The most outrageous, far-out, super cool probably never to be made dream item for the TI-99/4A...

[+Ksarul]

The current mapper used by the SAMS (74LS612) is already extensible to 16MB. Very little software has even attempted to max out the 1MB space on the current boards. When you put the buffers, the logic to mimic the 32K memory space, the mapper and a pair of memory chips on a board (this is done using through-hole components to keep it a project that can be assembled by an old-school hobbyist), you end up with a board about four inches wide and seven inches long. Horizontally, it will eat a lot of your precious desk space (about as much as the PEB connector). Vertically, it will be somewhat easy to knock loose. Note that you also have to add the necessary power supply connections too. I have considered a sidecar variant, but I really haven't seen a solid reason to go that way yet. A lot of people have the PEB card--yet there is still only a little software that makes use of it (although that has been slowly changing too, so having one is starting to become a good idea).

[wierd_w]

The idea is to use a more modern SRAM. Much of the space is taken up by the sram modules, but there are more dense modules to be had these days.

 

I was thinking more along the lines of a TSOP socket on a solderable DIP. Price would go WAAAY up, but still.  Failing that, go ahead and just use tsop packages with SMT mounting like they are intended.

[+FarmerPotato]

3 hours ago, Ksarul said:

The current mapper used by the SAMS (74LS612) is already extensible to 16MB. Very little software has even attempted to max out the 1MB space on the current boards. When you put the buffers, the logic to mimic the 32K memory space, the mapper and a pair of memory chips on a board (this is done using through-hole components to keep it a project that can be assembled by an old-school hobbyist), you end up with a board about four inches wide and seven inches long. Horizontally, it will eat a lot of your precious desk space (about as much as the PEB connector). Vertically, it will be somewhat easy to knock loose. Note that you also have to add the necessary power supply connections too. I have considered a sidecar variant, but I really haven't seen a solid reason to go that way yet. A lot of people have the PEB card--yet there is still only a little software that makes use of it (although that has been slowly changing too, so having one is starting to become a good idea).

I took the 4x7" figure as a challenge.

 

Here are my rough sketches for a 1MB SAMS on the JediMatt42 connector, which fit into 100x100mm for DIP, and 70x100mm for some surface mount.

The logic chips are condensed into one ATF22LV10C. This uses my favorite parts.

 

The 100x100mm is the magic figure for DirtyPCBs $2 per board.

 

I know there are some things missing - I said it was rough.

 

BOM - $25 using some SMD.

 2 PCB 
 4 LS612
 2 LVC245A x 4
10 AS6C4008 x 2
 2 ATF22LV10C
 2 2x20 Long Pin Female socket
 1 PLCC-28 socket
 2 misc
==
25

 

 

     

 

Then I found some LS612 for under $1  too good to be true in PLCC44 - and thru-hole PLCC44 sockets are easy. So here is an iteration.

 

 2 PCB 
 7 LS612FN ?
 2 LVC245A x 4
10 AS6C4008 x 2
 2 ATF22LV10C
 2 2x20 Long Pin Female socket
 1 PLCC-28 socket
 1 PLCC-44 socket
 2 misc
==
29

 

 

 

 

[Tursi]

11 hours ago, pixelpedant said:

 

I guess on the flipside (of cost and complexity), I feel like the best possible solution for cheap mass distribution of a two button joystick option would be a very simple, external 5V powered (micro-USB probably) Genesis controller adapter, with with the 5V doing nothing but keeping Select high.  Absolutely no signal processing or dealing with the mux required, since Up, Down, Left, Right, B and C are all exposed on unique pins with Select High. And Genesis controllers are abundant these days, in both cheap as dirt knockoff and high quality reproduction forms. Only gets you two buttons. But I figure it’s pretty good, for a solution which involves nothing but some traces/wires connecting two DB9 ports and a 5V source.  

  

I’m biased though.  Genesis controllers are my preference, and I use a modified Genesis controller for my own TI-99 gaming. 

Many Genesis controllers will just work on the TI, even though it doesn't provide 5v, for the four directions and one fire.  (You do need the atari adapter, of course). I used a Genesis pad for years.

 

[Asmusr]

My dream for the TI-99/4A would be an easy to install, cross-platform C/C++ (or similar) compiler with built in, seamless SAMS support for code and data. The compiler should have libraries for all common hardware including the F18A/MK2. The emphasis would be to remove restrictions and support creativity. Then my next wish would be to have some large, open source, community driven game projects using the compiler where everybody contributed with what they were best at (code, graphics, sound, story).

