Sega Dreamcast Code Checker Dev Hardware?

I dunno if this is known dev hardware (if it is, please enlighten me), but my friend sent me some pictures of a Dreamcast he received from another friend a while ago.. Apparently it was a Sega internal code checker of some sorts? He said "When you submitted a game for certification they would do a play through on this thing. And if it did something weird then that thing on the side would beep and the title will fail" Yeah I know what you are all thinking. "What's underneath that fan/heatsink?" Well I dunno..lol..

Anyone know anything about this?

 

I'll say it's dev hardware as the idea of pass/fail and a way to check code without looking at it in detail is a very handy tool.

What makes me think is the nut/bolts on the board as that seems potentially dangerous if the unit isn't protected well.

Do you have a way to check if it functions? Take a game and make a setup and give it a go.

 

I dunno how he would test if it functions, not really sure how it checks or what it checks for.

 

I'd guess this (quite simple) piece of hardware detects and indicate access to various address space areas.
some of them normally never accessed by game code, it is CS2, CS6, EXTxx and BOOTROM.
but if this thing's LED(s) will be lighting/blinking - game code had accessed mentioned areas, and this means there is something very wrong with this code (usually its because of non initialized pointer usage, or result of some wrong pointer arithmetic, or some other kind of memory trashing/corruption was happened) -> test failed )

 

Interesting. Thanks for the info!

 

That is actually really interesting, I would love to see what is under the fan.

It appears to be a peripheral bus hardware debugger of some sort, with EXTxx being external interrupts, and I'm assuming the GDROM and modem were on seperate busses, as were the SH4 and ARM MCUs in the system. I don't have a hardware block diagram handy otherwise it would probably be easier to figure out.

There are also some memory-section related LEDs on there so it is possible it was used to assist in debugging the MMU, but that was all internal to the SH4 as far as I know so I would be surprised if any external tools were needed for it.

Switches two and three appear to be setup to clear the interrupt register (INTCLR) and reset the memory (MRESET), any idea what swtich one is labeled as? I'm not seeing it anywhere. Also interesting to see that this is routed through the left hand side of the unit and not the right-hand side where the BBA/modem were. Is it removable? Any chance of him removing to cover to examine the PCB interconnects?

TriMesh might know more about it.

 

If you look at Katana system architecture diagram or schematics you'll see - *ALL* LEDs is memory section related. each of them corresponds to some address space part (chip select signal)

 

It still looks like a fan to me, but I could be wrong. Do you have a schematic link handy? It has been a while since I have seen a good one.

I suppose you can say they are all memory related they will need addresses on the peripheral bus, but to me it seems more like a development tool rather than a memory address validation tool. Notice there is a connector up top for JTAG or similar (CN1). Notice also that some of these devices (modem, GDROM, etc.) would be perfectly fine for retail software to access during normal operation. So unless the LEDs are different colors it really doesn't help you much in that sense.

Looking at it a little closer we also see an on/off switch labeled INT... I wonder if this was used to halt retail systems during runtime and perform a boundary scan or similar. I would say its not outside the realm of possibility knowing that Hitachi/Renesas used their own proprietary JTAG protocol and tools.

 

Here's something surprising. That board is actually part of the main board! And it has it's own silkscreen name on the main board. It's called the Katana Simple Checker 837-13657.



The chip underneath the fan/heatsink is a Altera Max.



Pretty exciting stuff.

 

Reveal your secrets...

 

That's very interesting! Thanks for sharing!
Any chance you could power it up with a game and see what it does?

 

Oooh.
Never seen one of these before. Thanks for posting.

Can the person with the console PLEASE buy a USB Blaster, and read the code from the chip?

It's really simple to do (using the free Quartus "Web Edition" software), and the USB Blaster clones are dirt cheap...

http://www.aliexpress.com/item/alte..._1&btsid=1f1fc70d-148e-41bb-8d74-80de308b20dc


As MetalliC and others have said - it probably just checks that each memory range is accessed properly (using a test disk maybe?), but it may be worth making a backup of the FPGA bitstream for preservation.

The chip is quite small, so there won't be a huge amount of logic in there (only around 2,500 "gates").

What might be more interesting though is how the board connects to the rest of the main board.
It looks like it's all one big board of course, but I wonder if it gives access to some of the signals on the CPU which are normally buried? (eg. RAM chip select lines etc.)

OzOnE.

 

Are you able to dump the bios?

 

Interesting. Can't believe I haven't seen this before.

 

Can't wait to find out what this does/was actually for. I saw some pictures of the external PCB quite a while ago, but I shared this on my site in the hope that someone may see it and be able to offer more information (hope that's OK). Gave credit to this forum thread so people know where to come for more info: http://www.thedreamcastjunkyard.co.uk/2016/04/katana-simple-checker-innards-revealed.html

 

Yeah that's no problem

 

@CoolMod thanks for more photos

+1 on this.
this mobo have undumped MPR-21933 bios rom. curious what version it is ? most common ver 1.01d is MPR-21931

 

TriMesh got a hold of me via PM a few days ago, it is indeed a memory region checker. This was apparently a tradition on all the Sega hardware. The LEDs were all red, but only some of the regions were defined as illegal. On access to an illegal area the CPLD would halt execution with an interrupt, and execution could be resumed with the INTCLR button. The LEDs were latched, so I imagine MRESET cleared them all when pressed.

This makes sense given that the SH4 was a RISC core IIRC and would have had a fixed width instruction set. It seemed that the CPLD did runtime instruction decoding and ran pretty hot. I don't know that the bitstream would really be all that interesting too look at (still cool though), but yeah I would love to see the interconnects.

Cool find-

 

Is the bios not dumped?