CD Stupid Card 4.0

[elmer]

Still, as soon as the time is right, somebody should really contact krikzz with details about the Stupid Card. You can be sure that he is working on finalizing the TED2.0 right now, so it's a really good time to let him know.

I've just posted there to see if he'll tell us his design goals ... but it doesn't mean that he'll respond.

FYI ... he posted pictures of the prototype TED2 cards a while back ...

Version 2.0, dated 2nd Oct 2014.
Version 2.1, dated 22nd Dec 2014.

It looks, to me, like they contain ...

Code Select Expand

2 x ALVC16424S 16bit 3.3V-5V transceiver
1 x 50MHz oscillator
1 x Lattice/SiliconBlue iCE40HX1K FPGA (1280 logic cells, 64Kbit embedded RAM).
1 x FTDI FT245RL USB2 to parallel FIFO (up to 1MB/sec).
1 x Micron CellularRAM PSRAM MT45W?MW16
 either ... MT45W4MW16BGX 4MBx16 (8MB)
 or ....... MT45W2MW16BGX 2MBx16 (4MB)
 or ....... MT45W1MW16BGX 1MBx16 (2MB)

That's a powerful piece of hardware!

[TailChao]

I certainly can't see why they chose to do the external mapping rather than the simple 2-level mapping that's on the Stupid Card ... nor can I see why they couldn't have chosen to do the 2-level mapping with dram.

My assumption is that the DRAM they used was actually very (s)low grade. Having access limited only to ports means that the DRAM controller has all the cycles in between the actual reads to prefetch or store data.

From what TailChao has said it's already working in Mednafen ... so "yes"

The entire feature set isn't supported (obviously, since we're still designing the card in this thread). Right now you only get 512KB ROM + 512KB RAM using the patch I supplied in the first post since there's enough overlap between the (new) MCGenjin-CD and original MCGenjin mappers.

Luckily it shouldn't be too much work to add MGCenjin-CD emulation once we're finished with the card design and verification. But I will leave that to you guys and the Mednafen author if she wishes to do so.

Having said which ... I'd still really like one of these for development, whatever happens with the TED2. Combine the Stupid Card with a USB-based Develo-board and you've got a really nice development system.

Exactly, and I'm glad someone picked up on this. While the CD Stupid Card does allow you to use it as an enhanced Super System Card 3.0, it's really just another option for developers.
No one says you have to use it as a system card, you could pull off the BIOS and replace it with something that does bitbanged comms with a PC over the controller port. Then download your test programs to the 2MB of RAM.

FYI ... he posted pictures of the prototype TED2 cards a while back ...
...
That's a powerful piece of hardware!

Yeah, once you start using lower voltage (i.e. modern) components and doing large production runs some really cool stuff can be shoved on that little HuCard form factor.
The FPGA on there is huge though. Maybe the interface will be as nice as the Everdrive N8 where there's literally a bunch of RBFs in a directory which store mapper setups. In that case Stupid Card support will be trivial.

[TailChao]

Alright everyone, we've hit our feature lock date. Now it's time for me to move on to board layout.
I'll be allocating two weeks for this, and will report progress on Mar 20, then hopefully complete the design by Mar 27.

Thanks everyone for participating in the design phase!

[elmer]

FYI ... I just got a reply from KRIKzz on his forums ...

Re: Turbo ED V2 - Questions for fan translations and homebrew.
« Reply #1 on: Today at 05:36:35 PM »
turbo-ed v2 allow to dissable write protection, so, whole onboard ram (4Mbyte)  avaialble for developers.

No info on the access method ... but I'd still put money on the Street Fighter method, which is 512KB fixed and 512KB banked. Pretty much perfect for the translators wanting a System Card 3.0 + 512KB RAM.

[SamIAm]

Splendid! The user base for RAM-expansion translations increases!

[NightWolve]

Good, as I sold mine off in hopes of significant enhancements. Redesigning the thing just for load time improvement alone wasn't worth it, so you had to figure he'd add other feature(s) at the very least to help better distinguish it from the previous model.

[TailChao]

Re: Turbo ED V2 - Questions for fan translations and homebrew.
« Reply #1 on: Today at 05:36:35 PM »
turbo-ed v2 allow to dissable write protection, so, whole onboard ram (4Mbyte)  avaialble for developers.

No info on the access method ... but I'd still put money on the Street Fighter method, which is 512KB fixed and 512KB banked. Pretty much perfect for the translators wanting a System Card 3.0 + 512KB RAM.

