• Please review our updated Terms and Rules here

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

I was a little worried I was coming across as pestering.
Not at all :smile:
I just wish I had more time to tinker.

Yes, the output of the first inverter gate drives the second. Though it did take a little bit getting used to seeing the power LED that has been steady green for 30+ years flash red, the last thing I want to do is drill a hole in the case. But like I said, I've been quite happy with how it turned out.

As for the gutter, as long as the chips are mounted straight to the board (not socketed) you should be able to line them up with pin 1 on the PLUS header.
 
As for the gutter, as long as the chips are mounted straight to the board (not socketed) you should be able to line them up with pin 1 on the PLUS header.

The modem board has few components that violate that; there's a metal isolation transformer that goes as far as the pin 2 row near the connector itself, and there's a big fat... filter? on the other side that's .4 inches past the top of the PLUS header. (The latter overhangs part of top the battery holder on my RAM board; if it got as far as the positive contact it would actually be a problem, but as is it *just* fits. It's that I think might have had a problem with the '245, at least if it sat high in the socket for some reason.) Of course, it's not like I'd be likely to ever plug that board in again if the serial card works...

So I took a deep breath and submitted both boards to be run off just now, after going through them both one final time. The CF board is so simple I want to be confident about it and the serial board is at least *conceptually* fairly simple, but I'm not counting any chickens, or pterosaurs, before they hatch.
 
The modem board has few components that violate that; there's a metal isolation transformer that goes as far as the pin 2 row near the connector itself, and there's a big fat... filter? on the other side that's .4 inches past the top of the PLUS header. (The latter overhangs part of top the battery holder on my RAM board; if it got as far as the positive contact it would actually be a problem, but as is it *just* fits. It's that I think might have had a problem with the '245, at least if it sat high in the socket for some reason.)

I realized that they violated in horizontal alignment, what I was curious about was vertical clearance. The '245 on the Tandy DMA/RAM card is closer to the gutter than yours, in line with pin 2 on the header, about just behind where the phone jacks on the modem board sit (at least so it looks from the photos), but it's soldered directly to the board. With my current setup, the PLUS-CF card sits directly below the inverted ISA I/O card, so the components face each other. The lower board has everything socketed (for prototyping, so I wasn't frying the chips with all the re-wiring), and the ISA card has jumper pins, so there's only a couple millimetres vertical clearance.

For the final product, I'd have everything just soldered straight to the board, other than RAM, ROM, and battery. Of course, if we were going for SMD, clearance would be a non-thought, but that severely limits who'd want to buy such a board as a kit to assemble themselves. For the battery, you might be able to shave off a couple millimetres by going with a surface mount battery clip, and these are very easy to solder on by hand.

Also, because I want to put the '245 next to the data lines on the PLUS header, how many pins down the PLUS header does the transformer go?
 
Last edited:
I realized that they violated in horizontal alignment, what I was curious about was vertical clearance.
....
Also, because I want to put the '245 next to the data lines on the PLUS header, how many pins down the PLUS header does the transformer go?

The transformer goes basically *all* the way down; if you hold the board edge-on to you with the PLUS connector near you and sight across it sticks up almost 1/8th of an inch above the connector. The transformer has little metal bits sticking out of the coil in each corner looking like they're begging to short on something, and they're almost the lowest point; if I rest a ruler on them and lay it across the plus connector the inner edge of the ruler *just* crosses the hole for pin one.

So, in short you're *probably* safe if you line things up so the top of pin *2* is the "top" of the circuitry on the right half of the board, but not pin 1.

Also checking with the ruler, the "filter cap" or whatever on the underside of the left side of the board does protrude as far as pin 4 on the PLUS connector, but it only extends downwards about a quarter of an inch. So even socketed ICs under it *probably* wouldn't be an issue, but a tall cap or that flying lead on the battery holder is pushing it. (Said "flying lead" on my board is about lined up with pin 5 so, yeah, it was close.)
 
