I'm not sure how to break up quotes, so I'll just copy and paste:
"In my case, I was able to perform the fx-570ms upgrade hack multiple times in a row, with only a ON press in between. But, I do a 'matrix hack' in between in the case when it doesn't work.
Also, that is a different 'Matrix hack' than I was referring to, that one involves pressing ' 
...' (Here is an example video: https://www.youtube....h?v=3kPirRKpCgo)"
The matrix hack I used was actually done on the fx-991ms, but works as well on my fx-82ms: https://youtu.be/Rn0EvhtOANA
It seems I can't do the '3030' matrix hack; as there is something else in memory rather than what is shown on the video. I'll try to do a full battery reset first, since if this matrix hack only needs to be used once, that will make it much better then the one I'm using (which needs to be applied every time).
"The fx-95ms has the EQuatioN solver mode, but notice that the EQN mode is on the the same screen as the MATrix and VeCTor modes on the fx-570ms. So, those modes are likely disabled at hardware level, and likely cannot be enabled with a software level hack."
After a bit more messing about I think you are right about this. My theory was that the missing pins 88, 89 & 90 might have something to do with selecting the actual model of the calculator. I scraped away a bit of the coating on pin 90 and found it was tied to the negative terminal or ground on the calculator. During the wire bonding process (before putting that black gunk on the top of the CPU), perhaps a combination of pins 88,89 & 90 are wire bonded to this ground to select the model. Unfortunately I'm not really an electronics person and only have a rudimentary knowledge of electronic stuff. I started however slowly drilling into the black gunk between the CPU (very slowly with a screwdriver) and where I believe the wires might be bonded to to see if there were any there. I didn't hit any after making about 1mm of progress, but decided I didn't want to risk permanently disabling the calculator without actually learning anything useful that I could use on the fx-95MS.
"Same for me. Even removing the battery doesn't seem to clear the memory completely, I once left the calculator without a battery for a whole day so all capacitors are drained as well, and I still got mixed results.
So, the success of a hack is really dependent on whatever data is currently in memory."
That is unusual. I did short the capacitor for a minute or so to make sure it properly reset, and it usually did the trick since the '53535' hack wouldn't work unless I did the '131313' matrix hack first, indicating to me what I thought was a successful reset. Though I can't rule out that I'm not successfully resetting my calculator to despite how it seems.
"Not only the display settings are kept, but also the last used mode is still active, but will be in a broken state."
That is interesting to know.
"I have read somewhere (I can't remember what site it was) that older models of those calculators didn't have the big black piece in the middle of the PCB, and that the black piece is preventing us from hardware-hacking the part of the PCB that determines what modes can be selected."
Do you mean the COB (Chip On Board, that black gunk covered chip), was easily accessible?, that would be rather handy and make life easier as you could compare the wire bonding between different models. I figure that the actual calculator software is hardcoded on the ROM to save costs, but it would be interesting if the software was actually contained on a built in EEPROM.
EDIT:
I Can't actually do that '303030' hack even after shorting the capacitor for a minute or so without power. I just get different result from what is shown ('Stack Error' instead of 'Math Error' and nothing like the letters shown after deleting all the 3030's for example).
Edited by MJim, 27 November 2019 - 12:54 AM.
