Image Map Image Map
Page 39 of 52 FirstFirst ... 2935363738394041424349 ... LastLast
Results 381 to 390 of 512

Thread: I wish to create a new DMA/RAM expansion card for the Tandy 1000 line.

  1. #381
    Join Date
    Feb 2017
    Location
    Chilliwack, BC, Canada
    Posts
    450

    Default

    Yeah, I'd definitely recommend the '245 for the CF card section, because that thing gets very finicky, if you saw my post a few pages back on my protoboard CF-IDE.

    Speaking of Tetris, this is the PLUS version of Glitch's XT-IDE that I've been working on (I got his permission, it's open source anyways). Also adding a couple things of my own.
    Capture2.jpg
    MediaFire link here if that's too blurry

    This is still a work in progress, once I can get back to it. I was doing fine tuning and confirming my measurements, but now that I know how far the transformer from the modem card goes, I'm debating whether or not to pull the first two chips on the left inwards about .200" to clear the transformer.

    It is time consuming to "play Tetris" with a PCB, but very doable if you're patient.

    I find that you can get quite a lot into a small space by running horizontal traces on one side of the board, and vertical traces on the other. It also assumes you're not being charged by the number of drill holes, because it involves a lot of vias.


    And yes, there's two IDE headers. Same controller, but I wanted to have the option of having one end exposed to the back for an external drive, or to quickly plug in a CF card sled and transfer files to whatever is installed internally. The big exposed copper piece on the left is to be a separate piece of PCB to solder on at a right angle to form the back panel.
    My vintage systems: Tandy 1000 HX, Tandy 1100FD, Tandy 1000 RSX, and some random Pentium in a Hewitt Rand chassis...

    Some people keep a classic car in their garage. Some people keep vintage computers. The latter hobby is cheaper, usually takes less space, and is less likely to lead to a fatal accident.

  2. #382
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,537

    Default

    Quote Originally Posted by misterblack View Post
    yeah just like you said it's going to take some homework how to figure out how to share the select the 245 properly .... it'll need some gates that's for sure, so it's at least two chips.

    Would basically need to be something like SRAM CE or EPROM CE or CF_CS1 or CF_CS0 (to enable the 245) and IOW or MEMW for direction. (or the IOR/MEMR depending on which orientation the 245 is in.)
    It turns out the equations for the PAL that controls the buffer enable and direction for the '245 that sits in front of Big Blue and all the other motherboard devices is in the manual for these systems. These seem to be the important parts:

    bufenb = Imemios & romcs # memios & fdcack;
    bufdir = memr & imio & !fdcack # memr & Imio & memios
    # memr & !mio & ior
    # mio & ior
    # ior & fdcack & Imemios & Imemr;
    Unfortunately I'm not particularly solid on what the syntax is for reading logic equations in this format... but I'm sure it says something about what you need to worry about when trying to share a '245 between IO and MEM devices.

  3. Default

    Quote Originally Posted by misterblack View Post
    it's going to take some homework how to figure out how to share the select the 245 properly .... it'll need some gates that's for sure, so it's at least two chips.
    Why not just use two 245s then.. one for the CF, and one for the SRAM? I get the 245 is relatively large 74 series chip compared to some of the smaller gates, but it would certainly save the time of trying to work out the logic. I also think this may be required anyways, so that when the CF is being read.. the bus isnt trying to read erroneous data from the SRAM and vice versa.


    EDIT because I didn't see it above:
    Quote Originally Posted by Eudimorphodon View Post
    (Which essentially means if you want to buffer both functions you might as well use two '245s, because the combinatory logic will probably take just as much room.)
    Yea, what he said.
    Last edited by rkrenicki; October 23rd, 2019 at 09:09 AM.

  4. #384
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,537

    Default

    Quote Originally Posted by rkrenicki View Post
    I also think this may be required anyways, so that when the CF is being read.. the bus isnt trying to read erroneous data from the SRAM and vice versa.
    The SRAM will be tri-stated when it doesn't have chip enable so it shouldn't produce any spurious data when the CF has the select if they were behind the same buffer. That said, I pretty much agree that a second '245 will probably equal about the same board space as the logic to sort out sharing one between memory and I/O.