Last edited:
With my current setup, the PLUS-CF card sits directly below the inverted ISA I/O card, so the components face each other. The lower board has everything socketed (for prototyping, so I wasn't frying the chips with all the re-wiring), and the ISA card has jumper pins, so there's only a couple millimetres vertical clearance.

I've been thinking about the whole ISA/Plus adapter thing; are you using one of dJOS's PCB adapter boards or a homebrew cable? I'm curious what the exact dimensional clearances are.

The reason it crosses my mind is I was vaguely wondering if there'd be any point of investigating how much of a "humpback" you'd need to add to an HX or EX in order to accommodate another card, either another Plus board or an ISA adapter, on top of the stack. After all, if you're actually turning out new-build Plus cards it would in principle be easy to offer the option of having a stacking header (IE, a long-tailed/wirewrap female connector) instead of a normal connector on a card targeted for the upper slot, thereby allowing the stack to continue right out the roof. In theory at least you could probably 3D print a replacement for the original expansion-bay hatch that'd add a "camper shell" sticking out far enough to house however many cards you want, although I'm guessing it would be a pretty expensive print. (I need to remember to ping my 3D-printer-guru friend one of these days...)

Granted there are pretty limited use cases for such a monstrous idea. Next time I have the machine open I do want to check and see if there'd be an issue with closing the original hatch if you had a card with a pass-through header plugged into the top slot. I'm *guessing* not, if only because I do know you can close it with the modem card in the top slot and I think its speaker, et al, stick up higher than adequate-length header pins would.
 
It's a home-brew cable. That was the only way to get everything to fit. I actually had the ISA I/O card installed before I built the PLUS-CF protoboard card.
IMG_20190823_170507214.jpgIMG_20190823_170528373.jpgIMG_20190823_170536159.jpgIMG_20190823_170947737.jpgIMG_20190823_173434713.jpg
The blue wire was my attempt at a shielding weave, which turned out to be barking up the wrong tree anyways. So ignore that bit of arts and crafts -a normal ribbon will work fine. The issue turned out to be that I'd placed the '245 on my protocard too far away from the bus, so it was getting too much noise with my wiring job.

It might not be a bad idea to cook up a 3d-printed box, in which one could add a few ISA cards with a small backplane. Say about the size of an external CD-ROM or two. The board would be a pretty brain-dead design, just need to add some terminating resistors to keep the noise down and maybe a 5-volt regulator, so an external power brick can be used if needed, depending on how many cards you'd like to add. Ribbons area easy to come by, but Digi-Key doesn't carry 62-pin IDC connectors, hence my lovely solder and hot-snot. They're available elsewhere on the internet though.
 
It looks like you're using exactly the same 40 pin->CF sled that I ordered for my crack at the CF board, that's reassuring. (From the chip count I'm going to guess we're using either the same or very similar circuits; my board of course lacks the ROM because it's on the RAM card.)

It might not be a bad idea to cook up a 3d-printed box, in which one could add a few ISA cards with a small backplane. Say about the size of an external CD-ROM or two. The board would be a pretty brain-dead design, just need to add some terminating resistors to keep the noise down and maybe a 5-volt regulator, so an external power brick can be used if needed, depending on how many cards you'd like to add. Ribbons area easy to come by, but Digi-Key doesn't carry 62-pin IDC connectors, hence my lovely solder and hot-snot. They're available elsewhere on the internet though.

I do wonder just how much abuse the built-in power supply can take.

If you did want to add an additional 5v from an external regulator I'm a little curious how exactly to make that electrically kosher. That idea reminds me of the expansion box that came with the Amiga 1000 I never play with; it came with this two (or was it three?) slot box that has a 2MB RAM card occupying one of the slots. The *scant* documentation I was able to dig up claims the slots are Amiga 2000 Zorro compatible but says you shouldn't use them without attaching an external power supply to the barrel connector on the back of the box... and nowhere does it say what the specifications of that power supply should be. One of these days I should take that thing apart and trace exactly how that wallwart plug is connected to what.

