Cartridge ROM/GROM image cleanup

[+mizapf]

I'm currently doing a thorough cleanup of our cartridge images for the purpose of identifying original, pristine dumps. In MAME/MESS we have currently two cartridge formats for the TI: the ZIP format from the MAME core, and the RPK format mainly for the homebrew cartridges. In the next release, the NVRAM-based cartridges like MiniMemory will also be available as ZIP, so only the newer multibank modules are missing.

 

While we can do anything we like with the RPKs, the ZIPs should contain the actual dumps, preferably verified in some reasonable way. I found that I can use the checksums of the ROMs to verify their integrity.

 

However, this endeavor revealed to me that we have a big deal of faked dumps, unfortunately. I'll have to mark those as "bad" in the next MAME release.The RPKs will not be affected, it's just the "official" software list for the ZIPs.

 

Many of the bad dumps are hacks to remove the autostart feature, as we recently saw with the Plato cartridge; the same is the case for most Scott/Foresman modules. Hopefully I can remove these hacks easily without redumping.

 

I'm posting for one special point here: Was there ever a cartridge "River Rescue", or was this another hack to push a disk-based assembly language game on an cartridge? We have a great number of such cartridgified programs, and it's not clear whether they ever lived in a plastic shell or were just used to be loaded into a GRAM card.

 

I found this at the beginning of River Rescue:

 

 

6042: ALL   >20
6044: MOVE  >001C BYTES FROM GROM@>6026 TO VDP@>02C2
604B: MOVE  >175E BYTES FROM GROM@>60A2 TO @>CFF8
6053: MOVE  >1800 BYTES FROM GROM@>8000 TO @>E756
605B: MOVE  >00B0 BYTES FROM GROM@>A000 TO @>FF56
6063: DST   >CFF8,@>8300
6067: XML   >F0

 

So the sole purpose of the GPL code is to upload the machine language program into the 32K expansion. This is not a real cartridge, is it? In this case I will throw out that cartridge from the "official" list - but I'll keep it as an RPK, of course.

 

 

Second, I'm in bad need of some information on MBX game cartridges. If you have one, like Sewermania, Terry Turtle, Space bandits, could you please open it and tell me what's inside or post a picture?

 

I noticed that the MBX cartridge type in my emulation is broken; the banking does not work as expected. There is a mapper at 6ffe, and values 0-3 can be written on it, but I don't know what is supposed to happen. Right now it switches to some second 8K bank, but this fails everywhere because there is no second 8K dump. So either it means something different, or we just lack the dump for the second bank.

[+acadiel]

This might help you with the MBX stuff:

 

http://atariage.com/forums/topic/184672-mb-cartridge-with-1k-of-ram/?hl=mbx&do=findComment&comment=2418416

[Fritz442]

Here's Sewermania...

[+mizapf]

Yes, that's indeed helpful! Good photo quality.

 

So we have

• at the left two 1024x4 SRAMs (MCM2114); this is the 1024 byte cartridge RAM, unbuffered
• some kind of ROM? What is a NCR 8342A?
• bottom edge: space for three GROMs (could be stacked, of course); leftmost position empty
• bottom right: this is the D-flipflop that is set/reset by writing to 6ffe. There is a 13-input NAND above the leftmost GROM and another NAND next to it, supposedly to decode this address

What remains is to find out where the outputs of the D-Flipflop(s) in the 74LS175 are connected to. Unfortunately I cannot look below the GROMs, and before you start soldering, maybe you have a multimeter to find out where the traces go to?

 

So we have real GROMs in the MBX cartridges.Is it true that all cartridges that have a PHM number and data in GROM space are using real GROMs? Would TI assign a PHM number to a cartridge that fakes GROMs? If this is the case (no fakes) then all cartridge images that have significant content in the blind spot of the GROMs obviously offer no original contents but are hacked versions.

 

 

Edit: Ah, just noticed that matthew already discovered some things in the linked posting.

Edited May 5, 2016 by mizapf

[Fritz442]

What remains is to find out where the outputs of the D-Flipflop(s) in the 74LS175 are connected to. Unfortunately I cannot look below the GROMs, and before you start soldering, maybe you have a multimeter to find out where the traces go to?

You just let me know what pins are in question and I'll let you know. If still needed.

Edited May 5, 2016 by Fritz442

[+mizapf]

Sure, yes, this seems to have remained an open question...

