bladamson
Veteran Member
Positive pressure on the inside of the enclosure.
Keeps the magic blue smoke in the chips. :3
Keeps the magic blue smoke in the chips. :3
- VCF South West - June 14 - 16, Davidson-Gundy Alumni Center at University of Texas at Dallas
- VCF West - Aug 2 - 3, Computer History Museum, Mountain View, CA
- VCF Midwest - Sept 7 - 8 2024, Schaumburg, IL
- VCF SoCal - Mid February 2025, Location TBD, Southern CA
- VCF East - April 2025, Infoage Museum, Wall NJ
-
Please review our updated Terms and Rules here
I wish to create a new DMA/RAM expansion card for the Tandy 1000 line.
- Thread starter blackepyon
- Start date
My first prototype of my 3-in-1 board arrived yesterday, and I quickly assembled one. It turns out that the footprint for my CF card slot was completely wrong for the part, so I did not assemble the XT-IDE portion of the board. I also apparently put the PLUS connector 2mm too far to the left, shifting the whole board to the right, so I needed to trim the board slightly to make it fit... but hey, that is what prototypes are for.
But, in the meantime, the SRAM and RS232 portions of the board seem to work well, and I am currently working on optimizing the layout to be a little less of a jumble.
blackepyon
Veteran Member
So, regarding the original intent of this post, I have a T512CLK-A1 here that I still intend to finish reverse engineering for it's DMA logic, which is entirely TTL.
I think we're all satisfied with the 00/32 RAM logic at this point, and people are building their own cards based on that, but once I finish up my current project, is there interest in having DMA function as well for "insane" EX/HX builds or is DMA just a curiosity for you guys? The RAM portion of the T512CLK is inconsequential at this point, and we've no need for DMA-0 since we're using static RAM, so it's basically down to whether or not you want the extra floppy performance and to run a sound card, scanner card, or something-whatever that can actually take advantage of DMA.
And then with the other DMA schematics (Sergey's 8088, the IBM 5150 technical reference manual, etc), we should be able to come up with a useful DMA circuit in case somebody wants to populate that section of the board.
I think we're all satisfied with the 00/32 RAM logic at this point, and people are building their own cards based on that, but once I finish up my current project, is there interest in having DMA function as well for "insane" EX/HX builds or is DMA just a curiosity for you guys? The RAM portion of the T512CLK is inconsequential at this point, and we've no need for DMA-0 since we're using static RAM, so it's basically down to whether or not you want the extra floppy performance and to run a sound card, scanner card, or something-whatever that can actually take advantage of DMA.
And then with the other DMA schematics (Sergey's 8088, the IBM 5150 technical reference manual, etc), we should be able to come up with a useful DMA circuit in case somebody wants to populate that section of the board.
blackepyon
Veteran Member
Looks nice, but you've got the external connectors on the wrong side of the board.
This only matters if you intend to have anything on top of it.By the way, what are you guys using for UART controllers? I've got an ISA multi-I/O card that I'm using for now, but I should probably work the RS232 into another card at some point to get rid of that annoying PLUS to ISA cable arrangement.
mojorific
Experienced Member
Great work! :thumbsup:
Looks nice, but you've got the external connectors on the wrong side of the board.
By the way, what are you guys using for UART controllers? I've got an ISA multi-I/O card that I'm using for now, but I should probably work the RS232 into another card at some point to get rid of that annoying PLUS to ISA cable arrangement.
It is intended to be a single-board solution, so there would be no need of anything on top of it. I made no provisions for the bus passthrough either.. but I suppose the connector could be replaced with a stackable header.. but what would one realistically install into an EX/HX that isn't already on this?
For the UART, I am using a PLCC packaged 16C550 and GD75232 transceiver combo, with some '138s for address logic.
Eudimorphodon
Veteran Member
I'm using the PC16552DV, which is the equivalent of two 16550s Siamese-twinned into the same PLCC-44 space as the standard 16550. (It manages this by getting rid of a bunch of connections that aren't used in standard PC applications and using the freed-up lines to drive the second SN75185N transciever.) I really like having the second serial port (the latest thing I've done with it is configured it to drive an Apple Imagewriter II printer), but I don't suppose it's really a must have if all you want is a mouse for games. The 16552 is pricey from sources like Digikey, but it seems to be pretty easy to get surplus.
(FWIW, if you don't mind the connectors being just recessed enough to be annoying I'd be happy to send you one of my PCBs for the "just serial" card for the price of shipping. I didn't include a buffer on it but it seems to work fine.)
According to the tracker my next gen boards should arrive today; If everything works the set will provide:
* 384k base RAM
* 0k-128k UMB RAM (128k max for an EX, 96k for an HX)
* 32k of Flash (Located at either C0000 or C8000)
* DS1215/1315 clock/calendar
* 44 pin IDE connector
* Dual serial ports
* 1x ISA slot
I guess that makes it the "Cadillac Option".
Need to hit the local shop soon and pick up some parts so I can build up the tester. (In small quantities I can actually get 74HCT cheaper locally than Digikey; I really hope the march of progress doesn't put an end to that.)
Huh, I had forgotten about the clock chip.. my HX already has one installed. I could go back and implement it into my next prototype, but honestly, it has very limited use for games of the era.
I decided to go the Monotech XT-CF route on my next prototype, and use an IDE to CF adapter mounted "upside down" to get the CF interface. Those are cheaper than buying the sockets themselves new, and are less of a headache than surface mounting them with their 0.635mm pitch. I do not have any problems drag soldering 0.635 surface mount chips, but the CF sockets are always a pain for some reason.. It will require some modification of the backplate, but I still think it kills two birds with one stone by providing a backplate at the same time as fixing my CF socket issues.