(Granted the only use I'd have for it would be to *maybe* try to convert the SCSI controller-based hardcard that came with the dead Amiga 2000 to work with the 1000, but my research on that card suggests success would be of a low order of probability. I should probably just sell the whole mess to someone that would actually love them more than I do.)
 
It looks like you're using exactly the same 40 pin->CF sled that I ordered for my crack at the CF board, that's reassuring. (From the chip count I'm going to guess we're using either the same or very similar circuits; my board of course lacks the ROM because it's on the RAM card.)
Well, I've had good results with this one. It's also a good size to measure against.


I do wonder just how much abuse the built-in power supply can take.
It could probably handle one or two extra cards without much issue, but if you tried to piggyback a hard drive onto it in addition to the two built-in floppies? You'd probably run into some issues. According to the datasheets, the power supply is rated for 3A on +5v, 1.25A on +12v, and 0.1A on -12v. So just need to put the multimeter to that and figure out what it's drawing.

If you did want to add an additional 5v from an external regulator I'm a little curious how exactly to make that electrically kosher

Remember when peripherals came with a grounding tag on the cable? Chassis ground is the signal ground as well (aka, common ground), and that's what the voltages are biased against. I'd probably have a switch, or perhaps a barrel jack for the power brick that has a switch built in, so you can disconnect the bus power when using the brick, but so long as the ground is common between the two supplies, I don't see any issue. Without the switch, the voltages would need to remain exactly the same, else one would be feeding into the other, so being able to separate them is the best idea.

The Tandy doesn't give out -5v, and not much uses -12v, so I wouldn't worry about those, but a +5v regulator from a 12v power brick would allow an external hard drive to be run in that case as well.

I've been thinking of getting myself a 3-D printer at some point, if nothing else, for replacing plastic parts that are hard to find (like for my dishwasher). Now I've got a good project idea for it. Probably make it large enough to house 2 or 3 8-bit ISA cards, or a 5-1/4" floppy drive, with ribbon connector knockouts so you just knock out the one for the ribbon you need.

First though, got to finish up my current projects.
 
Remember when peripherals came with a grounding tag on the cable? Chassis ground is the signal ground as well (aka, common ground), and that's what the voltages are biased against. I'd probably have a switch, or perhaps a barrel jack for the power brick that has a switch built in, so you can disconnect the bus power when using the brick, but so long as the ground is common between the two supplies, I don't see any issue. Without the switch, the voltages would need to remain exactly the same, else one would be feeding into the other, so being able to separate them is the best idea.

Maybe it really is just that simple in the Amiga box. What I could see of the circuit board containing the power jack left me scratching my head because I don't remember seeing anything that looked like power regulators, nor did I see anything that looked like switches to detect if power was getting routed in via that port. As I mentioned, the only tiny scrap of documentation I could find for the chassis (it was made by some small outfit near... Seattle, I think?) on the Amiga hardware database, a magazine ad for it, *could* have been interpreted as saying that any slots other than the one the RAM card was plugged into were effectively disabled unless the PSU was present. That would make sense if the *only* 5v supply to them was via the barrel jack and there wasn't actually any circuitry for combining/switching the inputs.)

I've been thinking of getting myself a 3-D printer at some point, if nothing else, for replacing plastic parts that are hard to find (like for my dishwasher). Now I've got a good project idea for it. Probably make it large enough to house 2 or 3 8-bit ISA cards, or a 5-1/4" floppy drive, with ribbon connector knockouts so you just knock out the one for the ribbon you need.

I was vaguely picturing something that would slide into the notches for the existing hatch lid and give the computer a kind of "camel's hump", which inside could accommodate either additional plus card(s) (by using a card with stacking headers in the top of the stock 3 slots to keep the stack going) or an ISA backplane for half-length cards; you could fit the "hump" with a backplate appropriate for either by making that a separate piece. It'd look pretty weird and keep you from stacking a monitor on top, but it would avoid needing to cable the bus. Granted I'm kind of stuck on what *I'd* actually use it for; a network card, maybe, although if I can get the dual serial card to work a serial port is *probably* good enough for my needs. But some people do want to go truly nuts. ;)

Another thought for a "hump" top might be to provide a front-accessible CF or SD slot by mounting the PCB from a suitable ISA blanking plate-mounted adapter cabled to the CF card. Something like that could be pretty low profile.
 