    Sharing the same '245 between the SRAM and the ROM portion of the circuit is easy, you just need to AND the respective chip selects together for CE and the same direction signal is valid for both.

  5. Default

    Quote Originally Posted by Eudimorphodon View Post
    The SRAM will be tri-stated when it doesn't have chip enable so it shouldn't produce any spurious data when the CF has the select if they were behind the same buffer. That said, I pretty much agree that a second '245 will probably equal about the same board space as the logic to sort out sharing one between memory and I/O.

    Sharing the same '245 between the SRAM and the ROM portion of the circuit is easy, you just need to AND the respective chip selects together for CE and the same direction signal is valid for both.
    That is quite true.. unless the SRAM was some off-brand chinese fake.. then who knows if they actually go into a tri-state. I also hadn't noticed that you had made the same point about taking up the same amount of space on the last page, so I edited my comment.

  6. #386
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,537

    Default

    Quote Originally Posted by rkrenicki View Post
    I also hadn't noticed that you had made the same point about taking up the same amount of space on the last page, so I edited my comment.
    Post/Edit temporal collisions are the best.

    Anyway, I'm pretty sure those Chinese chips *must* be tri-stating correctly because they worked in that PCjr. I took a peek at the jrIDE schematic and it looks to me like the data lines for all the memory devices are directly on the data lines pulled from the bus connector. If they "leaked" bad things would definitely be happening.

    (YET ANOTHER EDIT) It's still my gut feeling that the reason they are not working reliably in the Tandy is more of a drive strength/noise issue than "speed", per se, but the situation is still pretty wacky, especially the part where the EX and HX behave differently. I'll do some benchmarks to make *sure* when I get my HX test mule sorted out but I don't believe the HX runs any faster than the EX.
    Last edited by Eudimorphodon; October 23rd, 2019 at 09:51 AM.

  7. Default

    If you need some comparison testing on an HX feel free to call on me. Also, I got the box, thank you very much!

  8. #388
    Join Date
    Mar 2011
    Location
    Atlanta, GA, USA
    Posts
    1,497

    Default

    Quote Originally Posted by Eudimorphodon View Post
    I took a peek at the jrIDE schematic and it looks to me like the data lines for all the memory devices are directly on the data lines pulled from the bus connector.
    Yeah, 6 loads is not ideal, but the JR-IDE is really the only side-car most people run and the 245 on the MB can handle that fan-out. The parts on the JR-IDE have typically less input impedance than period parts as well.

    6 loads, yes a 245 was needed. But I don't understand the need for a 245 when there is only 1 downstream load like on the Plus serial and mem/dma boards. It just adds another 20+ ns of prop delay.
    "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. #389
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,537

    Default

    Quote Originally Posted by eeguru View Post
    But I don't understand the need for a 245 when there is only 1 downstream load like on the Plus serial and mem/dma boards. It just adds another 20+ ns of prop delay.
    I assume it's there as a "better safe than sorry" measure, which is why I decided to use it. (No implicit criticism of the jrIDE intended, BTW. It's such a compact circuit board and, as you say, it's almost certainly the only board someone's likely to bolt onto their PCjr.)

    All of the the original Radio Shack boards for the EX/HX have buffers on the data lines. (The schematics for the RAM, serial, clock/mouse, and network cards are in my copy of the EX manual; I just checked and they all have them.) Radio Shack had a reputation for being notoriously stingy in cutting out "unnecessary" parts so I can't help but think they must be there for a reason. Could the weird geometry of the "tree" expansion header arrangement just be fundamentally noisy?

    (Frankly I wonder why Radio Shack didn't just use a stacked PC-104 arrangement. That's basically what the PCjr's sidecars amount to.)

  10. #390
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,537

    Default

    Quote Originally Posted by RetroGaming Roundup View Post
    Also, I got the box, thank you very much!
    Awesome. Hopefully that documentation I threw together is enough to get things soldered together, it'll be awesome if they all "just work" for you.

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
  •