[cbmeeks]

@FarmerPotato

 

Have you considered using either an ATF1504 or ATF1508 CPLD?  They are still produced, 5V tolerant and come in PLCC-44 or 88 pin packages (easy to hand solder).

 

I'm no HDL expert, but I was wondering if you used the ATF1508 (64 I/O) you might be able to remove your GAL and the 245 buffers.

[+FarmerPotato]

54 minutes ago, cbmeeks said:

@FarmerPotato

 

Have you considered using either an ATF1504 or ATF1508 CPLD?  They are still produced, 5V tolerant and come in PLCC-44 or 88 pin packages (easy to hand solder).

 

I'm no HDL expert, but I was wondering if you used the ATF1508 (64 I/O) you might be able to remove your GAL and the 245 buffers.

Seems I wrote a more detailed reply than I intended. And it's tough reality, not in the spirit of this "I wish for" thread.
 
When choosing a PLD for my various projects, I looked at the whole ATF family.

 

My goal was to reduce chip count without raising cost. Hence, the $1.33 22LV10.

 

The ATF750 and ATF1500 are overkill for simple projects.  It is a waste to use a 1500 as a line buffer (you would consume its 32 I/Os in pairs.) The 4A has 16 address lines, 8 data lines, and 6 control lines to buffer. The 1504 barely covers this.

 

Using more PLDs removes the product ever further from "I can see what it does by looking at it" principle.

 

In favor of the 1500:

 

1. Available for $3
2. PLCC44
3. Elegant capabilities and registers. It is a pretty good match at being a logic chip (in the SAMS it would replace a bunch of chips.)
4. The 1504 is an upgrade to it for a few cents more.

 

The drawbacks of the 1500 are:

 

1. Requires a $300 investment in a programmer to use the 1500. It is not supported by the cheap TL866.
2. Horrible programming tools stuck in the Windows 95 era and never debugged.

 

So, the 1500 barely makes sense, if 1 or 2 22LV10s can do the job. The 1504 looks better but has those drawbacks too.

The 1508 has all the above and more drawbacks:

 

1. High cost ($12) compared to what you get. MachXO2 and ICE40HX CPLDs for the money give 10X to 100X the capability, also replacing the LS612 in software. Overkill for the job.
2. Monster size. PLCC84 is the smallest. 

 

The only advantage the 1508 has over chips that cost less than $12 is being more 5V safe.

 

[Omega-TI]

One Idea I've had for a number of years that I never mentioned, as it's not practical for everyday use... a Cybiko-type RF node sidecar for the TI.  A few years ago, Corey demonstrated a two player tank game at Fest West that got me thinking about it again, then later the FinalGROM was released and now people are having more and more TI get-togethers, so a device like this might be fun.

 

Basically what this device would do is just link two, possibly more TI's together that are within the same general proximity (a few hundred feet) over a radio frequency.  There would need to be a TI based setup program, (possibly within a game).  Imagine a little device the size of a TIPI connected wirelessly with another TI to play a head-to-head games.

 

I have no clue how much something like this would cost to develop and sell, and it would not get as much use as our other other stuff (except at get togethers), so it probably still falls under the 'never gonna happen' column... but I think it would be fun.  In the end, it would not even have to be a sidecar, heck a cartridge device with a cartridge interface for a FinalGROM might be able to do the job just as well.  

 

 

 

[+adamantyr]

I'd be happy with a sidecar SAMS device, possibly integrated with TI-PI.

[+adamantyr]

9 hours ago, Asmusr said:

My dream for the TI-99/4A would be an easy to install, cross-platform C/C++ (or similar) compiler with built in, seamless SAMS support for code and data. The compiler should have libraries for all common hardware including the F18A/MK2. The emphasis would be to remove restrictions and support creativity. Then my next wish would be to have some large, open source, community driven game projects using the compiler where everybody contributed with what they were best at (code, graphics, sound, story).

It would be nice to have this indeed!

[+FarmerPotato]

35 minutes ago, --- Ω --- said:

One Idea I've had for a number of years that I never mentioned, as it's not practical for everyday use... a Cybiko-type RF node sidecar for the TI.  A few years ago, Corey demonstrated a two player tank game at Fest West that got me thinking about it again, then later the FinalGROM was released and now people are having more and more TI get-togethers, so a device like this might be fun.

 

Basically what this device would do is just link two, possibly more TI's together that are within the same general proximity (a few hundred feet) over a radio frequency.  There would need to be a TI based setup program, (possibly within a game).  Imagine a little device the size of a TIPI connected wirelessly with another TI to play a head-to-head games.

