. Although... I tried the test in subzero conditions one night at Shangri-La at 3200m altitude and still got the same results as I did at Mt Cook in NZ in summer (ie ISO 400).

Lastly some cameras do automatic dark frame subtraction in-camera (the Panasonics do) just to make matters more complicated.