Previous Thread
Next Thread
Print Thread
Page 34 of 76 1 2 32 33 34 35 36 75 76
Re: New dumps [Re: Darkstar] #96865
11/07/14 02:19 AM
11/07/14 02:19 AM
Joined: Oct 2006
Posts: 974
S
Stiletto Online content
Senior Member
Stiletto  Online Content
Senior Member
S
Joined: Oct 2006
Posts: 974

Re: New dumps [Re: Darkstar] #96866
11/07/14 02:37 AM
11/07/14 02:37 AM
Joined: Mar 2002
Posts: 1,066
H
hap Offline
Very Senior Member
hap  Offline
Very Senior Member
H
Joined: Mar 2002
Posts: 1,066
er, nevermind, the chip pinouts are all alike (yeah, I don't know much about TMS1000 :P)

The game looks alive in MESS in a preliminary implementation here. I can press the "R" key and see it writing patterns. I'll look more into it and post an update when I know more.

Re: New dumps [Re: Darkstar] #96868
11/07/14 04:11 AM
11/07/14 04:11 AM
Joined: May 2010
Posts: 1,049
S
seanriddle Offline
Very Senior Member
seanriddle  Offline
Very Senior Member
S
Joined: May 2010
Posts: 1,049
Yeah, I checked pin 6 several times since it looks out of place. Many of those TI chips have similar pinouts. Another difference between Comp IV and the 0972 is pin 22- N/C on the 0972 but an output on Comp IV.

I put a post-acid die shot on my web page. It's not the best pic - I overheated the die and didn't agitate the acid enough - but it should be useful.

4A0970 is on all 3 dice, but I think 4A is just a layer label- in various open areas around the die there's also 1C, 2, 3Q (?), 5A, and 6A. Comp IV's die has 0970D-04A, Code Name: Sector has 0970F-05B and Wiz-A-Tron has 0970F-07B. I think D and F indicate the die revision, -04, -05 and -07 indicate the ROM contents, and A or B are the ROM revisions.

Re: New dumps [Re: Darkstar] #96870
11/07/14 05:21 AM
11/07/14 05:21 AM
Joined: May 2010
Posts: 1,049
S
seanriddle Offline
Very Senior Member
seanriddle  Offline
Very Senior Member
S
Joined: May 2010
Posts: 1,049
Oops- the infinite loop issue I mentioned above is with the Wiz-A-Tron ROM dump, not the Comp IV ROM dump. Comp IV looks very good- it initializes RAM, sets the R outputs, then loops toggling O0, O1 and O2 to strobe the keypad. But there is no RSTR opcode anywhere in the ROM dump to reset the R outputs, which the 10 LEDs are connected to. I'm not sure if the hex value changed for RSTR or if resetting R is a by-product of some other opcode.

Re: New dumps [Re: Darkstar] #96877
11/07/14 08:17 PM
11/07/14 08:17 PM
Joined: Mar 2002
Posts: 1,066
H
hap Offline
Very Senior Member
hap  Offline
Very Senior Member
H
Joined: Mar 2002
Posts: 1,066
I've added a preliminary comp4 driver to the MESS source, see: https://github.com/mamedev/mame/commit/95a6f484e51b1b915aefe1e26bf3fb86ced62d0a

add something like this at write_o to see activity:
if (data > 2) printf("%X",data);
You will see it writing different values while you play the game. Maybe a couple of LEDs are connected here.

As for write_r, yes, something's iffy with the R outputs. It always writes the same value.

*edit* if you speak a little German, here's a good video of the toy: https://www.youtube.com/watch?v=IKtuvonAuIE


Last edited by hap; 11/07/14 08:20 PM.
Re: New dumps [Re: hap] #96879
11/08/14 12:07 AM
11/08/14 12:07 AM
Joined: May 2010
Posts: 1,049
S
seanriddle Offline
Very Senior Member
seanriddle  Offline
Very Senior Member
S
Joined: May 2010
Posts: 1,049
Cool! They "misused" the R- and O-outputs since they don't have any 7-segment LEDs. Normally each R-output would select a 7-segment digit and a keypad row, and the O-outputs would output the segments for the selected digit. The K-inputs would connect to the keypad columns. In CompIV, each of the 10 LEDs is connected to a separate R-output, and 3 bits of the O-output strobe the keypad columns. One more O-output bit is the common anode for the LEDs so it's easier to flash them all at once. The K-inputs are connected to the keypad rows.
So while it's waiting for keypad input, the O-output should cycle between 3 values as it strobes the keypad columns. Unfortunately, I can't see an obvious output PLA like the TMS-1x00 has, so I'm not sure how writes to the O-register map to the O-output bits. Since there are no 7-segment LEDs, you'd think they'd just map the 4 bits directly, 01 02 04 08, but I'm seeing 00, 01, and 02 written to the O-register.
I found the opcode decoding for SETR on the die, but not RSTR, which is odd. If I toggle the K inputs sorta randomly with the little emulator I modified from Paul Robson's TMS1100 emu, I can see memory locations change, but I never see any of the R outputs written to again after the reset routing sets them all. And the O-register data changes from 00, 01, 02 to 10, 11, 12 after I mess with the inputs, which I assume is the LED common anode.