 

I have no clue how much something like this would cost to develop and sell, and it would not get as much use as our other other stuff (except at get togethers), so it probably still falls under the 'never gonna happen' column... but I think it would be fun.  In the end, it would not even have to be a sidecar, heck a cartridge device with a cartridge interface for a FinalGROM might be able to do the job just as well.  

That's a good niche for TIPI and wifi.

[pixelpedant]

11 hours ago, Tursi said:

Many Genesis controllers will just work on the TI, even though it doesn't provide 5v, for the four directions and one fire.  (You do need the atari adapter, of course). I used a Genesis pad for years.

 

I've got five original era Genesis controller models myself (two Hori, three Sega) and the only one which works for me with the Wico adapter is a three button Sega pad from which I've simply removed the mux and wired the contacts directly as desired (which I originally did for the sake of controller-modding my Game Gear, but it does the job here too).  Of course, all of these work correctly with my 2600, since that provides 5V on pin 7.  And a Master System pad also works correctly without modification (since there's no mux there).  But it seems unsurprising that the two Hori controllers (Hori Fighting Stick Multi, Hori Mega Commander) and two Sega pads (Sega MK-1653, Sega MK-1470) do not function, without 5V provided to Select. 

[+FarmerPotato]

43 minutes ago, pixelpedant said:

I've got five original era Genesis controller models myself (two Hori, three Sega) and the only one which works for me with the Wico adapter is a three button Sega pad from which I've simply removed the mux and wired the contacts directly as desired (which I originally did for the sake of controller-modding my Game Gear, but it does the job here too).  Of course, all of these work correctly with my 2600, since that provides 5V on pin 7.  And a Master System pad also works correctly without modification (since there's no mux there).  But it seems unsurprising that the two Hori controllers (Hori Fighting Stick Multi, Hori Mega Commander) and two Sega pads (Sega MK-1653, Sega MK-1470) do not function, without 5V provided to Select. 

 

 

I don't have a Genesis, but I have Saturn. So I looked up the schematic for those. https://gamesx.com/controldata/saturn.htm

It has two 74153 chips to select 4 of 16 inputs. https://gamesx.com/grafx/saturn.gif

My understanding is the TI joystick port can output -0.5 (active) or +5V (idle) on one select line at a time.

http://www.unige.ch/medecine/nouspikel/ti99/joystick.htm   

 

Pins

 

TI Saturn
NC 1 VCC
8  2 D1 Down, C
3  3 D0 Up, B
7  4 S0 (TI Joy#1 active)
2  5 S1 (TI Joy#2 active)
NC 6 presence detector
9  7 D3 Right, Start
5  8 D2 Left, A
?  9 GND

I don't see where you get a GND from on the joystick port. I guess this kills the idea. At least on the Genesis it's possible to ignore the mutiplexer chips and still get usable functions.

 

For VCC I think you could charge up a capacitor+Schottky diode from TI pin 7 or 2. This avoids an external +5V. Then with the two Joy Active lines, you could select either of the rows ABCSt or UpDnLtRt. But not XYZR (can't activate both outputs at once). With no joystick select, you get the Left trigger. In all you get joy + 3 fire buttons + the Start button. Maybe the left trigger.

 

SS Reads
00 Z Y X R
10 B C A St
01 UpDnLtRt
11 - - - L

The 4A would need custom code to read the buttons. By default, it would work as a 4 way joystick with no fire button, but the buttons would read as directions on joystick#2.

 

One area I don't have my head around is what happens when a joystick has a chip that actively drives an input? The 4A expects passive switches. The L trigger could interfere with keyboard scan - probably acceptable.

 

But as I said, this seems dead if there is no GND pin, the multiplexers would go haywire.

 

 

[Omega-TI]

On 10/16/2019 at 2:44 AM, TheMole said:

I also added support for 2-button joysticks to Alex Kidd, but I don't think anyone has ever used it...

On 10/16/2019 at 9:37 AM, pixelpedant said:

Especially given that almost nobody uses the original painsticks, so a lot of people will tend to use only one joystick, as it is.  Though I dare say most people here have more than one Atari compatible joystick kicking around, and a two input Atari joystick adapter. 

It would be relatively easy to remove the painsticks and use the dual cable for a single homemade joystick, but using both button leads.

 

[cbmeeks]

On 10/17/2019 at 10:30 AM, FarmerPotato said:

 

The only advantage the 1508 has over chips that cost less than $12 is being more 5V safe.

 

Those are all valid points.