A "camel hump" would be useful for that 1000HX max performance build that Retrogaming Roundup is going for. I'm hesitant because for years I was that schmuck with the monitor sitting directly on top of the HX with a couple blocks at the back to keep it balanced. Built myself a hutch for it only about a year ago.
 
Speaking of monitors, I should probably open a new thread about it when I get stuck into actually worrying about it, but mucking with a monitor was yesterday's project and it's raised (well, actually renewed) concerns that the RGB output on my EX might be blown. TL;DR, I opened a thread a while back about a dead Commodore 1084 that I got running by wiring around the switch, but upon connecting it to the EX with a long skinny printer cable found that it appeared that the "Green" line of the RGB port was either not registering at all or was doing so very weakly. Yesterday I built using shielded cable cut from an old KVM cable a short "proper" CGA cable, and the problem still persists. So either the output of the EX is hosed or something's wrong with the input on the 1084. I think I have my excuse to pull out the oscilloscope, since I don't have another RGBI monitor nor another computer similarly equipped to do unit-level testing on.

(Maybe I'll shelve that emergency until hopefully my HX board arrives to play with. In a sad way I'm kind of enjoying the "Low-Fi" look of running on composite anyway.)

A look at the schematic shows that the RGBI port is driven through a set of '244s, so at least maybe if it is the RGB port it's just one of those buffers, not Big Blue.
 
Was this a serial printer cable? Check that all 9 pins are connected completely through.

Check also that you don't have corrosion in the DB-9 connector at the back of the EX. Given the age of these machines, that's not an uncommon problem, especially if you got it second hand or it was carelessly stored.

I've got a Tandy CM-5 (another project on my to-do list), currently apart (displaying a horizontal white line), while I try to figure out where the fault is along the vertical driver circuit. Given how rare these TTL monitors are becoming, it's definitely worth keeping them going as long as we can. You even see "broken" TTL monitors on eBay for "parts/repair," because to people with the know-how, they're worth fixing. Heck, with the way the prices are going up on these old machines, sometimes it's worth buying a broken computer or monitor just to avoid spending a fortune on them.

The satisfaction of fixing and restoring them is half the fun anyways :p
 
I checked the continuity through the serial cable before trying to use it originally and it checked out, so building my own cable was kind of a hail Mary. The port on the computer looks clean so... deeper investigation is indeed called for.

(The monitor was also working in analog RGB mode with the Amiga it came with and it displays green fine with composite input. What worries me about the Tandy is if I flip the 1084 to analog mode I get the too-bright 8 color display I'd expect *with a green component present* but it looks a lot weaker than the other colors. I have this suspicion that when I hook a scope to it I'm going to see a messy/malformed signal on green compared to red and blue.)
 
... I wonder if it could be as simple as a bad pullup resistor somewhere, possibly in the monitor. Need to dig up the schematics for that end.
 
Good luck finding schematics for a monitor. It's hard enough to find them for the computers themselves.

But anyways, this is getting a bit off topic. If you want to banter ideas with me regarding monitor repair, feel free to PM me about it. If there isn't a thread for monitor repair, it might not be a bad idea to start one.
 
A little tired from the adrenaline crash to waste too much time on extraneous detail, but... latest update: the CF board seems to work perfectly:

20190826_213511.jpg

The package of them and the serial boards showed up this morning, so as soon as I could knock off work around 5:00 I frantically soldered one together and stuck it in the machine. The machine *would* hang when it tried to boot from the used CF card I stuck in it, but by hitting the "A" key I was able to get into DOS 3.3, and after a quick FDISK and Format C:/s later... it was still hanging on boot. So... I grudgingly tracked down a set of 720k DOS 5 boot images and tried them plus a second CF card together, explicitly forcing an "FDISK /MBR" before repartitioning and formatting. And, there you go, it booted just fine. (And, man, these XT-CF cards boot uncannily fast...)

