1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Making ure own cpu or mobo

Discussion in 'Hardware' started by Bboy_Jon, 26 Jan 2005.

  1. Bboy_Jon

    Bboy_Jon Minimodder

    Joined:
    28 May 2004
    Posts:
    1,390
    Likes Received:
    0
    dunno if this should be in electronics but whatever, how hard is it to make ure own mobo? (im not gonna do it cuz i suck, but ive been wanting to know for a while) and what is the cpu? is it just a lil chip/dye on a pcb with some pins and stuff? couldnt u make ure own lol
     
  2. play_boy_2000

    play_boy_2000 ^It was funny when I was 12

    Joined:
    25 Mar 2004
    Posts:
    1,559
    Likes Received:
    100
    you can make your own computer if you have alot of time, money and alot of knowlege of how eletronics work. that said, anything you could build would be about 1/20000th as powerfull as your average desktop computer.
     
  3. whypick1

    whypick1 The über-Pick

    Joined:
    17 Oct 2002
    Posts:
    2,014
    Likes Received:
    2
    Basically what play_boy said, but I'll make it a much more daunting task.

    Processors in today's computers are made up of transistors. Seems simple, but you have to consider there are MILLIONS upon MILLIONS of them, crammed in the space no bigger than a couple of postage stamps, if that. A human hair is a couple of microns thick (IIRC). Transistors in processors right now are at .13 microns. Besides trying to cram them all in such a tiny space without destroying any of them, you have to connect them together, which means you'll have to layer the transistors so that you have space for the interconnects. Uh-oh! Transistor leakage ahead! Watch out for electron tunnelling as well.

    Motherboards...not any easier. On top of the traces that you can see on the bottom and top of the motherboard, there's also another 2 to 4 layers of traces sandwiched in between. You can't just slap the traces together so that they all get to where they need to go. With some of the signals carrying data with GHzs of bandwidth, you're going to have to worry about signal timing issues. That means making sure each any every trace in a bus is exactly the same length. For 3 DDR DIMMs, that's 552 traces total, each one needing to connect to the northbridge (or processor in the case of AMD64s). Add in PCI, which has over 50 pins, AGP, which has a higher pin count, oh, and don't forget about the processor. If you think 478 pins is a lot, consider dealing with 939. At least that means the northbridge has less pins to worry about, since the memory controller has shifted from it to the processor. Oh, and there's the surface mount components. Lots of them.

    So, still want to try?
     
  4. arrowswake

    arrowswake What's a Dremel?

    Joined:
    3 May 2004
    Posts:
    112
    Likes Received:
    0
    OMG, I've made like...20 mobos and cpu's j00 guys... If j00 hav3n't, then you = teh suck.

    Just kidding.......lolz

    Anyone who can make one, even if it isn't powerful....my hat is off.
     
  5. Krikkit

    Krikkit All glory to the hypnotoad! Super Moderator

    Joined:
    21 Jan 2003
    Posts:
    23,712
    Likes Received:
    492
    :hehe:

    [consumed by laughter]Make... Your own... motherboard![/consumed]

    Seriously though, mobo's are like 8-layer PCB's with so many components on them it's unreal. Imagine just the layout for something like that. Not really possible to make a modern motherboard at home, not even for a 486DX or something as those boards were four-layer. You'd have to make it for a single/dual-sided board which is like a Spectrum Z8 or something...
     
  6. 1st time modder

    1st time modder What's a Dremel?

    Joined:
    21 Dec 2003
    Posts:
    544
    Likes Received:
    1
    he must be on some type of drug/ maybe drunk?
    questioning this because the # of posts he has
     
  7. Bboy_Jon

    Bboy_Jon Minimodder

    Joined:
    28 May 2004
    Posts:
    1,390
    Likes Received:
    0
    heh, ya i know it would be close to impossible (for me at least) but its been bugging me for a while, cuz i only know of the transistors (normal sized) that we use in our electrnocis class, didnt think they had tiny ones like in the procs

    also so the socket is called 754/939 because of how many pins it has? ooooooooo i just thought it was a random number that didnt have anything to do with the pin count
     
  8. RotoSequence

    RotoSequence Lazy Lurker

    Joined:
    6 Jan 2004
    Posts:
    4,588
    Likes Received:
    7
    With the right equipment and access to chipset components, making a motherboard is actually quite feasible - I have a friend at my school who does it. However, making anything that remotely resembles todays computers in construction methodology without millions of dollars invested in designing an extremely complex architecture and central processing unit design is simply not possible.
     
  9. whypick1

    whypick1 The über-Pick

    Joined:
    17 Oct 2002
    Posts:
    2,014
    Likes Received:
    2
    Well, considering you wouldn't be able to actually *see*, let alone work with a transistor like the ones used in processors, I can see why.

