There are loads of ways to kill yourself if you don't know the knobs 
Early on with my new plane, with an instructor too, we flew an ILS into Exeter.
Except we didn't fly the localiser. We flew the track of the last GPS leg 
That GPS track was very close to the localiser track, but the glideslope capture didn't work...
Nowadays, in addition to the normal ILS setup, I usually set up the airport as a DCT, pop the GPS into the OBS mode, set the OBS bearing to the localiser inbound track (which is not necessarily the runway bearing - see EGMD for example) and that gives me a nice line on the MFD which the plane ought to be tracking once it captures the LOC. (Caveat: the DCT to the airport is not necessarily the touchdown zone of the instrument runway you are flying to).
we flew an ILS into Exeter
Funny you should say that:
Last Sunday I was flying into Exeter, making a visual approach straight in from the east. Unusually, I decided to activate the ILS approach on the GTN (not for any particular reason - just practicing).
A few miles further on I was shocked to look down and see my CDI was the best part of half scale to the right. Chastising myself I turn to correct...
It took me about a minute to notice that the system had switched over to the ILS. I guess this goes to show that, even in very low workload VMC conditions (this was about the easiest approach I've ever done - straight in visually from FL60 in perfect VMC with no one else around) we can be caught out by the systems when we aren't thinking about them.
My excuse is that I was looking out for traffic...
Some CDIs have a source indicator builtin, that is actually a great feature. I have the same indication on my Aspen glass cockpit but there is so much on it that is by far not as obvious as an indicator light inside an analog CDI.
I like Peter's approach using the GPS OBS as a depiction of the localizer. My Garmin 695 shows the extended runway line for most airfields. In most cases that's also the localizer but not in all.
I have the Aspen as well - as you say, a bright light would be more obvious, but the text is usually enough when I'm eyes down on an instrument approach.
The GTN also displays extended runway lines - I use this to sanity check instrument approaches.
-- Johannes
The other handy thing about knowing when to expect the localiser is that, with some systems (and that includes my KFC225) if you press APR too early, it may (or may not) turn onto some spurious "LOC" signal.
One should not press APR until one sees the LOC bar moving in.
I don't know the cause of this, but I've had it a few times - at some locations, anyway.
there are a few conditions that have to be met. First of all, you have to enable the automatic GPS/VLOC switch in the settings of the Garmin device.
How do you do that? I have a 430, and didnt realise that was an option
I think the key even with auto switch is to check the switch occurs when you intercept the localiser. If you intercept a bit close for example or forget to flip the autoset loc freq it won't happen.
In the GNS530W (not sure about the 430/430W) the setting is called 'ILS CDI Capture' which has to be set to 'Auto'. This setting is in the AUX > Setup 1 Page.
The installer can disable this setting in the configuration pages when its not authorised in combination with your autopilot, such as with the KAP140 and KFC225.
Based on on the name of the setting it would seem auto switching wouldn't happen on a VOR approach - can't remember whether it does or doesn't but I do know you cannot rely on this auto switching and always need to verify CDI setting before you descend down the approach.
Personally my SOP is to switch as soon as I am getting vectors to final or, when flying the full procedure, as soon as I'm on the final approach course.
Jorgen
The installer can disable this setting in the configuration pages when its not authorised in combination with your autopilot, such as with the KAP140 and KFC225.
Why would these autopilots have a problem?
I don't know much about this stuff but I can see a lot of possible variation in installations.
For example you could install a GNS box and have the autopilot driven from a relay which selects between NAV (the output of a VOR/LOC receiver) and GPS (the output of the GNS box). The relay is operated with the NAV/GPS switch.
A lot of GNS boxes have been installed in that very simple non-integrated way - it's the cheapest option.
If the NAV/GPS switching is done inside the GNS (which must be the context of the original Q) then I can't see why an autopilot should have a problem - unless the need to press the APR button (which merely increases the lateral sensitivity) is a certification issue.
Personally my SOP is to switch as soon as I am getting vectors to final or, when flying the full procedure, as soon as I'm on the final approach course.
I have a little "ILS checklist" which starts off by setting up a DCT to the airport and setting up the OBS mode, so I have a separate confirmation of where the localiser lies.
There is some obscure reason for not pressing APR (to exit the HDG mode, for APR ARM) too early. Sometimes one seems to get spurious LOC signals...
Every GA airplane I have ever flown that was fitted with a IFR GPS was full of suicide switches. It is absolutely imperative that you create and follow a rigid approach SOP including GPS status calls (eg "confirming" term/LNAV/LPV/VLOC etc etc) at all the key approach gates.
Enroute I do a "1 min to waypoint" check. This involves going to the flight plan page to check that the next waypoint is in fact the one you expect and noting the next course to make sure the autopilot turns in the right direction.
