What's new
By:
Filters

Effort to get the GP9001 VDP chip, part of Toaplan V2 hardware, decapped

Thchardcore said:
We have to find 100 of me and you and maybe him out there. As much as I want this, I think there are other areas where the money might be better spent with respect to preservation.

With that said, in for a go fund me. Never set one up, but happy to take the lead.
Click to expand...
Yeah, that's my thoughts on it as well. Oh well, maybe down the road it will get cheaper to do this process!

Also, doesn't the Dumping Union or another group like that do things like this already?
 
Very promising feedback, thanks y'all! :thumbup:

So I've been mulling over this, and I came to the conclusion that I won't be starting a crowdfunding campaign for this in the near future: the stars are not really aligned currently, so to say.

Some factors are hard to predict, and I think we could collect more when we do this at another time.
I'll be keeping track of the way the situation evolves for now - from my point of view, this is the more reasonable option at this time.

Aside from that, I still think it doesn't necessarily have to cost as much. I'm in contact with a university, let's see if something results from that :)


brizzo said:
This is my first time reading this thread, and my first response was going to be "decapping and imaging is the easy and cheap part", because delayering and tracing into netlist is extremely time intensive and therefore expensive.... but you got that answer!
Click to expand...
The advantage on the other hand is that, once you have the images, you don't need bulky and/or expensive specialized equipment for the rest of the work. You can do it on personal computers and parallelize the work across many individual contributors working together.

@xtrasmiley: from what I gather, they also resort to more specialized decaping groups for stuff like this (e.g.).

To the voices saying the money could perhaps be better spent elsewhere: yes, certainly. maybe. It's a matter of opinion. But primarily, it's off topic :P
 
Although I agree that right now is a bad time for crowdfunding campaigns, I'm quite surprised at the idea that the money for decapping and analyzing the GP9001 could be better spent elsewhere. After all, the GP9001 is used in a LOT of prestigious, expensive shmup boards, including Battle Garegga, Batrider, Battle Bakraid, Batsugun, Dogyuun, and Truxton II.

Given that the GP9001 is the primary custom on all of these boards and that it's known for failure, I think most arcade shmup enthusiasts would pitch in for a decap and reconstruction knowing that it would (A) improve the quality of MAME emulation on those games (B) open the doors to FPGA simulation of those games (ie MiSTer) and (C) reproduced customs for owners of the OG boards.

After the Covid crisis is behind us all, I think a crowdfunding campaign would be quite successful as long as potential donors were made aware of all of benefits that could result from it.
 
I already handed off a spare GP9001 to a friend to pass on to be decapped, but more samples can't hurt. I don't know what kind of schedule that permits or if/when it will be decapped, but I can maybe retrieve the sample IC if we want to do something else with it.
 
Just for info, a french guy is reversing the CPS2 custom chips by analyzing the die extracted from the decapped chip. Here is a very interesting article he wrote about the "after-decapping" process:
https://medium.com/@WydD/diving-into-silicon-for-the-first-time-73086018e7de

I'm happy to contribute in this project as far as I can


PD: I have a PiPi and Bibis pcb somewhere which eventually could be donated to use its GP9001 to be decapped
 
6t8k said:
By the way, there are different four-digit numbers on the fourth and last line of the GP9001's imprint. Not aware of any difference those would make; they're likely just batch numbers, but I suppose it wouldn't hurt to decap at least two different ones while we're at it, no? But that decision would also somewhat depend on the additional cost involved, I suppose.

So far I've seen these (most certainly very incomplete):

#game(s)
9044Tatsujin Oh
9150V-V, Fixeight, Mahou Daisakusen, Tatsujin Oh
9152Dogyuun, Fixeight
9232Dogyuun
9235Knuckle Bash
9335Batsugun
Click to expand...
I'm pretty sure they are only manufacturing numbers, meaning fabrication year-week.

I have a Batrider pcb with "9717" labeled in its GP9001 (week 17 of 1997. Batrider is a game released on 1998). In fact, it seems to be made in Hong Kong, instead of Japan, as it is also stamped in the custom chip.


I bet that also Garegga and Bakraid have those Hong Kong produced chips but unlike Batrider, they have the Raizing label on it and it's not directly visible.
 
Raizing purchased the rights and required IP from Toaplan (or whoever held Toaplan's IP after they closed) to manufacture the GP9001. After Raizing used up the old Toaplan stock, they produced new ones at a different fab, hence the later ones that have the brighter text and triangle markings on them.

