Simple, crude "black tech."

A recent fire of a web text, called "Apple has what black technology," one of the post-look very feeling, excerpt down:

The motherboard of a product is very mini, the component density is very high, so the interface used for motherboard testing is only mung bean size. A test cord (FPC) should be designed to test the board status on this interface during Proto, which is a technical background. Because the interface is too small, the pin pin on the cord is dense, the design tolerance is only +/-0.01 mm, that is, the size is larger, smaller or position 0.01 mm, this line can not be used, and even the board directly short-circuit burned. Test lines are generally prepared by the foundries themselves. Generation factory to find a lot of FPC manufacturers to do this test line, each manufacturer feedback said can not do, precision can't control so high. Factory no way, to find Apple to report the situation, asked whether the test port can be bigger, Apple said you leave it, this line we do it ourselves. more than 10 days later, two qualified test lines were sent to the foundry, where the electronics engineers were heartily impressed by Apple's black technology. As for how to do it, I understand that: Apple found an FPC manufacturer, a breath of production of 2000 lines, and then one by one on the board to try, and finally pick out the use of the total of the two root. The plan was to do 2000 more if you couldn't pick it up. The so-called real high-tech comes from simple rudeness.

Some people would think that this is what kind of high-tech ah, Apple will only be simple rough ah, but I see Apple's broad approach to solve the problem. Technical personnel often fall into the technology of the dead end, technical problems should be solved with technology, often hit badly beaten, or spent too much cost, is not worth the candle. As the saying goes Black Cat white cat, catch mouse is a good cat, the same, tube it black technology or simple rough, with the least cost to solve the problem is a good way. Many manufacturers have said that the line can not be done, Apple more than 10 days to give a qualified sample, even if the simple rough, electronic engineers should be heartily praised.

A while ago encountered a different language UTF8 codec difference problem, also used a simple rough method to solve. A simple description is to transplant an algorithm, involving a random byte[] to UTF8 string and then back to byte[],c# and Java processing results in some cases is not consistent, to find out the law of the transformation. Start by trying to chase the core code to the bottom to see how the difference is achieved, but a little bit deeper to find the difficulties. Then the direct poor lift, put all possible values in two languages than the one, the differences to find out, and then summarize the law. In fact, not too much, hundreds of millions of times than the right, the final rule only 21, successfully solved the problem.

This article is from the "Rabbit Nest" blog, please be sure to keep this source http://boytnt.blog.51cto.com/966121/1882268

Simple, crude "black tech."