After a lot of mucking around in DOS 5 I tried sticking the CF I formatted with 3.3 back in, booting from the DOS 5 boot disk, and running FDISK /MBR on it, and now it's booting too with Tandy DOS 3.3 from the SL disk set. So... yay.

that-was-easy-button.png

Trying not to think about which direction this might bode for the serial card. (Which I'm waiting for RS232 transceiver chips to come in for before I can beat myself senseless against that wall.)
 
Yay!

Yeah, the CF cards need to be "cleaned" before they can be used. Something about how the modern machines format the partition table or the MBR (which ever) that XT class machines don't like.

I've got Dos 5.0 running on mine. Took me a while to figure out that since DOS 5 recognizes the HX as an XT-class machine, the setup menu only recognizes the arrow keys on the numpad, and DOSSHELL doesn't recognize the backslash key (but it works fine from the prompt). Other than that, if you need the large format, DOS 5 works just fine too.

Now I just need to get back to working on mine.
 
Last edited:
Yay!
Took me a while to figure out that since DOS 5 recognizes the HX as an XT-class machine, the setup menu only recognizes the arrow keys on the numpad, and DOSSHELL doesn't recognize the backslash key (but it works fine from the prompt).

IIRC, the left arrow works as \ on the HX in Dos 5.
 
IIRC, the left arrow works as \ on the HX in Dos 5.

I actually was in DOSSHELL briefly last night (it totally reminded me of why it earned the nickname "DOS HELL"...) and ran into the backslash thing. Bizarre thing about it is the backslash works okay in edit.

The placement of backslash and pipe are the two things that make me understand why people slag on the 1000's nonstandard keyboard. I'm mostly charmed by the differences, even the stupid "hold" key, but, man, does it need a dedicated ( | \ ) key.
 
Last edited:
Yeah, the CF cards need to be "cleaned" before they can be used. Something about how the modern machines format the partition table or the MBR (which ever) that XT class machines don't like.

Yep. I'd heard it could be an issue, but since I think the last flash-card multi-adapter I had that could do CF probably disappeared into the aether sometime around 2012 wasn't able to 'dd' them clean with zeros before installing them. "fdisk /mbr" did the trick; it's a shame that fdisk versions prior to five don't have it. (You can give the 3.3 fdisk the switch and it just ignores it; I wish it had thrown an error.) I guess back then the problem of having an existing alien format on your drive was probably a novel one given you would have been expected to have run a low-level format to match your controller before ever getting to partitioning.

(An interesting addition to the XT-IDE firmware might be a bare "disk tools" menu that'd let you "low level format", IE, zero fill, an attached device...)

Next thing on my list, I guess, while I'm sitting on my hands waiting for serial board parts is poking around to see if I can make better use of my 128k of UMBs than that Radio Shack RAMdisk driver. I don't have much in the way of device drivers to load into them, but it looks like loading part of DOS *may* be a possibility, maybe? That would make DOS 5 as a daily driver that much more attractive.

Anyway, I'm still pleased that the devious plan of sticking the Flash on the RAM card worked. Although, really, considering the CF adapter is only three chips I should revisit the possibility of integrating it *on* the memory board. I'll need to get a lot better at routing traces to get the density high enough to do it on a 100x100mm board.

Speaking of the memory board, I had a forehead-slapping realization that I could simplify it slightly (IE, take one level of gate latency out of one path) by subbing a 7411 for the 7408, and in the process make it theoretically possible to map the second page of UMB RAM at either E000 *or* C000. (Which would only really be useful in an HX without a mass storage device, granted.) Adding that to my list of "if I ever make more of these" items.
 
Back
Top