Image Map Image Map
Page 1 of 2 12 LastLast
Results 1 to 10 of 18

Thread: recreation (soon to be) missing ICs

  1. #1
    Join Date
    Jan 2017
    Location
    Zürich, Switzerland
    Posts
    17

    Default recreation (soon to be) missing ICs

    Hello

    I did stumble on an older post in this forum about (soon to be) missing ICs. link

    Vintage ICs, that are no longer in production will eventually vanish, as chips will be lost, recycled or burn out over time. It will be very unfortunate and sad, when we will loose countless of the beautiful and classic ICs forever.
    I would like to discuss, if we as a community, have a chance to remake old and soon to be lost ICs. What would we actually need to accomplish this?

    To be a bit more clear, I am aware that is not possible to understand or clone e.g a Northbridge of a Pentium chipset. I am thinking more about ICs like the Commodore SID chip or simple processors like the 8008 (and many more).

    I have collected some thoughts about the idea. I believe, to start with, a good model (cycle accurate) is needed. This could even be the most important part, as this would additiolly help to preserve the function of the chip at least. I am not sure how hard it is to get a complete model, but there should be various sources available.
    • Datasheet
    • Function in system
    • Traces from logic analysers/oscilloscopes
    • Insider knowledge from people working on/with the chips


    What are your opinions or ideas concerning this topic? Is it even feasible/possible?

    I would love to discuss this topic a bit more in detail and especially get as much input as possible


    Thank you in advance and have a nice day.

    Regards,
    Thomas

  2. #2
    Join Date
    Oct 2016
    Location
    Cincinnati, Ohio USA
    Posts
    427

    Default

    I'm far from an expert on this, but it seems like a FPLA and some interface parts could probably easily duplicate any of these older chips. The hard part would be figuring out how to configure it so the inputs and outputs exactly mirror the original chip. You might even be able to make versions that are capable of running at higher clock rates than the originals. This would allow things like a mod to a C128 that enabled the 40 column screen to work when the CPU was in 2 MHz mode.
    PCjr, DTK PC-XT Turbo, 386DX 33, 486 laptop, Pentium 120, Pentium III 500, various old laptops, Commodore Colt, all working. I also have a 286 that I need to see if I can repair.

  3. #3
    Join Date
    Sep 2017
    Location
    Principality of Xeon
    Posts
    1,861
    Blog Entries
    1

    Default

    I would imagine it's emminently possible to understand a North bridge or any other modern ic. Of course limited by your means to disassemble and photo graph what's under the hood. Of course it's pretty impractical to attempt (LOL) to clone or mimic the operation of one. And what would be the point? The cost and effort, assuming it were even possible with modern fpgas, wouldn't be justified.

    But as to older ic's, I could point you to an old Circuit Cellar Ink article where the author breathes new life into an HP 41cv. Successfully, and claims it was a lot easier then he expected. From memory. But I can't point you to the article as I don't know where it is .

    I'm actually interested in participating or even starting a sig group based on this subject. I do not want to participate or much less start a group where there isn't gung ho participation. You will need to hit the books and be willing to get your hands dirty. Count on devoting at least 1 hour a week to learning (i.e reading) or experimenting hands on. For many of us that means starting at the beginning ... so it turns out there are this many valence electrons in an atom of ... get the picture? If anyone is *seriously* interested, send me a pm.

  4. #4

    Default

    You simply cannot make a SID with digital logic alone. You can approximate one, but it won't really be a SID.

  5. #5
    Join Date
    Apr 2015
    Location
    Austin, Texas
    Posts
    1,640

    Default

    Quote Originally Posted by ThommyThomaso View Post
    To be a bit more clear, I am aware that is not possible to understand or clone e.g a Northbridge of a Pentium chipset. I am thinking more about ICs like the Commodore SID chip or simple processors like the 8008 (and many more).
    It is entirely possible to clone a Pentium chipset, and it was done. There were plenty of Chinese and Taiwanese 3rd party and counterfeit chipsets. If you got the money and connections, it can happen.

    Making perfect clones of old chips is extremely difficult though because not only do you have to make them cycle accurate, you have to emulate all of the undocumented features that are known. Like the CMOS 65C02 is very different from the earlier NMOS 6502. The CMOS variant fixed bugs, removed undocumented instructions and introduced new instructions and address modes. They were supposed to be compatible on a software level, but older software making use of the undocumented features of the 6502 wouldn't work on the newer CMOS part.

    Many of these old chips are slow enough that you can emulate them in an FPGA, but getting them cycle accurate and functionally accurate is a different story.