 

Here is an "xray" that I created from the above photographs using GIMP. I marked the traces that I am interested in. Could you try to find out where they are going to or coming from?

 

Thanks,

 

Michael

 

[RXB]

I'm currently doing a thorough cleanup of our cartridge images for the purpose of identifying original, pristine dumps. In MAME/MESS we have currently two cartridge formats for the TI: the ZIP format from the MAME core, and the RPK format mainly for the homebrew cartridges. In the next release, the NVRAM-based cartridges like MiniMemory will also be available as ZIP, so only the newer multibank modules are missing.

 

While we can do anything we like with the RPKs, the ZIPs should contain the actual dumps, preferably verified in some reasonable way. I found that I can use the checksums of the ROMs to verify their integrity.

 

However, this endeavor revealed to me that we have a big deal of faked dumps, unfortunately. I'll have to mark those as "bad" in the next MAME release.The RPKs will not be affected, it's just the "official" software list for the ZIPs.

 

Many of the bad dumps are hacks to remove the autostart feature, as we recently saw with the Plato cartridge; the same is the case for most Scott/Foresman modules. Hopefully I can remove these hacks easily without redumping.

 

I'm posting for one special point here: Was there ever a cartridge "River Rescue", or was this another hack to push a disk-based assembly language game on an cartridge? We have a great number of such cartridgified programs, and it's not clear whether they ever lived in a plastic shell or were just used to be loaded into a GRAM card.

 

I found this at the beginning of River Rescue:

6042: ALL   >20
6044: MOVE  >001C BYTES FROM GROM@>6026 TO VDP@>02C2
604B: MOVE  >175E BYTES FROM GROM@>60A2 TO @>CFF8
6053: MOVE  >1800 BYTES FROM GROM@>8000 TO @>E756
605B: MOVE  >00B0 BYTES FROM GROM@>A000 TO @>FF56
6063: DST   >CFF8,@>8300
6067: XML   >F0

So the sole purpose of the GPL code is to upload the machine language program into the 32K expansion. This is not a real cartridge, is it? In this case I will throw out that cartridge from the "official" list - but I'll keep it as an RPK, of course.

 

 

Second, I'm in bad need of some information on MBX game cartridges. If you have one, like Sewermania, Terry Turtle, Space bandits, could you please open it and tell me what's inside or post a picture?

 

I noticed that the MBX cartridge type in my emulation is broken; the banking does not work as expected. There is a mapper at 6ffe, and values 0-3 can be written on it, but I don't know what is supposed to happen. Right now it switches to some second 8K bank, but this fails everywhere because there is no second 8K dump. So either it means something different, or we just lack the dump for the second bank.

 

You may be right about the Assembly version of the cart as most GROM carts could run fine without any 32K expansion and this version clearly will not.

 

This appears it might be a GRAMULATOR version of the MBX carts as I think the Originals had the same feature and C.A.D.D. set up a version of the GRAMULATOR that could run the MBX carts with just a dump.

Edited May 5, 2016 by RXB

[Fritz442]

Hope this helps... I checked my Bigfoot cart and it is the same.

 

where you have the 2 dots, pins 2 & 3 don't connect to anything I can find.

Edited May 5, 2016 by Fritz442

[Fritz442]

EDIT to post #8

I found pin3...

[Flottmann1]

here some Pictures from MBX Carts

MBX.rar

Edited May 6, 2016 by Flottmann1

[Tursi]

MBX banking is working in Classic99 (or at least Terry Turtle works, I didn't try any farther.)

 

My notes state that MBX has the lower 4k fixed and banks the upper 4k. Furthermore, the last 1k of the fixed bank is RAM. As you noted, there's a latch at 0x6FFE which latches the data bits written to select the bank (non-inverted in my notes).

 

So:

 

>6000 - >6BFF - Fixed ROM, first half of the ROM in my samples.

>6C00 - >6FFD - RAM

>6FFE - >6FFF - latch register (write data, least significant bits select bank, non-inverted (probably))

>7000 - >7FFF - banked ROM

 

I don't know if the latch also shadows RAM (that is, if you read the latch, do get get back the data written there?)

Edited May 6, 2016 by Tursi

[+acadiel]

Look at the link I put above. I put my disk of MBX dumps from Michael Becker in that old thread. Those are ones he dumped and are considered proper dumps. Hopefully that helps once you get the mechanism they are using down pat.

[+mizapf]

here some Pictures from MBX Carts

 

ah, very good - the pictures for Sewermania and Superfly show configurations where the U6 socket is empty, so we can see the traces that were covered.

