Just made an account to chime in and increase the number of affected cartridges, as my cartridge/shell was exhibiting the same behaviour too. And to document it somewhat for posterity.
At first I was considering to try Alex's solution
but I believe Utgardaloki76
is more on point on this. So I think Alex's approach is one right solution but the wrong answer overall.
To confirm, I compared the functionality of the cartridge between PCBs only and PCBs + spacer (not even the other parts of the shell), as Utgardaloki76 did in their post. In fact, I compared it before I even read that post, so I was extremely confused until then, since the PCBs booted fine on their own, but after just inserting the spacer it didn't...
(Oh! Also, it didn't boot up to a white screen but a hyper-pixelated version of the NEOSD-Menu, like you could make out the UI elements for the list (white border rectangle or lines in upper region) the artwork (fully coloured square lower left) and the buttons prompts (fully coloured rectangle lower right), but only the "background" of it. It kinda reminded me of an overblown Atari-era game. I wish I took a photo of it, since it was really weird, kinda spooky even since it also showed some overall noise which scanned over the picture fast, then slow until it went black after like a minute or so. I think the colours were unstable too. EDIT: In fact it's a lot like the video fredjss
Anyway, again, with just the PCBs it works as expected. Insert the spacer (i.e. _just_ the spacer) between the two boards and boot it again, and it won't work, no matter how high or low the spacer is.
Therefore, the culprit wasn't the insertion length or shortage thereof, but the spacing _between_ the boards.
Alex's solution just may work, but only because Utgardaloki76 explained why in their post: the raised spacer can alleviate the strain between the PCBs, and may consequently provide better contact for the socket.
To double-double-check I also just compared (marking with a fine sharpie) how much the PCBs do get inserted with and without spacer. No difference, as you would expect now.
So, Terraonion, I would suggest to remove or at least comment Alex's solution before anyone starts digging in the wrong place (obligatory call to rewatch Raiders of the Lost Ark
there). I'd also revise the blanket decision to ship all future cartridges with the spacer raised, since it apparently isn't the right fix for everybody, but then again I'm in no position to do so. At least we got to hear a charming Spani... - excuse me - Andorran-English accent (sorry :P).
As for the right answer (at least imo) look at Utgardaloki76's post
Regarding measurements, the spacing between my board connectors is pretty much 10mm and the spacing between the two spacer slots was 11mm. The measurements are from centre to centre. It's a usual NEO-AES board btw.
(Those serve as "tangible" but by no means accurate measurements as they were taken with a simple ruler.)
So I had to shave off around 1mm worth of plastic, or eyeballing .5mm per slot respectively towards the middle. To make it easier I also cut parts off the plastic "nubs" (perpendicular in relation to the slots, probably intended for stabilisation, four in total) since I had to shave along the whole side of the slot. I never manufactured things like this, and have no knowledge of any allowed tolerances but 1mm still seems a little bit too much. And if it's not a product of manufacturing error it's a design error, but I'll leave that to smarter gals&guys.
In any case, I got it working this way. Incidentally I first suspected the power supply (formerly 2.5A) and am now in possession of an overpowered 4A one, but that's on me.
So in the end, for other users experiencing the same: Double-check yourself if the "official" solution (raising the spacer) is even the right way to go, and if not read Utgardaloki76's post
and go from there.