 

The very reason I am learning to use the 1504/8 is because of the 5V tolerance.  That, and I think it's big enough to put some glue and simple VGA frame buffer.  I have some FPGA's but I like the idea of a decent size CPLD that "feels" more like a vintage, custom ASIC to me at least.  Especially since you can get through hole sockets for them.

 

Anyway...I was just curious.

Edited October 18, 2019 by cbmeeks

[+FarmerPotato]

I'm posting my thoughts on this topic instead of Ksarul's thread about SID Master

https://atariage.com/forums/topic/298131-sid-master-and-all-software-packages-by-marc-hull

 

2 hours ago, jedimatt42 said:

 

I think the cause and effect is confused. 

 

Fewer developers have the thing. And the percentage of people in the 'market' that are developers is tiny. In my opinion, developers should tinker with what they want to explore, and ignore this market nonsense. 

 

I think it is the constant drumming of this maligned sentiment that influences, and creates a self-fulfillment.  

 

If you are developing and care about market, remember, tech demos don't move hardware, kill-app-games do. This has been proven over and over again. It is a shame mad-marven didn't exclusively use a SIDMaster99 sound track.  

 

And nobody should be concerned with fragmenting the community. The 4A community is already massively fragmented. As are all the pre-internet/hive-mind era computer communities. 

 

-M@

I agree with Matt, go ahead and build the thing that interests you.

I have, and still, fret over the worry "but how will I share this thing if other people don't have X hardware?" 

 

For instance, I put years into developing an RPG in the 80s and ultimately ran out of memory to put it on the 4A (I was working on the Geneve in 56K by then. Pretty small audience there.) Here's adamantyr close to finishing his RPG, and it does need more memory, so I'm going to put what effort I can toward getting people an AMS capability. (perhaps that just means soldering current SAMS boards and testing them for people!)

Suppose I were crazy enough to redesign the side-car videodisc controller, only this time for RCA Selectavision disc players, just so I can make a Rankin-Bass "The Hobbit" game that uses external video, to play on the 4A. I think I would be the only person with a 4A and a Selectavision to run it on. OK, I do think that much crazy, but I'll never get around to it. 

I also like standards. I think a DSR callable from Basic is a bedrock requirement for any gadget. I like that about the TIPI. I'm trying to write my first DSR for FORTI-2 (status update on that later.)

 

I also think a lot about whether using super modern technology is cheating on a 99/4A. Everybody can have their own view on that. I bet some people are not even concerned about it.

 

Fabrice Montupet uses only contemporary 80s parts! More power to him. I decided that using a 16MHz TI MSP430 coprocessor was too much, but adding an extra TMS9995 coprocessor was ok. Others decide that an Atmega is a good solution (UberGROM).  Matthew is putting a monster FPGA and mindbogglingly up-to-the-minute advanced technology in an F18A mk2 but we will eat it up because it does our favorite video chip! (well, assuming the 9918A is your favorite.) A Pi3 way outclasses the 4A, but, Pi is cheap and serves some big needs, so... and TIPI is well-engineered to become a "standard". Some people are content with emulation (I'm not, but I still use it all the time.) 

 

 

So, go ahead and build the thing that interests you.

 

[+FarmerPotato]

This thread needs this sound track!

 

 

 

[+FarmerPotato]

On 10/18/2019 at 1:33 PM, cbmeeks said:

Those are all valid points.

 

The very reason I am learning to use the 1504/8 is because of the 5V tolerance.  That, and I think it's big enough to put some glue and simple VGA frame buffer.  I have some FPGA's but I like the idea of a decent size CPLD that "feels" more like a vintage, custom ASIC to me at least.  Especially since you can get through hole sockets for them.

 

Anyway...I was just curious.

 

I think I was too harsh on the 1500s. It's obviously meeting your needs! Would you mind sharing which tools you're using? (compiler, programmer)

 

I wonder if it could replace an LS612 memory mapper? Needs at minimum 8 x 8 bit registers to be useful. I guess the 1508 has 128 flip flops (1 per macrocell?)
 

 

 

[cbmeeks]

Well, I've only tinkered with it a couple times using WinCUPL.  Then, we packed up the entire house and moved to another location.  So right now, I'm un-packing my house again.  ?

 

 

[+FarmerPotato]

On 10/13/2019 at 4:13 PM, pixelpedant said:

And I feel like even just distributing a blank card makes plenty of sense, given keyboard building is a pretty popular, well-supplied, and well-documented hobby these days.