I decided to go the Monotech XT-CF route on my next prototype, and use an IDE to CF adapter mounted "upside down" to get the CF interface. Those are cheaper than buying the sockets themselves new, and are less of a headache than surface mounting them with their 0.635mm pitch. I do not have any problems drag soldering 0.635 surface mount chips, but the CF sockets are always a pain for some reason.. It will require some modification of the backplate, but I still think it kills two birds with one stone by providing a backplate at the same time as fixing my CF socket issues.
dJOS
Veteran Member
having an onboard clock is mostly useful for folks with XT-IDE as it helps manage the contents of the hard disk (from a human pov).
Eudimorphodon
Veteran Member
I mostly use my EX for "general hackery" instead of gaming so I really appreciate having the clock. But it certainly wouldn't be necessary otherwise.
(The reason why I want an ISA riser is for a network card so... I suppose I am aiming to scratch the more technical itch.)
Hopefully there's enough people out there who need EX/HX expansion cards to absorb a few different designs, otherwise I'm going to have a bunch of nice PCB coasters lying around.
(The reason why I want an ISA riser is for a network card so... I suppose I am aiming to scratch the more technical itch.)
Hopefully there's enough people out there who need EX/HX expansion cards to absorb a few different designs, otherwise I'm going to have a bunch of nice PCB coasters lying around.
dJOS
Veteran Member
Hopefully there's enough people out there who need EX/HX expansion cards to absorb a few different designs, otherwise I'm going to have a bunch of nice PCB coasters lying around.
I'm mainly messing about with my design for fun and learning - I'm not going to muddy the waters with another product. I'm happy to keep making my various simple adapters as long as folk want them - tbh I barely have the time to build those these days. :D
bladamson
Veteran Member
Adlib, on a 90 degree ISA riser.
bladamson
Veteran Member
Hopefully there's enough people out there who need EX/HX expansion cards to absorb a few different designs, otherwise I'm going to have a bunch of nice PCB coasters lying around.
I am trying to dream up a way to design my eventual printed board such that it has an ISA edge connector (for standard PC/XT systems), but also has a spot where an optional PLUS header can be soldered in. The ISA edge connector would have a perforated spot where it could be scored with a box knife and broken off on a bench edge, to make it fit in the EX/HX after the PLUS header is added. Although with as cheap as it is to produce boards in small quantities nowadays, I wonder if such a design would really be practical.
Eudimorphodon
Veteran Member
Heh. Yeah, I mostly just imagine it'd be nice to just be able to unload a few to recover a bit of what I spent over the saga and keep them from going to waste; the idea of going all-in on making a side business of it fills me with a bit of existential dread. Inventory management for parts alone looks like a non-trivial challenge.
dJOS
Veteran Member
That's easy, just include the Plus header and utilize my Riser Card which has the correct ISA spacing:
https://www.tindie.com/products/cyberneticsys/tandy-1000-ex-hx-plus-isa-riser-card/
dJOS
Veteran Member
I totally understand I was honestly shocked how many folks wanted my adapters (still am) that I made for myself out of necessity. It might be worth your while to team up with Texelec like James did so you don't have to worry about that side of things?
Inventory mgmt is a total PITA, I suffer from that issue frequently.
bladamson
Veteran Member
Yeah, I guess you really don't lose any board space with your riser, do you? I was coming at it from the angle of "gotta cram in as much as I can in a limited footprint", heh.
Now if we could just find something that could be made into a functional and not-too-ugly PLUS bracket, we'd be all set eh? I've been known to make stuff like that with an improvised sheet-metal brake, a band saw, a drill press, and a set of taps, but I'd hate to do that for something that had to be made in quantities larger than one. D:
dJOS
Veteran Member
You only lose the small amount of space needed for the PLUS connector which isn't much.
I'm the same, I just mod ISA/PCI brackets.
Eudimorphodon
Veteran Member
The big gotchya, previously mentioned, is Plus boards *should* have their panel connectors on the opposite side from the components, unlike a standard ISA card, and that has some knock-on effects with regard to where it's kosher to place components relative to the back. Granted this doesn't matter if you're intending your board to be the only one in the machine. I've tried to make my boards technically possible to intermix with original Tandy cards, but I'll be the first to admit that it's unlikely anyone who *doesn't* have the Tandy RAM board to have a card that they'd want to reuse after upgrading their RAM.
(If someone has just one card other than the RAM card it's probably either the useless modem card or the somewhat less-useless serial card. Technically you could mix either with my cards in a stack but it'd be pretty redundant. I do have a stock of the IDE-only daughterboard if someone did have the Plus serial card, or wanted to use the ISA riser for one they already bought.
I wonder if anyone actually still has the "Digi-Mouse" Plus card and the matching mouse... and had it in a 256k machine.)
Eudimorphodon
Veteran Member
The other night I finally sat down determined to come up with a BOM for what it would cost quantity one to build up a bare boardset from Digi-Key and, quite frankly, it was pretty awful. The killers are of course the DUART and the DS1315, those together are more than $30. If you're willing to buy a batch of those from *slightly* sketchy sources you can make that part of the BOM come *way* down, like into the $4 ballpark for the same two chips... but then you're stuck asking yourself "sure, I can buy ten of these for less than the price of two from DigiKey, but will I actually ever use the dang things?!". I guess when I'm done building the prototype and, knock on wood, it checks out, I need to get off my rear and see if anyone is actually interested in buying the rest of the batch and in what form. (Bare boards, kits, completely assembled.)
I kind of feel like I can't really go to Texelec until I actually have a "finished product", but... yeah, given they sell open-source things and my plan is to open up the GIT for my boards anyway if I can actually point to a satisfied customer or two maybe they'll talk to me.