I am exactly the same as Cobalt, he just explained it better.
I think the reason this is confusing is that with the KI256 you have almost no “horizon” (especially with the FD bars obscuring most of it) so one is using the two triangles (circled in yellow) instead
Whereas with the lower presentation you can use either the big horizon (which is probably what all “glass” pilots use, without actually realising it) or the two triangles but these triangles work in the opposite direction!
See this video which I have just done
But once you consciously use just the big horizon then it is OK to fly with.
Interesting to see there’s a slight difference between the angle of bank what the mechanical one and electronic AI show (especially while in motion). I’ve noticed this difference between the Garmin G5 and vaguely post-war Sperry Artificial Horizon, too.
Interestingly, the default G5 presentation is the same as the Bendix King mechanical instrument – the pointer stays fixed on the display and the arc above it moves.
That video was with the autopilot on. I did some hand flown but they were messy, especially trying to hold the phone such that my face was not reflecting in the KI256 
I vaguely recall that there is an adjustment in the KC225 to balance up the left and right roll angles (they are supposed to be 22 degrees IIRC, for the TB20). I have not had this done because there is only one man in the UK who knows how to do it Actually it should be possible via the laptop menu; also I have the KC225 bus extender. That’s one issue.
The other visible issue is that the KI256 clearly responds faster than the SG102. My guess is that the delay is due to the ARINC429 digital data stream from the SG102 to the SN3500, which runs at some number of packets per second, so the update rate is not super quick. That feature of the Sandel EHSI is called “reversionary AI” and not certified as a primary AI. It would not surprise me if years ago they were looking at that market but today is too late, with the EA100/GAD43 solutions for the EFD1000/G500 etc. I think Sandel should have done it because the KI256 is extremely widely used and costs about $10,000. King’s version – KI300 – is vapourware.
For sure, if you want to use the lower AI style to recover from an unusual attitude, you need to be awfully current with it! Or just remember to use the big “horizon” and forget the pointers.
I’d be intrigued to know how many pilots actually use the pointers?
We have established that I don’t and Peter does, but is there any way of finding out on a wider basis, say by poll on here?
This discussion was interesting to me at first, as I don’t recall being taught to look at the pointer, other than to check bank angle (to set rate one turn for instance). But then, I think I normally either apply the control and put the triangle on the ~19degrees. Basically I think I mostly use for fine adjustment.
I might also have forgotten what I was taught!
I use it on checkrides for the left then right turns.
JasonC wrote:
I use it on checkrides for the left then right turns.
To actually find out the direction of the turn/angle or fine tune your bank angle?
The problem (for me at least) is not reading the current bank angle, which is obviously easy. It is interpreting the dynamic situation. To control the plane in roll you need to react instantly to the dynamic situation, and the above video shows that the two are the opposite of each other.
Noe wrote:
To actually find out the direction of the turn/angle or fine tune your bank angle?
Fine tune. I know which way I turned. :)
