Leif, I tried to replicate this but with only partial success. In EV3 I got a credible looking tile exactly centres on the N69 00 0.0 E024 00 0.0 user waypoint I had set up. So I think that matches your successful EV3 check. In EV4, for the same waypoint, like you this time I got a tile with a "white centre", but credible looking terrain round the edge. So this corroborates your "corrupt tile" finding. However, unlike your experience, in my EV4 (both on iPhone and Win 10) that (corrupt) tile was still exactly centred on the user waypoint, where is should be. See attached image. So I cannot corroborate your "displaced co-ordinates" finding.

I have no idea what might be causing your problem, the only thing I could think of (and I admit it's a long shot!) is to enquire whether you have recently been using the new "bearing/range offset" facility in teh Search function (e.g. that will allow you to create a waypoint say 5 nm NE of a fix)? perhaps an offset you might have used a while ago is still being applied ???? (I've never used that facility so I cannot compare my experience with it.)

Interesting.  We have a different approach.

you first created the user waypoint, and then selected google from the WP menu.   That works fine !

I entered the coordinates into the „search with google“.  This will translate them to a place name, which turns out to be offset.

Thanks for explaining your method, Leif! 

I think the point is that the EV4 Search function is not exactly a "go to" function it is a SEARCH (Usually a search through all our Aerodata items, but in the special case of the Google search, a list of all the places Google Earth knows about), and finds anything it can that can somehow be matched with what you have typed. If it recognises what you have put in as a lat/long, it shows the nearest thing it can find to those co-ordinates, but statistically it is very unlikely that it will find something at EXACTLY those co-ordinates (especially in the far-north region you have chosen! - or at a mid-ocean location!) so it can only show the nearest thing it can find to that position, which may be a considerable distance away. It then allows you either to look on the map at what it has found, or to set a Google tile there (or both).

Doing a search for 49, -37 gives a similar "curious" but in its way logical outcome! 

I don't, though, have any theory about the "corrupt tiles" that such searches seem often (but *NOT* always) to generate!

Note the "google search" option in EV4 is NOT a latitude/longitude search like in EV3. the used Google API simply tries to find a feature based on what you are typing (things like "white house","brandenburg tur" etc are all possible, even with spelling errors), and doesn't know that you actually mean latitude/longitude coordinates. 

The white "cutouts" are interesting. It looks like for those areas no highres tile is available for download, only the lower res. I need some time to investigate....