Previous Thread
Next Thread
Print Thread
#108733 - 02/05/17 07:36 PM Consider uncapping TMC0430  
Joined: May 2007
Posts: 542
mizapf Offline
Senior Member
mizapf  Offline
Senior Member

Joined: May 2007
Posts: 542
Germany
We are currently relying on behavior analysis of these ROM circuits, commonly known as GROMs. There is an abundance of those circuits around in cartridges, and many of those cartridges are not really rare enough to be of particular value. So I believe there are enough people around who would donate their cartridges.

I do not have any experience and neither the means for uncapping circuits, so my question to those among us who know more:

- Would it make sense to uncap such a GROM? Can we find out things other than the memory of 6 KiB, in particular the addressing and line control logic? The memory is not so interesting since we can easily dump it in-system.

- Would anyone be interested to do it? As I said, it should be pretty easy to find a "donor" in the near area.

#108736 - 02/06/17 12:19 AM Re: Consider uncapping TMC0430 [Re: mizapf]  
Joined: May 2010
Posts: 1,041
seanriddle Online content
Very Senior Member
seanriddle  Online Content
Very Senior Member

Joined: May 2010
Posts: 1,041
I've got some TMC0430 info from the Milton game here: http://www.seanriddle.com/mc6805p2.html

I can do more testing by controlling it with a PIC MCU and capturing the signals with a Logic if you need it.

I think I decapped one; I'll take a look.

Edit: I found an ugly top metal die shot and a whinked die shot:
http://seanriddledecap.blogspot.com/2017/02/blog-post_5.html

Edit2: It looks like the TMC0430s in Milton are a little different from the ones used by TI-99/4A.

Last edited by seanriddle; 02/06/17 05:29 AM.
#108741 - 02/06/17 04:42 PM Re: Consider uncapping TMC0430 [Re: mizapf]  
Joined: May 2007
Posts: 542
mizapf Offline
Senior Member
mizapf  Offline
Senior Member

Joined: May 2007
Posts: 542
Germany
We recently got some graphs from a logic analyzer that seem to prove some of my assumptions, but it could be interesting to see it in hardware.

However, I don't know whether it is actually possible to "read" the logic from the die shot. I can imagine that decoding memory cells is much easier.

#108742 - 02/06/17 04:53 PM Re: Consider uncapping TMC0430 [Re: mizapf]  
Joined: Mar 2001
Posts: 15,954
R. Belmont Offline
R. Belmont  Offline

Very Senior Member

Joined: Mar 2001
Posts: 15,954
USA
You can translate a die shot to logic (e.g. visual6502.org), but it's a specialized skill.

Torturing one with a micro controller as seanriddle offered to do is probably a lot more effective for this type of thing.

#109085 - 03/06/17 05:06 AM Re: Consider uncapping TMC0430 [Re: mizapf]  
Joined: May 2010
Posts: 1,041
seanriddle Online content
Very Senior Member
seanriddle  Online Content
Very Senior Member

Joined: May 2010
Posts: 1,041
I did some more TMC0430 tests: http://www.seanriddle.com/grom.html

I bought a TI 99/4A DOS Manager 2 cart and dumped the 2 GROMs. Contrary to what I originally thought, they appear to operate identically to the GROMs in Milton.

First of all, the Milton GROMs are only 6K, as seen on the die shot. When you keep reading past the end, the output looks like normal data; it's not all 00 or FF, isn't random, and isn't a duplicate of previous data. So I thought I was getting 8K.

Addresses are written and read MSB first. One source (http://www.unige.ch/medecine/nouspikel/ti99/groms.htm) says LSB first, but another (https://archive.org/details/tibook_hardware-manual-for-the-texas-instruments-994a) says MSB first.

But both sources say pin 1 is D7 and pin 8 is D0, while my tests show pin 1 is the lsb and pin 8 is the msb.

Let me know if anyone has any other tests they want me to run.

D'oh - I see now that the TI 99/4A convention is that D0 is the msb and D7 is the lsb. So everything I've seen is consistent with the hardware manual I linked above.

Last edited by seanriddle; 03/06/17 05:15 AM.
#109087 - 03/06/17 10:34 AM Re: Consider uncapping TMC0430 [Re: mizapf]  
Joined: May 2007
Posts: 542
mizapf Offline
Senior Member
mizapf  Offline
Senior Member

Joined: May 2007
Posts: 542
Germany
Thanks for your reply, any help is certainly appreciated. Before you dig too deep here, please have a look at src/devices/machine/tmc0430.cpp to find out how far we already got by behavioral analysis. Using a logic analyser, we recently got some more information that has not been added to that file yet.

The reason for my original question about uncapping was that we could possibly prove some of our assumptions by the real hardware.

#109131 - 03/11/17 10:52 PM Re: Consider uncapping TMC0430 [Re: mizapf]  
Joined: May 2010
Posts: 1,041
seanriddle Online content
Very Senior Member
seanriddle  Online Content
Very Senior Member

Joined: May 2010
Posts: 1,041
No problem; I wanted to fix errors on my web page anyway. It makes sense that the unused page 3 is the OR of pages 1 and 2. I thought up a few more things to test; I'll put those on the page when I'm done.

It shouldn't be too difficult for someone with the skills to capture the logic from a TMC0430- it's not a large die, and something like 80% of the die area is ROM array, address decoding, bit multiplexing and I/O pads. And it looks like another 10% in the upper right of my die shot is registers for the 3 GROM ID bits and the 13 address bits.

I'm not proficient at IDing transistors and figuring out the logic, but I'm pretty sure the uppermost 8 pads on my die shot are the address/data pins. Since the pins are numbered counter-clockwise, that makes the pad in the upper right corner pin 3.

The bits are arranged in a simple order, so if there's ever a bad chip that won't electronically dump, it would be easy to visually dump it.


Who's Online Now
3 registered members (B2K24, seanriddle, Pernod), 39 guests, and 3 spiders.
Key: Admin, Global Mod, Mod
Shout Box
Forum Statistics
Forums9
Topics8,543
Posts111,677
Members4,795
Most Online225
May 26th, 2014
Powered by UBB.threads™ PHP Forum Software 7.6.0
Page Time: 0.031s Queries: 14 (0.010s) Memory: 4.9790 MB (Peak: 5.1792 MB) Zlib enabled. Server Time: 2018-06-20 18:56:59 UTC