how to repair a computer system?

Efficient or at least effective diagnosis is the key thing - after all - given a bottomless box of spare parts a trained monkey could swap things in and out until a broken machine started working again.

Before you can diagnose problems effectively you need to know the basics of the data flow in and around a machine and the bus structures - not in great detail but just how the data moves about.

Data gathering and analysis are important - what is wrong with what - A Trouble Statement. It follows that if you can accurately determine this you can also determine the things that are probably not broken (unlike boiler engineers)...

Next you develop a most probable cause - or several probable causes - it can be a wise approach to tackle these in order of speed and cheapness i.e. You may believe that an expensive or difficult to change part is the most probable cause of the fault but you may also suspect something that you can eleiminate in a few seconds at no cost - go for the easy option first because it might save time and money.

Breaking a data flow - Sometimes you can see something working at one end of a system but not working as it should at the other end - a useful technique is to examine the half-way point. In doing this you immediately eliminate half of the fault areas. Again you break the remaining half into two quarters and get ever closer to the true problem.

Form a trouble statement - develop a hypothesis of the most probable cause(s) - test your hypothesis - confirm the fault - repair the broken component - test the repair - test the whole system just in case there's something else wrong.

You should now get TOP MARKS for your school assignment...

Good luck...

Peccavi
October 2009