It is not mine although I would like to try if it is true that the Neos is exactly like a MSX mouse....

By the way, what graphic package went with the original(patkány) Enterprise mouse? Had it some utility?

This is the Joystick counterpart:

http://www.ebay.es/itm/MSX-ML-50JY-JOYSTICK-Mitsubishi-Japanese-Controller-Excellent-NEW-BOXED-/331160150811?pt=UK_Controllers_Attachments&hash=item4d1aaee31b

It has a point... And the shape is similar.

I've found another microcomputer that used a MSX mouse. Exelvision was its name. I remember that was sold in Spain.

This is a good page abut it.

I've made a pin converter to connect the MSX mouse to the Boxsoft, and the result is discouraging. The mouse interacts at the movement but the pointer moves randomly and doesn't obey the button. (Isn't as easy to do a Boxsoft interface and use a MSX mouse instead of a Neos....) May be looking into the MSX mouse reading code gives us the key of what is wrong.

The next thing I'll do is to play with the two jumpers inside. In the schematics they say "JP1 ON:MSX  OFF:OTHERS    JP2  ON:JOYSTICK  OFF:NORMAL".

Seems the Mitsumi MP01 A01 processor is a clone of the MB88201.

Searching that code I've discovered here that the CD-I series of CD-players made by Philips and others also worked with a MSX mouse.

Turbo EPs can be adapted to the MSX mouses modifying the mouse.xr driver or writing a routine in SymbOS. The problem is: recollecting the direction bits from different rows surpasses the necessary delays on basic EPs.   

I think, in my humble opinion, a better solution would be to create a new MSX mouse port on one of the expansion bays, and then take advantage of that cheap Arduino adapters to be able to use PS2 mouses. It could have the benefit of freeing that surplus processor time that is needed to read the Neos. Another benefit can be to free the joystick port and bring homogeneity in an aspect(mouse on EP) that is new to almost all of us.

Searching some MSX schematics I've found the joystick ports are attached to the AY-3-8910, a version of the "universal" sound chip with a 8 bit input/output general purpose port.

But here isn't necessary something so complicated, only 6 bits(four directions and two buttons) in a row on a Z80 IN port and only 1 bit OUT on for Strobe signal.

I'm still trying to make work the MSX Arduino converter, but have little time...

Now I know where to adjust the X and Y increments(dividing by 4 or shifting right x 2) so the pointer doesn't jump, but still have problems with correct reading, something related with the so different delays, I think.

Also the two buttons remain stuck on the pressed position.  Every correctly read nibble the program inside the Arduino clears the direction bits involved (like the Neos effectively does, try to read Joy1 in Basic while Mouse.xr is working), leaving the buttons bits intact. But something in the routine leaves garbage in  the register involved so the "noise" is sent to the Enterprise part. Or can be the reading is wrong (maybe because the incorrect delays truncate the operation) and then that four bits remain with the values that where tried to send.