 

I suppose that given ideally you'd have keycaps with all relevant TI functions on them (e.g.., a -/ key rather than a \| key, an alpha lock key, the correct symbols on the lower faces of the caps) rather than just semi-analogous IBM 101 keys, it would make sense to see if custom keycaps can be sourced, though.  And if it gets to the point where you're sourcing the caps and printing the board, you're close to a complete kit anyways.

 

In any case, I'd buy anything from a blank card to a complete kit in a second.  Though it'd be nice if the card came with a connector.

 

It seems like there are on demand custom keycap printing services out there which would work for the individual installer, if accurate keys were desired.  WASD Keyboards (art, text) and Max Keyboards (art), for example.

 

One way to economise on custom keys might be to just buy standard black cherry mx keycaps for the great majority of keys whose labels are identical to those on an IBM 101 keyboard, and only do custom keys for the exceptions, and any keys for which non-standard dimensions (e.g., a "skinny" shift or ctrl key) are required. 

 

And I suppose given that the symbols on the lower faces aren't actually actively touched during keyboard use, it'd be easy enough to replace those with self-printed white-on-black plastic labels, and not have them look half bad or interfere with the keyboard feel. 

 

I saw this keyboard build on Hackaday. 

 

https://hackaday.com/2019/11/18/sick-mechanical-keyboard-is-3d-printed/

 

You may have seen kits upwards of $250+ to build a CherryMX mechanical keyboard. This one shaves costs everywhere it can, to get under $50, plus $10 to infinity for your desired keycaps.

 

It's all handwired, so you might as well wire it like a TI keyboard.

 

I haven't found a source for a Caps Lock on/off switch, or a simple FCTN key (I think FN exists) but you can buy an ANY KEY. Also, if you were customizing, I think you could add F1-F12 keys that combined FCTN and 1 to =.  But I'm not sure having a one-keypress QUIT is the greatest idea.

 

Fun!
 

 

[Elia Spallanzani fdt]

we would like a small editor to change the backgrounds and sprites of rasmus games, quite simple even for those who can only use the basic.

[Mehridian Sanders]

Just a thought here, completely unresearched, does the joystick utilize the same controller as the keyboard? I remember reading about the alpha lock and joystick issue. If it does... Couldn't a button input show "button1+shift" allowing for an alternate input from the controller? Maybe outlandish or unfeasible.. Maybe impossible having to rewrite the joystick code to make such an allowance. Just a whimsical thought to spur ideas.

 

Sent from my SM-G955U using Tapatalk

 

 

 

 

[+FarmerPotato]

6 hours ago, Mehridian Sanders said:

Just a thought here, completely unresearched, does the joystick utilize the same controller as the keyboard? I remember reading about the alpha lock and joystick issue. If it does... Couldn't a button input show "button1+shift" allowing for an alternate input from the controller? Maybe outlandish or unfeasible.. Maybe impossible having to rewrite the joystick code to make such an allowance. Just a whimsical thought to spur ideas.

 

Sent from my SM-G955U using Tapatalk

 

 

 

 

Good idea, but:

 

The joystick port only has pins for the joystick functions. It can't report on the Shift line.

You do have joystick 2 fire button and the other directions to play with, so one augmented controller could be done that way.

 

Here are all the joystick details: 

http://www.unige.ch/medecine/nouspikel/ti99/keyboard.htm

http://www.unige.ch/medecine/nouspikel/ti99/pinouts.htm#Joystick

http://www.unige.ch/medecine/nouspikel/ti99/joystick.htm

 

In particular see the keyboard grid of row/column scanning. When a key/joy is pressed, there is a connection at a row-column intersection. The console activates each of the column lines and checks which rows it is connected to. The two joysticks have one column each. But there are three unused row intersections there! (As Thierry observes, it's a shame that the Up was placed on the same row as Alpha Lock when there were three more unused rows.)

 

What could be done with those, along with two unused pins on the port, is build up 2 more joystick row readouts: extra buttons, a paddle knob with 2 bit quadrature! 

 

Also, in Rasmus games AND who knows how many others, the joystick is scanned directly, testing for each row bit. They don't use the console ROM KSCAN.

 

 

 

[unhuman]

Better / External power supply rather than the hot thing under the cartridge port.

[pixelpedant]

Now that this is bumped back to the top: my top wishlist item right now? 

 

A low profile dedicated RS232 sidecar usable in combination with JediMatt's 32K sidecar (and consequently, TIPI), featuring full legacy compatibility.

 

Until such a thing exists, I think NanoPEB will always be my default expansion preference, as I use RS232 all the time.  So I'm hoping that comes along one day.