Re: New dumps [Re: Darkstar] #96889
11/09/14 07:28 PM
11/09/14 07:28 PM
Joined: Mar 2002
Posts: 1,066
H
hap Offline
Very Senior Member
hap  Offline
Very Senior Member
H
Joined: Mar 2002
Posts: 1,066
I saw that you uploaded pinout info of the other games to your 0970 page. Did you have the romdumps too? Or do you still have to double-check them with your die images? smile

Re: New dumps [Re: hap] #96890
11/09/14 09:12 PM
11/09/14 09:12 PM
Joined: May 2010
Posts: 1,049
S
seanriddle Offline
Very Senior Member
seanriddle  Offline
Very Senior Member
S
Joined: May 2010
Posts: 1,049
I just copied the Wiz-A-Tron ROM dump out there. I have to do some more work on Code Name: Sector.

The 0970 die shots are very similar to the 0980 die shots (TI-30, TI-Programmer, etc), other than the fact that the 0980 has twice the RAM and ROM, and the ROM is 9-bit rather than 8-bit. So I've been reading one of the TI-30 patents (4064554) to see what I can pick up.

I was able to locate the output PLA on the 0970- it's just above the RAM array, rotated 90 degrees from figure 13 in the TI-30 patent. The Wiz-A-Tron output PLA is pretty typical- 0-9 map to the segments needed to display them on a 7-segment LED, ABCD map to the special LED it has for digit 3 to display plus, minus, multiply and divide symbols, E is a lower case r and F is a blank.

I'm confused by CompIV's output PLA. It only uses 4 of the O output bits - 1 for each keypad column and 1 as the common anode for the LEDs - so I was expecting a pretty simple PLA, but all 8 output bits are defined and change, and I don't understand how the 4 bits it uses work.

Since the 0980's ROM is 9-bit instead of 8-bit, they have room for new opcodes (BRANCH and CALL take up 1/2 the opcodes on both chips). The 0970 seems to be working pretty well with the TMS1000 opcodes; I think we would see more issues if much had changed.

The TI-30 patent mentions that SETR does an implicit DYN (decrement Y), but the 0970 definitely doesn't - the initialization code does an explicit DYN after each SETR. The TI-30 patent doesn't list RSTR, which jibes with what we are seeing with the CompIV code. If I'm reading it correctly, TDO clears the R outputs, presumably some time after the LED segments are lit up.

Another similarity is the extra input. The TI-30 patent mentions K3 in addition to K1, K2, K4, and K8. It's called K3 because instead of adding another bit (there's no place to put it since it is a 4-bit chip), it turns on both K1 and K2. That's fine in a calculator, where the K lines are connected to keypad columns and only one is active at a time.

The 0980 has several new features that the TMS1000 doesn't. I'm not sure how many of these are in the 0970. It seems like most of them can't be, since they require new opcodes. I need to profile the CompIV code and see what TMS1000 opcodes that it doesn't use- they might have been replaced.

I really wish there was a patent for the 0970. The TI-30 patent should make it pretty easy to get the TI-30 emulated. I've also got the TI-Programmer and I just bought a Business Analyst, both of which use the same chip.

Re: New dumps [Re: Darkstar] #96900
11/10/14 04:43 PM
11/10/14 04:43 PM
Joined: Mar 2002
Posts: 1,066
H
hap Offline
Very Senior Member
hap  Offline
Very Senior Member
H
Joined: Mar 2002
Posts: 1,066
some useful links:
list of relevant TI patents: http://www.rskey.org/~mwsebastian/patents/patent_ti.htm
codename sector patent: https://www.google.com/patents/US4171135
stop thief patent(mess/drivers/stopthie.c) http://www.google.com/patents/US4341385

By the way, I had thought MESS supported TMS0980, but it looks like it's not finished yet. It'll take a while still to emulate the TI-30 then.

Re: New dumps [Re: Darkstar] #96901
11/10/14 05:10 PM
11/10/14 05:10 PM
Joined: Mar 2002
Posts: 1,066
H
hap Offline
Very Senior Member
hap  Offline
Very Senior Member
H
Joined: Mar 2002
Posts: 1,066
I've uploaded a very preliminary driver for wizatron: https://github.com/mamedev/mame/blob/master/src/mess/drivers/wizatron.c

Hmm, all these TMS1000 based handhelds are very similar they could all fit in a single driver, but I prefer keeping them separate per manufacturer and hardware generation.

Page 34 of 76 1 2 32 33 34 35 36 75 76

Who's Online Now
4 registered members (R. Belmont, Stick, Praxis, mfm005), 153 guests, and 3 spiders.
Key: Admin, Global Mod, Mod
Shout Box
Forum Statistics
Forums9
Topics8,664
Posts113,575
Members4,848
Most Online324
Dec 20th, 2018
Powered by UBB.threads™ PHP Forum Software 7.6.1.1
(Release build 20180111)
Page Time: 0.045s Queries: 15 (0.026s) Memory: 5.7403 MB (Peak: 5.9601 MB) Zlib enabled. Server Time: 2019-07-15 18:02:42 UTC