Sadly I can’t find a picture showing a bank angle, but these two presentations appear the opposite of each other when turning.
Today I flew a NP IAP by hand, in the dark, to see how well it works just on the lower one. It does take quite a while to get used to it.
Peter wrote:
Sadly I can’t find a picture showing a bank angle, …
Those of us who have to alternate between the two presentations know what you mean :-( IIRC there have been several accidents already (up to airlier level) because pilots misinterpreted AIs which were unfamiliar to them. Will do some googling when in the right mood.
Personally I have misread the electronic AI several times already I must confess. Never the traditional mechanical one in your top picture (the mechanical ones from Russia – I am told – work the opposite way, just like our electronic ones). Luckily I always noticed my mistake after a couple of degrees and could reverse bank without anyone noticing. It really becomes a problem in case of an abnormal attitude.
@Peter, I don’t have time now to grab / link them, but have a look at this page for some graphic representations of the two AI types in a turn:
Yes… I need to get up there and do a movie showing the two in turns.
As WN says, in an unusual attitude this could be a real problem. For small angles one can “cover it up” before anybody notices 
I find it quite spooky and wonder if there is a way to “get over it”.
Flash Airlines B737 at Sharm el Sheik in 2004 was one of the accidents attributed to EADI misinterpretion. I will come up with more.
NB: The problem has already been recognised a long time ago. Here is a 62 (!) year old paper on the subject: http://www.dtic.mil/dtic/tr/fulltext/u2/034642.pdf
Many electronic displays such as the Garmin G500 and Aspen EFD1000 are configurable to either type of roll index pointer and a specific requirement in the STCs is to ensure both the primary and standby displays are in the same format.
That’s quite a find, WN. I have put a local copy here
The illustrations are not great. In this one, I think the leftmost one is the KI256 but I can’t tell which of the other four is the Sandel
I think it is one of the last two.
This table is telling:
The traditional AI is the best of them for roll control but the last ones are better for pitch. However all of them are OK for pitch and this is exactly what I find: the KI256 is best for roll control but the Sandel (with its sizeable and very clear horizon) is great for pitch control.
Many electronic displays such as the Garmin G500 and Aspen EFD1000 are configurable to either type of roll index pointer and a specific requirement in the STCs is to ensure both the primary and standby displays are in the same format.
I wonder how many people know this can be configured, before they fly with it?
I remember that this issue was covered in my IR theory. The weird one is called a sky pointer IIRC. When I learned my theory I remember that I told myself how bad an idea that seemed, and being glad that I thought that I would never have to deal with an “inverse” sky pointer… that might prove wrong one day.
Here is a study that was linked from a pprune thread:
http://www.humanfactors.lth.se/fileadmin/lusa/Sidney_Dekker/articles/2003_and_before/Horizon.pdf
Peter said “I wonder how many people know this can be configured, before they fly with it?”
It’s set up in configuration mode, not in a pilot-accessible page. So it’s up to the installer to make sure they match the new EFIS to the standby instrument.