 

@Tursi: Concerning 6FFE, if Michael Becker was right, it is accessed in parallel to the RAM, which means the value is written to RAM and to the latch at the same time, and when reading, the RAM will deliver the value.

[Flottmann1]

here Pics from the other MBX Carts

more mbx.rar

Edited May 6, 2016 by Flottmann1

[+mizapf]

In the Bunyard book, the Extended Basic module was described in detail, and he also stated that there was another, earlier design.

 

I suspect that there once was a three-chip ROM design (3 x 4K) because the ROM dumps have a CRC16 word at the end of each 4K block. The 4K ROM has a CRC16 at 0x0FFE, and the 8K ROM has a CRC16 at 0x0FFE and another at 0x1FFE. I think this only makes sense if you originally planned to burn two separate ROMs.

 

Do we have evidence for that? Are there any older Extended Basic cartridges left, maybe the v100?

[RXB]

As far as I know they never made a version 100 Extended Basic.

 

I have the original version 110 Source Code that I use for making RXB.

 

In the notes it seem to indicate version 100 had to many errors, so I do not think they ever released it.

[+Ksarul]

I know it was released, Rich--I have a copy around here somewhere. It did have errors, so when they were found the cartridge was recalled. During the recall, it became known as the great Extended BASIC famine. . .especially as the replacement cartridges were initially in VERY short supply.

[+OLD CS1]

I had a 100 version in my set at one time. I am almost certain I got rid of it, but while I had it I tried out some of the bugs just to see them for myself.

[RXB]

Yea long list of bugs in Extended Basic Version 100.

 

Many bugs fixed in Version 110 and I crushed many more in RXB through the years.

[+mizapf]

MBX banking is working in Classic99 (or at least Terry Turtle works, I didn't try any farther.)

 

My notes state that MBX has the lower 4k fixed and banks the upper 4k. Furthermore, the last 1k of the fixed bank is RAM. As you noted, there's a latch at 0x6FFE which latches the data bits written to select the bank (non-inverted in my notes).

 

The problem is that none of the MBX cartridges that I know have a ROM image larger than 8K, so I don't understand what is actually banked. Or do we have something like 4K + 4*1K? I'll have to check the new pictures of the MBX cartridges; maybe we can find out more.

 

Or someone manages to dump such a cartridge 4 times, by writing 0, 1, 2, and 3 to 6FFE. Then we can see the effect. I don't have a TI console, neither an MBX cartridge.

[Tursi]

The Terry Turtle on the CYC CD seems to have 3 GROMs (6000, 800, A000) and 4 ROM files (an 8k one at >6000, and three more 4k files). Are you sure the dumps you are looking at are complete?

[+mizapf]

Are you sure the dumps you are looking at are complete?

 

Hmm ... obviously not. That might explain a lot.

 

I thought I had a copy of the Cyc somewhere, but I cannot find cartridges on it. Could you mail me all the MBX files? TIA.

[Tursi]

Hmm ... obviously not. That might explain a lot.

 

I thought I had a copy of the Cyc somewhere, but I cannot find cartridges on it. Could you mail me all the MBX files? TIA.

Carts are buried a bit, look in the PC99 folder. They are stored by TI part number IIRC, so you'll need to go look that up elsewhere to know what you're looking at.

 

I can not send you all the MBX files, I only ever pulled out Terry Turtle (I'm not much of an archiver ). I can send you that after work if you give me an address to send to.

[+mizapf]

Alas, all the RPKs and ZIPs on WHTech for the MBX cartridges were incomplete all the time ... I once created them from the files in the TOSEC collection and from other sources I don't remember. And those RPKs with names starting with "mb_" are actually hacked versions; when you look inside you can even find a string "** DISK VERSION BY JNB APR87 **".

 

We need to clean up the MEss. As I said, we will keep the RPKs, but I want to recreate the genuine modules which will then be selectable by the softlist.

 

OK, so if anyone has complete sets of GROM/ROM dumps for the MBX cartridges, it would be highly appreciated if you send me a copy (ti99@mizapf.de).

 

If you have the MBX cartridges and enough hardware to create a dump (run machine language programs, switch the cartridge at runtime) you can also send me a note; I can quickly set up a dump tool.

Edited May 8, 2016 by mizapf

[RXB]

I never had the GRAMULATOR with the MBX mod built in.

You need to find the people that saved them for what you need.