Yet another homebrew saturn card

Project is a work in progress now. First public release is estimated before christmas 2015, but it's a very rough estimate.
Cart is based on Altera MAX10 FPGA with 256MBit SDRAM or DDR SDRAM onboard. Non-volatile data like saves/ROMs is stored on SD card. USB and BLE (bluetooth low energy) are used to communicate with external PC/smartphone.
Cart is open source and open hardware.

Feature list in current roadmap:
[WORKS, but slow] Power Memory simulation(from 4mbit/8000+ blocks, to 32mbit / 65000+ blocks)
[UNTESTED] 1M / 4M RAM simulation(as native as it will be possible, the maximum goal is to allow 1M games to work glitchlessly)
[NOT YET]ROMs from KOF95 and Ultraman simulation.
[NOT YET]Link to PC via USB and to smartphone via BLE (for downloading/uploading all of the flash content, changing cart type while in game, upgrading, etc.). BLE is optional (it can be excluded to reduce price), high-speed, normal and legacy bluetooth won't be supported.
[WORKS, but slow] Boot cartridge mode (run any code from file on SD card, including PseudoSaturn, development, homebrew etc.)

Possible features (not in this release) :
Modems simulation (both netlink and japanese one, modem mcu emulated from the saturn side, and a usb link to PC with a next layer of network emulation software, connected to the real internet)
CD-ROM emulation (like SAROOO project, but with lower speed)

Total parts and assembly cost for a 10 pcs. batch is somewhere around $50. For BLE add $5-10.

Preorder list, assemblers registration : www.wasca.info
Repositiory (for developers) : https://github.com/hitomi2500/wasca

 

Is it a wishlist, or the announcement of a new Saturn cartridge project ?

> (excluding visiting a psychiatrist)

In my case, I visit surgeon for head and neck cancer ...

> and will make it a bit difficult to fit into $50
Yes.

 

Great! Shut up and take my money

 

Some day all the original carts will be gone and making a replacement will be the only option. In the mean time RAM/AR/ROMs can be had for a few bucks and the modems not much more. I can get a lot of original cards for $50. Maybe it's interesting if pseudosaturn causes a shortage of AR cards?

 

Okay, okay. It is announcement all right. Open-hardware and open software. Codenamed wasca, specs will be publised soon.

No problem, bro. I will. If you insist

Well, i'm happy for you, but for me the picture is a bit different. I can hardly get two for $50 because of a far shipping. And the point here is not an low-cost solution (which it definitely isn't, being half the saturn retail price), but rather a cart that will be there for eternity, so nobody will ever need to touch that sensitive connector (though toughening the connector is not that difficult if you ask me). If you strip all of the functionality leaving only usb power memory, it will probably fit into $20-$25, but the need for this overpriced power memory is quite questionable.

Update : is there anyone familiar with Saturn's SCU's A-Bus timings? What i want to know is how exactly those peculiar signals like Function Code Bit 0/1 or External Access Timing 0/1/2 works. It appears i'm out of pins on FPGA after all, so i will either reduce SDRAM width to 8 bit, or add some A-Bus address/data multiplexing. Using Address Strobe and these Timing 0/1/2 should make it possible, but i'm unsure how they works exactly.... I can analyze it myself, but that's some quality time with oscilloscope or function analyzer. Maybe someone did that already?

Update again : pre-route schematic is finished. Still might contain some unfound bugs though. I will start route in friday. Everyone interested might check local supplier for parts cost. Haven't done the BOM yet, but there are only a few components there (excluding a bunch of caps and resistors, but they're cheap), and i ended up with $37 for 1-piece US-buyer prices in mouser: http://www.mouser.com/ProjectManager/ProjectDetail.aspx?AccessID=37893c883e , (WARNING: THIS LIST IS NEITHER COMPLETE NOR PROVED WORKING YET, BUY/ORDER AT YOUR OWN RISK) but i used a smaller memory (256 mbit instead of 512) because 512mbit costs surprisingly much, and a bigger fpga (10M08 instead of 10M04) because 10M04 are not available yet.
Add PCB cost ($3-$10), 3D-printed case ($3-$10), soldering cost (i have no idea) and you can probably fit into $50 for a 10 pcs amount.

 

It looks like an edit button is broken for a while. I will post progress in a new reply for now.
Board is routed(see pics), schematic fixed accordingly, github repo is https://github.com/hitomi2500/wasca
Today project entered a hardware validation stage, i will try to create valid FPGA design with current pins configuration. After that the PCB will be sent to fab.

 

If shipping to where you are is so expensive then shipping from there will likely also be very expensive. It's going to make it harder to sell a multi function card when shipping costs more than buying a few of the cards it replaces.

 

Agreed. That is one of the reasons i am not selling them, but sharing design sources, so that folks interested should be able to assemble them cheaper, or get them cheaper from the neigbour. If the project will be successfull enough, then maybe some chinese guys will start to produce and ship them cheap all over the world. But that "cheap" is still somewhere around $50 for now.

 

Has it been tested yet or still in the design process?

 

As a potential US buyer, it's great to see Helder interested in this thread. He seems quite capable of taking these awesome open source designs, getting them assembled, and delivered into the hands of the lazy video game consumer (e.g. me, lol).

