Intellivision TutorVision found w/games

[intvnut]

With some of the 125 games?

 

Theoretically, could you take the Exec that was dumped, and populate a newly designed PCB and effectively produce more Tutorvisions? Or take existing Intellivision PCBS and install some new chips or RAM (forgive my terminology) and do it that way?

 

If you start with a standard 2609 main board and do the following, you'll end up roughly at a Tutorvision that displays the classic Mattel font instead of the Tutorvision's rounded-looking font.

• Remove the existing GRAM chips. Attach 2K x 8 of static RAM in its place. A 6116 SRAM would do nicely. The necessary signals are already present on the RO-3-9503 GROM and the STIC graphics bus.
• Make a small cartridge with 16-bit RAM mapped at $360 - $4FF, and the additional WBEXEC ROM mapped at $2000 - $2FFF.

Note: This may work on systems with an RA-3-9600, but won't work on systems that have an RA-3-9600A. As far as I can tell, the RA-3-9600 correctly maps itself into exactly $200 - $35F, and doesn't respond to requests on $360 - $3FF. The RA-3-9600A, however, actively returns garbage on the range $360 - $3FF, and that would conflict with any RAM you might try to map there. The RA-3-9600A takes its bus control lines directly from the CPU, so there's no way to intercept and disable that address range, either.

 

If you were to design a new PCB using the old parts, you could also address the RA-3-9600A issue on that design. On such a redesign, you'd need to do something to replace all the GI ROMs, though. If it were me, I'd just put a JLP-like circuit down to handle the scratch RAM and EXEC, and maybe another to handle the GROM and GRAM.

 

But, at what point down this path does it stop being an Intellivision and start being an emulation?

[mr_me]

Maybe, if you had the wbexec, split wbexec in two files and use jzintv ecs support to enable both files.

 

As it is decle's modified jzintv would be helpfull in finding out which intellivision games would have issues with a Tutorvision.

[+Tarzilla]

Calm down. I'm still at MGC. I won't be able to release anything until I'm back in California. It'll be a couple days.

 

decle's modified jzIntv won't cut it. Other changes are required. (Not many, but still.)

While you are at it, add that GRAM feature we talked about at PRGE

[JohnPCAE]

Any chance of posting a pic of the mainboard (with U9 present)? At some point we'll be asking for a schematic too

Edited April 9, 2017 by JohnPCAE

[+Lathe26]

Joe, I am glad that you are getting the chance to analyze the bare circuit board. I made an effort to aquire the board myself, both as a collector and to have the opportunity to backwards engineer the board to contribute to the community. I lost out but my biggest fear in losing was that it would end up with some collector who would just mount it on a shelf and never touch it again. I am really glad that you got the chance to investigate it.

[intvnut]

Any chance of posting a pic of the mainboard (with U9 present)? At some point we'll be asking for a schematic too

 

I'll have to dig through my email when I get back to California. I did crop a close-up, though, of Chuck's STIC1A ASIC. The chip isn't terribly interesting by itself.

 

 

Joe, I am glad that you are getting the chance to analyze the bare circuit board. I made an effort to acquire the board myself, both as a collector and to have the opportunity to backwards engineer the board to contribute to the community. I lost out but my biggest fear in losing was that it would end up with some collector who would just mount it on a shelf and never touch it again. I am really glad that you got the chance to investigate it.

 

I found out from Earl that you were also trying to acquire the board, but only after-the-fact. He was also keen on ensuring the board would end up with someone who would actually derive some knowledge-preservation value from it, rather than just mounting it as a display piece, like you said. I get the feeling I beat you solely on proximity, being able to take a Sunday afternoon to drive out to meet him. LOL.

 

I've been putting together a rather comprehensive reverse engineering document that I will be sharing publicly soon. It has everything i've learned from the PCB and other related resources.

 

It turns out that even just the TutorVision title screen needs the extended GRAM to display properly. Also, the "softer" GROM font is mostly just a pixel-perfect copy of the CGA font, with a very small number of modifications.