    Nope, haven't used random numbers since the Slot-based processors (that'd be Pentium 2, early Pentium 3s, original Athlons and early T-birds) went bye-bye. After that, it's been Socket370 (later Pentium 3s and Celerons), Socket462 (also called SocketA, basically all AMD chips from Thunderbird on), 423 (early Pentium 4s), 478 (current Pentium 4s), 603 (Xeons) 754 (Athlon 64), 775 (latest Pentium 4s), 939 (latest Athlon 64s and Opterons) and 940 (first Athlon 64 FXs and Opterons).
     
  10. Zephyr

    Zephyr Go V-Boy, Go!

    Joined:
    1 Oct 2004
    Posts:
    2,024
    Likes Received:
    1
    I actually remember somebody, maybe on Bit?, making a motherboard/processor on his own. It took up an entire table, had tons of huge components on it, cost like over $300, and was around 3mhz. But, he made his own computer :hehe:.

    Just to put this in comparison, the processor in my calculator (a TI-89 Titanium) is something like 12Mhz...in a calculator...

    It would undoubtedly be interesting, but not very economical, easy, or quick.
     
  11. Bboy_Jon

    Bboy_Jon Minimodder

    Joined:
    28 May 2004
    Posts:
    1,390
    Likes Received:
    0
    hehe, ya for that price might as well buy the best mobo and maybe a cpu :D
     
  12. ChromeX

    ChromeX Minimodder

    Joined:
    12 Aug 2004
    Posts:
    1,606
    Likes Received:
    22
    :jawdrop: :wallbash: Man thats the sort of thing i'd expect my girlfriend to say! Not a Bit-tech resident!
     
  13. Altron

    Altron Minimodder

    Joined:
    12 Dec 2002
    Posts:
    3,186
    Likes Received:
    61
    The companies that make them use very precise and expensive equipment, in sealed off clean rooms in bunny suits. The amount those rooms are sealed off is like that of a place where radioactive materials are handled.

    To actually design a processor takes hundreds of people who know a hell of a lot more about electronics than you or me. They went to college to learn how to design processors. They build heavily on older processors. What you can do with a few transistors is make a machine with slightly more computing power than a power cord.
     
  14. Zephyr

    Zephyr Go V-Boy, Go!

    Joined:
    1 Oct 2004
    Posts:
    2,024
    Likes Received:
    1
    lmfao..I have to say that made me laugh a ton.

    But yeah, there is absolutely no way you can make a processor anywhere near as powerful as today's processors..hell, it would be probably be impossible to make a 386 processor.
     
  15. RotoSequence

    RotoSequence Lazy Lurker

    Joined:
    6 Jan 2004
    Posts:
    4,588
    Likes Received:
    7
    It took the entire Intel design team to create their first processor: the Z80, and they had experience making memory. Even a Z80 is likely out of the reach of anyone here without years of experience in the industry.
     
  16. Xiachunyi

    Xiachunyi What's a Dremel?

    Joined:
    3 Nov 2003
    Posts:
    1,404
    Likes Received:
    0
  17. Altron

    Altron Minimodder

    Joined:
    12 Dec 2002
    Posts:
    3,186
    Likes Received:
    61
    The Z80 was made by some engineers who left Intel, the first Intel processot was the 4004. I read the book about Intel.
     
  18. RobbieGSM

    RobbieGSM Banned

    Joined:
    26 Nov 2003
    Posts:
    1,616
    Likes Received:
    4
    Have you ever seen motherboards being made bud?

    Telling you, it's amazing to watch. I was watching a video, of the company, Abit, making their NF7S V2.0 boards. And wow! :jawdrop: .

    They have mechinces shooting, their compontents into the PCB, then hot ovens with liquid solder, which joins everything up.
    Very intresting stuff. I'd love to go there, and have alook though.

    As for making you're own, fully working board, for modern day desktops....well, i wish you the best of luck if you ever try it. :dremel:

    Rob.
     
  19. Da Dego

    Da Dego Brett Thomas

    Joined:
    17 Aug 2004
    Posts:
    3,913
    Likes Received:
    1
    I'd like to watch someone make a modern board by hand, for one. Wake me up when I'm 80. ;)

    Even designing halfway complex dual-sided pcbs are a pain, much less an 8 layer one with buried components and more traces than you have hairs. And chip technology, well, the parts ALONE for that would be incredibly expensive. But, this has all been covered already, so I'll just add the one thing that I know of the process to the pile...

    Actually designing the PHYSICAL (not the idea of the various gates) layout of the chip isn't that hard. At intel, the p4 northwood layout of those millions of transistors was developed by a college intern. We'll just ignore the fact that he had a nearly $500k program doing the majority of the hard work, which involves making sure none of those .13micro transistors gave off enough EMFs to interfere with the one next to it. That's the major issue with speed on todays chips, anyways. Sometimes the machine didn't lay the transistors juuuust right and the EMF generated interferes with one of the neighboring ones. Boom, interference just turned your 0 to a 1 or vice versa, and you just crashed your thread. Slow it down and you decrease the percentage chance that 2 transistors next to each other are both passing current at the same time, increasing the stability greatly.
     

Share This Page