Straightening CPU bent pins and the ZIF socket

This is a special episode where we look at the issues I've had with some of the tested CPUs and some mistakes I've done while creating the clips. We start by looking at the differences between PGA and LGA type CPUs, then we look at the ZIF socket and we conclude with some methods I've used to straiten bent CPU pins.

[@taggxoc]

I would disagree with PGA being more durable, PGA actually wears out quicker than LGA with many CPU changes, even faster if the pins on the CPU aren't 100% straight (most of my PGAs died after 100-200 CPUs), LGA can take ~300 to 800+ CPU changes no problem :) As for caps on the bottom of the CPU that's no big issue in most cases, especially with 775, the CPUs will run just fine regardless, though you can notice slight increase in possible clocks putting back ripped of caps.

[@TechLab47]

Wow, 200 CPU changes on PGA and 800 on LGA, I have rookie numbers, got to pump those numbers up. Somehow I imagined that bent pins may affect the contacts in the socket and I did force a few bent pins CPUs in sockets but never had issues afterwards. I guess it also depends on the socket since for example sk 370 or sk A have exceptionally strong pins :). Taking about rookies I've seen somebody drop a LGA cpu corner first into the socket from about 10cm (these were new CPU and motherboard). Some socket pins were bent and we never managed to get them back. Needless to say everything worked fine only the last row of dimms was not visible anymore( the 4th of 4) and also there was no dual channel support. This is why I said PGA is more durable but finally I agree with you (when you know what you are doing) a LGA is better. How did you put the ripped caps from a LGA cpu back in? With hot air?

[@taggxoc]

​@@TechLab47 With forcing just slightly bent pinst into a socket you're unlikely to notice after just 1-2 CPUs, but if you do it 50+ times you'll eventually bend one of the internal pins of the socket too far to make contact, socket 370 and 462 are indeed the most robust (semi modern) PGAs :) For putting back caps on the bottom of the CPU i ise regular soldering iron as i fear hotair might heat up the core too much and degrade it. Basically what i did is to put back the 3 big caps on the bottom of a e2200, after that the CPU was still missing 1/4 of the small ones 3/4 of the arrays and one blue resistor, but just the ones i put back gained me 60MHz core clock in superpi 32M :)

[@TechLab47]

@@taggxoc That's very impressive. I've never heard someone actually testing the difference those SMD make. But I did wonder about the overclocking potential of the Pentium 4 (478) as I had a 3.2Ghz that refuses to start with more than 10FSB (210FSB for example) and it had only a few SMD components on the back (I'm not sure those were capacitors) and another P4 2.8 that had every place filled with those SMDs that was able to do 3.6Ghz on the same motherboard.

[@taggxoc]

@@TechLab47 For 478 basics are that only Prescott features resistors on the back too, all other cores only have capacitor (arrays), but what you're talking about sounds like the famous Northwood 30 cap PCB, they are known for exceptional OC performance when compared to 12 cap ones, only chips that intel only delivered with this PCB was the 3,4GHz northwood as well as all Gallatin based extreme editions. The the 30 cap pcb can be found (occasionally) on 2,8GHz 3,0 ghz and 3,2ghz 800 fsb northwoods too, there's a slim chance that others allso exist but i've never seen any.