Edited April 10, 2017 by intvnut

[mr_me]

Is 8930 a date code or AY-3-8930.

I still wonder if this board didn't make into some late serial INTV Systems. Whoever entered it into the console5.com web site put it under "INTV System III".

[intvnut]

8930 is a date code, I'm certain of it. That bit of speculation on console5.com is likely incorrect.

 

I cover this in the reverse-engineering doc I've got written up. I'll publish that alongside jzIntv, probably tonight.

[intvnut]

OK, I finally got jzIntv updated for Windows and Mac. It turns out that my emulation wasn't complete, and I needed to make a few more changes. I discovered a few more quirks of the TutorVision system.

 

http://spatula-city.org/~im14u2c/intv/

 

Note that I have not yet updated the Linux build, as my Linux box is still offline. I hate living in this cramped apartment.

 

I've also posted extensive reverse engineering notes here. This is still a work in progress.

 

EDIT: The PDF posted in the next comment is much more readable.

Edited April 12, 2017 by intvnut

[intvnut]

OK, I finally got jzIntv updated for Windows and Mac. It turns out that my emulation wasn't complete, and I needed to make a few more changes. I discovered a few more quirks of the TutorVision system.

 

http://spatula-city.org/~im14u2c/intv/

 

Note that I have not yet updated the Linux build, as my Linux box is still offline. I hate living in this cramped apartment.

 

I've also posted extensive reverse engineering notes here. This is still a work in progress.

 

Hrm. The "publish to web" option in Google Docs doesn't preserve some of the layout, making a bit of a hash of some sections. Attached is a PDF version that's a little friendlier to read.

TutorVision_INTV88_Reverse_Engineering_Notes.pdf

[decle]

OK, I finally got jzIntv updated for Windows and Mac. It turns out that my emulation wasn't complete, and I needed to make a few more changes. I discovered a few more quirks of the TutorVision system.

 

http://spatula-city.org/~im14u2c/intv/

 

Note that I have not yet updated the Linux build, as my Linux box is still offline. I hate living in this cramped apartment.

 

I've also posted extensive reverse engineering notes here. This is still a work in progress.

 

Hi Joe,

 

Firstly thanks for undertaking this work. Would you also be willing to post the updated source please? In return I can offer you a set of Linux x64 binaries of the results

 

Am I correct to think that we are waiting on Stupus, Chuck, or someone else to potentially release the WBEXEC?

 

Just downloaded the PDF, looking forward to some serious bedtime reading

 

 

Thanks

 

decle

Edited April 12, 2017 by decle

[mr_me]

 

Hrm. The "publish to web" option in Google Docs doesn't preserve some of the layout, making a bit of a hash of some sections. Attached is a PDF version that's a little friendlier to read.

Thanks.

So the sound chip is inside the stic chip?

 

"Bits 5:4 behave unlike any other PSG observed to date for values 01b, 10b" "01b, 10b: Takes volume from envelope for one envelope step and then go to silence" Does this mean Intellivision games could sound a little different?

 

Edit:

The expanded gram gets filled with graphics automatically regardless of the cartridge? So an incompatible Intellivision game displays different graphics as opposed to no graphics.

Edited April 12, 2017 by mr_me

[intvdave]

OK, I finally got jzIntv updated for Windows and Mac. It turns out that my emulation wasn't complete, and I needed to make a few more changes. I discovered a few more quirks of the TutorVision system.

 

http://spatula-city.org/~im14u2c/intv/

 

Note that I have not yet updated the Linux build, as my Linux box is still offline. I hate living in this cramped apartment.

 

I've also posted extensive reverse engineering notes here. This is still a work in progress.

 

EDIT: The PDF posted in the next comment is much more readable.

 

 

THANK YOU Joe!!!! Finally, another piece of the Intellivision history is preserved!!

[mr_me]

 

 

THANK YOU Joe!!!! Finally, another piece of the Intellivision history is preserved!!

We still need the other 15 Tutorvision games.

[intvdave]

We still need the other 15 Tutorvision games.

 

12 if we believe 14 exists. 2 is better than zero.

 

Do you know where we can get the rest of the games?

[mr_me]

Intellivision Productions and Realtime Associates have 14 of the games, found on floppy disks many years ago. I've read that 16 games were programmed, and if the Canadian version of Time Trip is counted seperately that could make it 17. Two titles remain unknown.

[+nanochess]

Nice effort!

[intvdave]

Intellivision Productions and Realtime Associates have 14 of the games, found on floppy disks many years ago. I've read that 16 games were programmed, and if the Canadian version of Time Trip is counted seperately that could make it 17. Two titles remain unknown.

 

Good luck finding the remaining 15 games. I prefer stay away from speculation and am happy that we have two games. Sure, it would be nice to get the other 12 known games to exists. Even nicer if the 3 unknown titles are found (one being the Canadian version of Time Trip)

 

According to Intellivision Productions, they say 14 (15 counting the Canadian version of Time Trip)

 

http://www.intellivisionlives.com/media/newsletters/news030715.html.

 

What are the other 2 titles that were made?

[mr_me]

Yes that article says 16 games were to be programmed but they only found and have 14 of them. You'd have to ask Keith R or David W if they have records/info on the other two title names or if they know if they were worked on.

 

Edit:

At the 2003 cge Intellivision Productions had 14 Tutorvision games available for play. http://www.intellivisionlives.com/media/newsletters/news030831.html It sounds like they might have the Canadian Time Trip as well. I wonder what kind of cartridge they used at the 2003 show. Anyone know?

Edited April 12, 2017 by mr_me

[+intvsteve]

Awesomeness!

[intvnut]

Thanks.

So the sound chip is inside the stic chip?

 

"Bits 5:4 behave unlike any other PSG observed to date for values 01b, 10b" "01b, 10b: Takes volume from envelope for one envelope step and then go to silence" Does this mean Intellivision games could sound a little different?

 

Yes, the STIC1A contains video, audio, and address decoding for the entire system.

 

As for sounding different: There are already multiple PSG variants. The short answer is "yes, but most likely sounds like the original 1978 Intellivision." The long answer probably deserves its own thread. I wish AtariAge allowed for tables, as I would insert a feature matrix of what differs between the PSG variants here.

 

Here's a quick rundown of the parameter space:

• Behavior of "period == 0".
  • Early PSGs (and STIC1A) give a maximal count for period == 0.
  • Later PSGs give a minimal count for period == 0. This appears to be the source for "thin sounds" on some Intellivision 2s.
• Count down and reload vs. count up and compare. This affects behavior when changing pitches. The difference in circuitry has actually been observed in die shots. (I posted links to INTVPROG awhile back.)
  • Early PSGs have a simple down-counter circuit that reloads when the count expires.
  • Others count up, and reload whenever the count reaches or exceeds the programmed value.
• 6 bit vs. 5 bit volume register. Three known variants here. The first two are clearly visible in aforementioned die shots.
  • Original AY-3-8914: 6 bits, with mode in bits 5:4. 00 means volume in LSBs. 01, 10, 11 means envelope volume shifted right by (11b - bits[5:4]). ie. 11b is full envelope (0 to 15), 10b is envelope/2, 01b is envelope/4. The latter two envelopes are really quiet.
  • AY-3-8914A / AY-3-8916: 5 bits, with mode in bit 5. 0 means volume in LSBs. 1 means volume comes from envelope.
  • STIC1A: 6 bits, with mode in bits 5:4. 00 and 11 mean the same as original AY-3-8914. 01 and 10 are weird, as described in the doc.
• I/O port implementation.
  • Original AY-3-8914, also AY-3-8917: Bits 7:6 of channel enable register control I/O port direction.
  • Later: Bits 7:6 do nothing; I/O port is always input.
  • STIC1A: No I/O port. Bits 7:6 always return 0.
• Behavior of unimplemented bits in registers.
  • On some PSGs, in some bus cycles, one of the I/O ports "bleeds through" on unimplemented bits. This does not happen on STIC1A.