In these newer ones, you are likely to find some entirely unexciting and maybe logically insignificant changes.
 
I was unaware Raizing eventually went beyond using old stock and produced it themselves. I take it that the Raizing-produced ones look like this?

batrider_hongkong_gp9001_cropped.jpg

(couldn't find a better picture)

Was it one of these or a Toaplan-produced one you handed off @Hatsune Mike? In any event, definitely keep us updated!

@sergiopolog: very helpful article, thanks for bringing it up!


- - - - - - - - -

Here are some more, amateurishly compiled, bird's eye-view/introductory resources that I've found helpful for understanding the matter at hand:


Decapsulation and imaging (if you want to experiment with this yourself)

Peter Laackmann, Marcus Janke: Uncaging Microchips (A talk specifically about IC decapsulation. Slides.)
In the absence of professional tools, the most simple way is probably boiling the package in rosin/colophony. You still don't want to do this without a fume hood, however.

Pertinent article.

If you have the die, you can take a first (set of) image(s) of the topmost layer using a suitable microscope, which should at least:
  • use reflected light (meaning the light source shines onto the specimen from above)
  • have a mounting point for a digital camera (or have an integrated one)
  • together with the available lenses, provide a sufficient magnification factor (I suspect 50X-100X should be enough for the GP9001)
Some optional features aiding the process:
  • It being a stereo microscope, so you can actually see what you're doing
  • It having an X-Y table, ideally one that can be rotated within the horizontal plane, greatly simplifying the stitching together of many individual images
Here are some appropriate, mid-rangey examples:

mitutoyofs70.jpg
Mitutoyo FS70

zeiss_axiotron.jpg
Zeiss Axiotron

olympus_bx61.jpg
Olympus BX61

The die might have multiple layers. To comprehend its logic, you might have to deprocess one layer, take another (set of) image(s), deprocess the next layer, and so on, until you reach the bottom layer, where, typically, the logic gates are situated. There are various ways of doing this, one is to use a grinding machine, like the following one, in combination with a polishing slurry of the right grain size:

phoenix_4000.jpg
Phoenix 4000
Geology departments in universities potentially have those.


It helps to determine count and thickness of the layers in advance. This can be done by means of careful grinding into the edge of the die and looking at the resulting groove through a microscope, or by x-ray imaging. A professional in the field told me that for ICs with a sufficiently large process technology, which might be the case for the GP9001, it could be possible to detect all relevant features for netlist extraction just via x-ray imaging (decapsulation can be omitted, working order is preserved).

Analyzing the die

36C3 - HAL - The Open-Source Hardware Analyzer (Basic fundamentals of (IC) reverse-engineering + HAL project introduction. The software they present becomes interesting when you already have a netlist and want to understand/process/restructure it into usable blocks, and maybe, eventually, arrive at an HDL implementation. For working with die images, something like Degate could be used.)
36C3 - Understanding millions of gates

Diving into Silicon for the First Time (Seems like a good starting point; very insightful at least for me. Thanks again @sergiopolog for mentioning this.)
 
Last edited:
So it looks like Furrtek has decapped and imaged the GP9001 :thumbsup:

gp9001_decapped.png

[1], [2]

John McMaster so far also had been interested, but busy with other projects.

Not long ago I was only imagining it, and now... huge step for the preservation of Toaplan's legacy! :)
 
That's good but, sorry, you will never see a replacement of this custom IC because, even if we have the internal structure, you will be not able to fit all the logics in a single chip and even if you do it then you can' t think to mimic a QFP208 package also with a castellated design or flexible PCB :) The only option is to create another ASIC if you have some thousands of dollars to put in.
 
caius said:
you will be not able to fit all the logics in a single chip
Click to expand...
Curious about this. What is the biggest FPGA or CPLD in terms of logic-units, that works in 5V-levels?
 
@caius: Yes, without designing another ASIC, I see no way to directly replace the IC with a replica that fits neatly onto the solder mask geometry just like the original package. On the other hand, I think it may be feasible by designing an adapter board that acts as an interposer – let's not disregard that possibility. :) I don't know why fitting the logic should be a problem.

All the same, that sorta thing would bring the endeavor full-circle, and doesn't have to be the primary aspect I'd concern myself with just yet. If we can expect improvements in MAME and the MiSTer's emulation quality in the midterm for a start, then much would be gained already, wouldn't you say?
 
The big benefit at the moment would be the ability to use these boards with anything but an arcade/broadcast monitor. Currently, there is no workable option.
 
You must log in or register to reply here.
Back
Top