  6. #6
    Join Date
    Sep 2017
    Location
    Principality of Xeon
    Posts
    1,861
    Blog Entries
    1

    Default

    The January 1987 BYTE is in part devoted the PALs and their programming. You'll need a lot of pals/gals to implement anything substantial I would imagine, and modernish FPGAs are very dense, some even having a microprocessor core built in. Buy you have to start somewhere. Rome wasn't built in a week. The issue is free on archive.org.

    Being one chip I've been eyeing is designated a gate array, I'm guessing/hoping emulating it's functionality should be doable. This may be the first gate array, as specified, with a gold cover. The Mindset has 2 custom "processors", 1 for graphics, 1 for sound, that are identical to the if in question. Most asics/gate arrays have a standard ceramic packaging. All of which means nothing I guess. But a high performance gate array/processor seems to indicate a good deal of integration. And more then likely the gate array does also.
    Last edited by 2icebitn; December 17th, 2018 at 06:15 PM.

  7. #7

    Default

    6502 has been implemented in FPGA for some time. 65c02 has never been out of production. Now 65EC02...

  8. #8
    Join Date
    Mar 2011
    Location
    Atlanta, GA, USA
    Posts
    1,477

    Default

    If the idea is to make form-factor DIP replacements for augmenting older/existing PCBs, things get interesting. .3" and .6" standard widths of a traditional DIP footprint does not leave much board space for an equivalent circuit - even those from high density devices like PLDs and FPGAs. Using machined Mill-Max tapered pin-headers leaves almost the full .3/.6 width. But the PCBs stand pretty high over the DIP socket. Using Batton Allen BA3760 Dill Leadframes makes a PCB DIP module that looks more like a real DIP. However it leaves even less room on the board. A .3" narrow DIP only leaves ~.25" of usable width.

    Not many packages will fit in those widths unless you go chip-scale - which is even more difficult.

    Some successful part choices I've used to make DIP modules:

    5V .6" modules: Atmel ATF1504 in a TQFP-44, 22v10 and 16v8s in TSSOP
    5V .3" modules: 16v8 in TSSOP if you are willing to drop a couple pins. And my new fav, Silego (now owned by Dialog) GreenPAK4+ devices in STQFN20 (SLG46722 and SLG46121 in particular).

    3V .6" modules: (requires level translation) Lattice iCE40up5K in QFN48, Lattice MachXO2 1200 in QFN48 and 256/640 in QFN32.

    -Alan
    "Good engineers keep thick authoritative books on their shelf. Not for their own reference, but to throw at people who ask stupid questions; hoping a small fragment of knowledge will osmotically transfer with each cranial impact." - Me

  9. #9
    Join Date
    Jan 2017
    Location
    Zürich, Switzerland
    Posts
    17

    Default

    For fully digital ICs this is certainly the case. I am currently working on implementing a Zilog Z80 like processor on a FPGA and I target a clock frequecy of about 100MHz.

  10. #10
    Join Date
    Jan 2017
    Location
    Zürich, Switzerland
    Posts
    17

    Default

    Quote Originally Posted by 2icebitn View Post
    I would imagine it's emminently possible to understand a North bridge or any other modern ic. Of course limited by your means to disassemble and photo graph what's under the hood. Of course it's pretty impractical to attempt (LOL) to clone or mimic the operation of one. And what would be the point? The cost and effort, assuming it were even possible with modern fpgas, wouldn't be justified.

    But as to older ic's, I could point you to an old Circuit Cellar Ink article where the author breathes new life into an HP 41cv. Successfully, and claims it was a lot easier then he expected. From memory. But I can't point you to the article as I don't know where it is .

    I'm actually interested in participating or even starting a sig group based on this subject. I do not want to participate or much less start a group where there isn't gung ho participation. You will need to hit the books and be willing to get your hands dirty. Count on devoting at least 1 hour a week to learning (i.e reading) or experimenting hands on. For many of us that means starting at the beginning ... so it turns out there are this many valence electrons in an atom of ... get the picture? If anyone is *seriously* interested, send me a pm.

    As I understand 'more modern' chips cannot longer be analized by taking pictures of the die, as the transistors are just too tiny and there are too may metal layers on top of the transistors, which cover them up.
    I guess the never the less feasible it will be. I guess as a rule of thumb, it should be possible with DIP packages.

    That would be great Thanks. As for memories, this should be easy to replicate with SRAM.

    I know it will be some work to invest, but it might open a whole field of interesting stuff to understand and discover. I guess you can only understand a circuit in its entirety if you have studied in detail or maybe even reverse engeneered.

    Someone needs to understand how the chips work, draw block diagramms, timing characteristics,... . This would allow to make a simulation of the chip or even a discription in a HDL (Hardware description language). As soon as we are this far the function can be implemented on a FPGA or even on a custom Chip (ASIC).

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •