Okay I came back around to messing with the robot. I took some time off to write the chapters of the thesis thinking I would come back and look at the problem with a fresh mind but I guess I was wrong.
 
Still I can find no logical explanation as to why the hell I can not read from the sensor but at the same time I can write to it. Adn the problem is that I do have some kind of deadline and being stagnant like this really does not help me. Maybe I should scrap the optical sensor altogether. I am not sure. Anyone managed to interface such a sensor with the SRV?
 
To make matters worse out of the blue my source code does not compile. I get the error:
 
make: ldr.exe : command not found. Error 127
At this point of the compile : ldr.exe -f -T BF537 -c srv1-c.ldr srv1-c.bin , creating the loader.
It worked at the end when I wrote that last line of code by hand into the console. Super strange ...
I changed nothing in the makefile nor in the directory. Suffice to say that ldr.exe is already in there. I did not remove it ofcourse. Everything has to go wrong, doesn't it?

yes I did indeed install some programs since then but it is quite strange since I thought that LDR = ldr.exe in the makefile meant to search for ldr.exe in the same directory the makefile is.
 
I guess this was not the case. Which search path was altered? I changed the makefile from LDR = ldr.exe to LDR = ./ldr.exe and it worked fine. But I am not sure which path got changed.
 
Still that is not the true problem I have, this is but a small discrepancy, the real problem is this darn optical sensor. I can't see how to make the blackfin read from the sensor's registers. The code is very very simple , I have checked it again and again every day but it does not seem to be wrong.

Thanks for the advice henry but if the pin was blown up shouldn't I be unable to write to the sensor too? I can still use PH15 to write to the sensor. I am using PH15 as data dual-way line and PH14 as clock.
 
So it might still be functional but the blackfin can't handle the 5V? I will do the pin test when I get back and will try and fix it. I so so so so SO much hope that will be the thing to correct everything and allow me to read. Problem is I had the same exact problem with the Atmega8 but Atmega can handle 5V. I guess I will have to go back and try it.  
 
Thanks again for the advice. Will post back with the results.

Thanks for the encouraging words henry. I tested the pin and it seems to be okay. It can read both 3V and GND as 1 and 0 respectively. I then proceeded to add the zener as you suggested. What I got is something like this circuit :
 

 
The difference is that I have got another R1 from the diode's cathode to the 3V logic in and that R1 is 2.2K Ohm.
 
The problem now? While previously everything I read used to be FF , hence all 1 now everything I read seems to be 0. I tried to measure with a voltmeter but ofcourse the time between writes and reads are so small that the voltmeter can't catch them. I see very small fluctuations of 0 to 1.4 V , but that has to be the blackfin sending out the register's address and not the sensor responding. I will try to experiment until I get some new idea.
 
As for the sensor's ic I can not check if it functions as a mouse anymore since I have removed the mouse's microcontroller :p

You could try using a voltage level shifter such as PCA9306

Kind of a strange thesis that one. I recognized many of the diagrams coming from the ADNS2610(might be the same for the whole family of sensors) datasheet.
 
I come bearing good news! Today I managed to read from the sensor at long last. The problem is I am not 100% sure what did the trick. It can have been two things mainly. I changed many PortH pins , tried another mouse's sensor, checked the circuit of the mouse thorougly e.t.c.
 
But what I think did the trick was : Reducing the waiting time between the register's adress and the data sent by the sensor. The datasheet said at least 100us (100us min ) . And since the datasheet said so I never questioned it. I dropped it at 80us. It then worked like a charm. Could have been only this change or some changes I made on the sensor's circuit ( connecting the wires directly to the sensor IC instead of a nearby pin.
 
The 2 main registers I am now interested in are ofcourse DeltaX and DeltaY (and finding a nice way to mount the sensor on the SRV). In the datasheet it says that the register contains :
Data Type: Eight bit 2’s complement number.
 
Which means? It has a small diagram in there from which I can not make much sense. Bits 7-6-5-4 are for the negative change in X/Y and Bits 3-2-1-0 for the positive?

Oh that's two complements. I see , I see I am used to the greek definition. Yep I know that form, thanks for pointing it out spikey. Should have googled >_< ,