Image Map Image Map
Page 1 of 2 12 LastLast
Results 1 to 10 of 17

Thread: A "new" XT clone - ATX board

  1. #1

    Default A "new" XT clone - ATX board

    Some time ago I presented you my first XT clone that would fit into the original IBM 5160 case or an equivalent (mini-aT). Now I present you the same clone but in ATX format. In short it is IBM 5160 clone with some extras like up to 960 KB of static RAM, 128 KB EEPROM, a turbo mode and an onboard DS12885 RTC.

    ATX-schS.pngATX-brdS.png
    Click the image for a what larger version.

    Why one in ATX format? The most simple reason: my mini-AT cases already all have a board and I have still some mini-AT boards laying around that could need a case as well. But I still have some empty ATX cases ....

    Like before, the bigger images are available on my site:
    Schematic Board

    Q: Did I create the design from scratch?
    A: Of course not. It is based on my previous 5160clone but that one was based on IBM's original schematics.

    Q: Did I use ideas from others?
    A: The hart of the turbo circuit I copied from Sergey but in contrary to his design I used the F/C and EFI inputs of the 8284. I just hopes this works as well, time will tell.
    I reduced the number of dip switches from eight to four plus one jumper for the presence of the 8087 co-processor. Triggered by Sergey's design but I have seen it on various clone boards as well.
    I added a circuit that can handle an ATX power supply.

    Q: Did I add things of my own?
    A: First, the design of the board is my own, no question about it. Just have a look at the placement of the parts. Second, the way the second 512 KB SRAM can be used is my own design as well. The CPU will only see 64 KB of ROM but it can be swapped withe the second half of the EEPROM in flight if needed.
    ISA cards with only a RTC on board are rare. You have various MIO cards with one on board but the onboard Floppy Disk Controller most of the time cannot support 1.2 and 1.44 MB FDDs. Using an onboard RTC means I can use a newer FDC like one used by 286+ PCs and thus be able to handle 1.2 and 1.44 MB drives as well.
    I replaced all resistor arrays with single resistors. Pure out of laziness, placing single resistors near where they were needed meant less work then to draw lines form where ever to the array. I also replaced the 373s by 573s; it simplified the design of the board as well.
    I replaced two 8-bits ISA slots with 16-bits ones. The data bits D8..15 are not used of course. The main reason for having them is to support physically 16-bits VGA and FDC cards that can be accessed 8-bits wise. The extra power connections on the extra little slots won't hurt as well.
    I added the keyboard lock as promised. OK I cannot remember ever having seen an
    ATX case with a keyboard lock. But if they exist, the lock circuit is available.

    Q: 960 KB of RAM, what about that strange number?
    A: The 8088 can support up to 1 MB = 1024 KB of RAM. Minus 64 KB for the ROM makes 960 KB. I enabled the user to select parts of the second 512 KB RAM in sections of 32 KB. Theoretically a used could select all available RAM and use the board as a video-less system.

    Q: What about the 128 KB of ROM?
    A: First, I didn't want to use an EPROM for the very simple main reason that they cannot be reprogrammed very quickly. So an EEPROM/FlashRAM was my next choice. Due to a bit of luck I have many AM29F010s and 020s at hand so the choice was made quickly.
    Second, not using dynamic RAM meant I didn't have to deal with parity errors and the circuit that was needed for them. That freed output PB4 and I decided to use it give the board capabilities to swap the ROM. There is a risk of course that a program fiddles with output PB4 and in the worst case a jumper makes sure that in such a case the user can fix his choice to one 64 KB part.
    For what can we use it? Two ideas:
    - You can use both halves to test different versions of a BIOS.
    - Load the top 8 or 16 KB of both Halves with the BIOS and use the rest ROM as ROM drive to load a DOS in a very fast way.

    And again many thanks for Sergey!

    The Eagle SCH and BRD are freeware. Just PM me using the email on my site.

    Please give me your comment, ideas and criticism. I hope to hear from you!
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

  2. #2
    Join Date
    Feb 2018
    Location
    Toronto, ON, Canada
    Posts
    44

    Default

    Is it possible to make the motherboard work with both AT and ATX cases, like the many Pentium 1 and some Pentium II boards were? (example: ASUS P2B-B)

    I'm asking because it's fun to be able to run XT in the XT case.

    Another question: ebay is full of very cheap NEC V20 that are rated up to 16MHz. How hard is it to make turbo at 16mhz? My fastest XT is 12mhz Juko, so I know it's possible to go beyond 10mhz.
    Last edited by dhau; September 14th, 2020 at 11:24 AM.

  3. #3

    Default

    I have no idea at what speed this board could run. I'm sure that you have to replace certain ICs by faster types but I for myself already would be glad if it runs at 8 MHz. If it does, then we can go looking for faster speeds.
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

  4. #4
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    2,913

    Default

    My very vague understanding is the fastest available flavor of the 8284 clock generator is the "A-1" version that's rated for 10mhz; later integrated XT chipsets, like the ones used in those Juko boards, integrated the 8284's functions into an ASIC. So going much faster than 10mhz with a completely discrete board might be a problem.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  5. #5
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    35,125
    Blog Entries
    18

    Default

    For a time I wondered how the clock would be supplied for the V20HL-16 CPU if the fastest 71011 clock generator was rated for a clock output of only 10 MHz (not true--keep reading). Then I realized that the duty cycle of the V20 is 50 percent, not 33 percent of the 8088. That makes clock generation a lot simpler--you don't have to go to an x3 frequency input, but only an x2. So a 71011-C10 part could, theoretically, generate a clock for a 20MHz V20. That we can run V20 swap-ins for 8088 without changing the 8284 clock generator is just dumb luck.

    Compare the block diagram of the D71011 with the D71084 (8284 clone).

  6. #6
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    2,913

    Default

    Quote Originally Posted by Chuck(G) View Post
    That we can run V20 swap-ins for 8088 without changing the 8284 clock generator is just dumb luck.
    Do you have a datasheet that *specifies* that 50% duty cycle? Because what's really odd is after looking through a few V20 datasheets I either find no mention of duty cycle, or they actually say in the AC specifications section that the minimum "low" clock pulse *is* longer than the minimum high pulse. For instance, this one. They don't throw out that "ideal" 33% duty cycle number that's in the Intel datasheet but it's not symmetric either.

    ... I'd find it a bit odd if NEC made the chip otherwise pin compatible but built it to choke on the clock you'd normally find in an 8088 machine.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  7. #7
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    35,125
    Blog Entries
    18

    Default

    I think the idea is that Intel does call out a 33% duty cycle as being "optimum". Provided you can stay within the datasheet clock margins (rise and fall times are critical), you could run an 8088 a 50% and still be within spec.

  8. #8

    Default

    I think I have still some work to do. I ran into this: Techniques for reliable high-speed digital circuits in the Hardware section on the forum of 6502.org. Terminators, ground plane, etc., et., etc. There was only one light in the dark for me: I checked various XT clone turbo boards and many of them were just two layers, no plane filling and no terminators. And they work at 8 MHz. So there is still hope for my designs.
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

  9. #9

    Default

    How would one doing layout in Kicad, or Eagle, check propagation delay of the traces, or even check setup/hold for signals on a bus, etc?

    I am not a pcb designer, but I know some about chip design. I think that is what you'll start doing if you work on higher speeds. I don't think you'd want to do this all by hand.

  10. #10

    Default

    I haven't given that a thought at all and also don't know how to check that. I just give it a rest and if I know more, I will improve the designs. And towards Christmas I will run the risk to produce a board and we will see how it will perform.
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

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
  •