 

Still designing. Right now i'm developing a pin-accurate FPGA firmware to validate current routing. In a week or two i'm sending PCB to fab, then add 3-4 weeks for production and shipping, 1 week for soldering, and some debugging (most of the firmware will be tested on simulator without hardware). So if you want a hardware-test-proven design, you should wait for 2 months (it's an optimistic approximation).

 

Some random questions/comments :
- The FPGA you're using isn't longer produced by Altera ? Do you have a place where to purchase it ?
- What is the part number of the SD card socket you're using ?
- Do you plan to provide API for SD card access from SH-2 CPU ? Also, you should provide some other registers for version info, etc so that firmware can detect cartridge type in use.
- Segasata'n Shiro

 

Nope, it isn't YET produced by Altera. MAX10 is a brand new series of their max family, and although all of the parts are documented and supported by tools, only a few are available at the retail market now. So 08s you can buy today, 04s will be available later.

Just a "standard" card holder from kicad library. Thanks for pointing that out, i will check if there are some available holders with this footprint before producing PCBs.

Providing these interfaces for SH-2 is not a problem, but i need some SH-2 software to talk with these regs to debug it decently. Since i'm a complete zero in a Saturn development now, i'm not planning to write this software myself, not at the first stage of the project. Are you interested in this?

Hell yes! But i'm afraid that the fab i will be sending design to (seeed or dirtypcbs) won't be able to print Sanshiro-sama precise enough to recognise.

 

(short answer from smartphone)

Please let me be your software/debug guy :]

Another remark : square in usb dev cart comment layout indicates that no IC should be here. Is this OK with your cart ?

Another question : what is the meaning of wasca ? 僅 ?

 

There's no way i can reject you ^_^

Oops, you are right. I already edited some overlapping parts, but as i look at it now, some capacitors will prevent card ejecting/inserting. I will fix that.

「なぜ他のサターンカート再び？」英語で。

Update: SD card is a standard one after all. Compatible with 5638 series from Kyocera or 10100 series from Amphenol. Some of these still might require mechanical adjustments, i will publish 100% compatible part numers later.
Board is in production now at dirtypcbs, viewable and orderable (THIS IS A FIRST VERSION, IT MIGHT NOT WORK WELL, AND MIGHT NOT WORK AT ALL, BUY AT OUR OWN RISK!) here: http://dirtypcbs.com/view.php?share=7875&accesskey=3cd3896f732b1946b81be99289a8c1a1

 

Sorry for the delay. I only have limited access to internet/PC on weekends, and was busy on last Monday/Tuesday ...

You're really fast at prototyping ! I started to modify USB dev cart sources 3 years ago, and my project is still WIP ^^;
Sorry for the confusion about MAX10 FPGA.

If your gerber is on dirtypcb, I suppose PCB manufacturing has started ? If so, I should hurry up in order to make test program for your board.
Can you let me know about what is required for your testings ?

If was thinking about something like this :
- one shot/continous read/write of specified register
- allow to enter register address and size from pad
- hex display of register values
After you get USB connection working correctly, I can add PC-Saturn transfer, so that you will be able to perform the same tests from PC.
Please let me know if you need something more.

Your segata sanshiro logo on silkscreen looks nice I hope result on PCB will be OK.

 

Hey, take it easy, we're all busy people here. Take your time and do your best.

Yes, they reported that the board was sent to factory, yet you shouldn't hurry that much - i haven't purcharced all the parts yet. I expect to start software tests somewhere around august.

That would be great. By size do you mean access width (8/16/32) or array size? Reading/writing arrays might be helpful to test consecutive transfers. And i will need to check a DMA transfers too (both from SH-2 DMAC and SCU's internal DMA), so i will need some DMA reads/writes later.

I think we won't need bus tests anymore at that point.

 

OK, this should give me enough time for your test program ^^

I suppose you will need access for single register while not touching others ones, so I will add testing features for both single register and arrays.

OK, but it will be add non-DMA things first.


Some questions :
- Do you use Windows ? In order to know if I should add emulator with test tool, so that you can verify its behavior before turning on your Saturn.
- There's no test pads/LEDs on your PCB ? Personally, I can't debug without this kind of extra things, but it's probably due to the fact I'm a beginner in vhdl
- Where do you plan to implement SD -> SDRAM file load routine ?
Everything on vhdl ?
Or do you plan to copy a bootloader program from fpga's flash memory to SDRAM, and then bootloader will load file from SD ?
(as long as fpga can load 64KB-ish boot code, the remaining tasks can be done on SH-2 side)

 

Yes, i have two Windows PCs, one as laptop, and one at work.

I ended up with only 1 unused pin after routing. I connected it with a big enough via, but it was deleted accidentally during some re-route. I can still solder a wire to this pin for debug purporces, or use in-fpga virtual logic analyzer via jtag.

This will be done by a Nios soft CPU. It's a ready-to-use solution from Altera, and Altera's IP will generate a vhdl code according to what features i want in my CPU, what buses i have around it, and what peripherial i add, Most of the work done by Altera tools, i just fill boxes with values, and use off-the-shelf compiler to generate firmware. Firmware is stored in fpga's internal flash.
I want to leave SD stuff to Nios. If, by any chance, Nios performance won't be enough, i'd rather add another mcu than transfer some work to SH-2.

 

OK, so it will really be a 100% hardware solution
I started to write the tool and plan to show a beta next week.