 

The STIC1A behaves most similarly to the original AY-3-8914 for sound generation. I don't think any games rely on the behavior of bits[5:4] = 10b or 01b, since already that behaves rather different across the spectrum of PSGs.

 

 

The expanded gram gets filled with graphics automatically regardless of the cartridge? So an incompatible Intellivision game displays different graphics as opposed to no graphics.

 

Not so fast.

 

The extra GRAM is always active. It will most likely have garbage in it at power-up, although in jzIntv it is blank. Incompatible games that specify a GRAM card 64 - 255 will get whatever garbage is there, which may be all 0s and display as blank. (This is especially true in jzIntv.) So, some games, like Space Patrol, may appear to be missing many graphics. That's what happens in jzIntv.

 

The tiles loaded by WBEXEC only get loaded only if something calls the WBEXEC code, such as a TutorVision game or a homebrew. Otherwise, the WBEXEC code remains dormant.

[intvnut]

Yes that article says 16 games were to be programmed but they only found and have 14 of them. You'd have to ask Keith R or David W if they have records/info on the other two title names or if they know if they were worked on.

 

Edit:

At the 2003 cge Intellivision Productions had 14 Tutorvision games available for play. http://www.intellivisionlives.com/media/newsletters/news030831.html It sounds like they might have the Canadian Time Trip as well. I wonder what kind of cartridge they used at the 2003 show. Anyone know?

 

If the photo caption is accurate, that looks like a normal-sized brown or black Mattel cartridge in the Intellivision while they're playing. The Intellicart could have done the job (it was available at the time). Or, maybe they already had PCBs with the necessary ROMs on them.

 

I don't think they had an EPROM-based protoboard that fits a Mattel shell. The Activision EPROM protos, if I recall correctly, require the slight additional space offered by an Activision shell.

 

I can also say that's definitely not a T-Card plugged in in that photo.

[intvnut]

 

The difference in circuitry has actually been observed in die shots. (I posted links to INTVPROG awhile back.)

 

 

Ah, here's my INTVPROG post linking the die shots.

 

 

Now here's some fun. I found die shots of the AY-3-8914 and AY-3-8910. I suspect the later PSGs in the Intellivision are closer to the AY-3-8910 than the original AY-3-8914.

https://siliconpr0n.org/archive/doku.php?id=digshadow:gi:ay-3-8914

http://privatfrickler.de/blick-auf-den-chip-soundchip-general-instruments-ay-3-8910/

The second link includes a version of the die with the individual functions identified. You can see some of the modifications made between the AY-3-8914 and AY-3-8910. If you look along the top, they highlight the 5-bit registers that store the volume setting for example. If you compare to the same location on the AY-3-8914 (which, due to rotation, is in the lower right), you can see those structures have 6 bits. This definitely corresponds to the changes made between different variants of the chip.

The logic in each bit-slice is definitely different between the two, although the general floor-plan is the same. In particular, there does seem to be more logic in the AY-3-8910, again suggesting a more complex design. (The comparator for upcount-and-compare would be more expensive, for example.)

These die photos hold more definitive answers on what can/can't be done with this chip, if we care to work out the logic.

[Tempest]

This is all really interesting stuff.

[mr_me]

 

If the photo caption is accurate, that looks like a normal-sized brown or black Mattel cartridge in the Intellivision while they're playing. The Intellicart could have done the job (it was available at the time). Or, maybe they already had PCBs with the necessary ROMs on them.

 

I don't think they had an EPROM-based protoboard that fits a Mattel shell. The Activision EPROM protos, if I recall correctly, require the slight additional space offered by an Activision shell.

 

I can also say that's definitely not a T-Card plugged in in that photo.

I think it's an Intellicart. I think I can see a white cable coming out of the cartridge going to the Mac right beside it.

 

Now about the Tutorvision sound chip. When you say it sounds like the original 1978 Intellivision, is that different than the 1980 Intellivision?

 

And regarding jzintv did you put a switch to turn on/off extra gram like Decle's -G2 switch or is it only automatic?