At least we know the amount of RAM on the card, now.

Personally I'm hoping for some method to reconfigure the FPGA on his cartridge. This is provided on things like the Everdrive N8 where the SD card literally has a folder with different RBFs for different mappers. It's pretty nice for new development (not to mention improving emulation accuracy).

[TailChao]

Alright, end of the week update-
Attached is the layout status, shouldn't have any issues finishing the routing and dropping on a few more caps by the end of next week.

Edit:
-The large 39F040 is where the BIOS / Whatever ROM'd code you like will be stored. This will be socketed.
-The two LEDs on the top right of the card indicate whether the SWAP and LINEAR modes have been set to simplify debugging.

[elmer]

-The large 39F040 is where the BIOS / Whatever ROM'd code you like will be stored. This will be socketed.

Cool!

It's just crazy how the flash chip is so much larger than 2MB ram and the CPLD ... I guess that's why everyone moved to surface-mount.

Is there any chance of a ZIF socket fitting on there for the flash?

Attached is the layout status, shouldn't have any issues finishing the routing and dropping on a few more caps by the end of next week.

It's all that signal strength/coupling and power supply smoothing stuff where you totally lose me.

I'm glad that you're here to do this!   

[TailChao]

It's just crazy how the flash chip is so much larger than 2MB ram and the CPLD ... I guess that's why everyone moved to surface-mount.

Is there any chance of a ZIF socket fitting on there for the flash?

Yeah, it's pretty astonishing how much IC sizes have shrunk in the last few decades. There aren't any BGA parts on this board, either. The RAMs are only TSOPII's.
Funny thing is that if I used a smaller package for the flash it would have been way easier to route.

I'm actually not sure if there is enough clearance on the left side of the card to fit a ZIF socket which won't bonk against the card guide in every single PC-Engine variation. What I'd actually recommend is just taking a ZIF socket and then placing it in a wire-wrap socket (which have very long legs). This can then be placed in the board's normal socket and allows you to reuse the ZIF socket in other cards / projects (they're expensive, so I stopped soldering them down years ago).

It's all that signal strength/coupling and power supply smoothing stuff where you totally lose me.

Thing is that for these old machines you rarely have to worry about such things. The data speeds are too low to be super-vulnerable to interference. So what I am doing is really just a glorified connect the dots.

On some of my earlier cards, I had no proper decoupling caps at all. But they still worked.
A good rule of thumb is one small (.1uF-ish) cap per VCC pin on each IC and one larger cap (maybe 10uF) for the whole card. But I'm not going to fret over leaving out one decoupling cap if I can't fit it anywhere.

[elmer]

What I'd actually recommend is just taking a ZIF socket and then placing it in a wire-wrap socket (which have very long legs). This can then be placed in the board's normal socket and allows you to reuse the ZIF socket in other cards / projects (they're expensive, so I stopped soldering them down years ago).

That sounds like a great idea ... if a bit ugly.

I guess that in practical reality, there's not a lot of reason to flash anything other than the regular system card bios onto the Stupid Card. The would give maximum compatibility for unmodified games ... and the modified ones just load up whatever changes they want.

I guess that a question that really should be asked is ... does it even make sense to take up your valuable time and energy in actually going ahead and manufacturing these cards at this time?

I'd hate to see you get stuck with a lot of boards and parts.

There doesn't seem to have been a lot of interest from translators/developers looking to use/buy a Stupid Card so far.

The translation guys have already got a Mednafen-capable version of your MCGenjin that they can develop with in preparation for the volume-produced TurboED2 ... and developers like myself are going to find the TurboED2's 4MB RAM and its built-in USB port and SD card to be irresistible draws.

Yeah, it's pretty astonishing how much IC sizes have shrunk in the last few decades. There aren't any BGA parts on this board, either. The RAMs are only TSOPII's.
Funny thing is that if I used a smaller package for the flash it would have been way easier to route.

I've just found out how far the FPGA scene has come ... it looks like it's now possible (and afforable) to emulate complete old computers/consoles in hardware on an FPGA. It looks like there's at least one PCE-on-an-FPGA in development.

There are even people selling things like the MIST board, specifically for retro computing/gaming ...

https://code.google.com/p/mist-board/

There may be a Terasic DE1-SoC in my future sometime soon. The idea of being able to put together a complete computer without having to do any surface-mount soldering is pretty appealing. ...

http://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=English&No=836

[TailChao]

I guess that in practical reality, there's not a lot of reason to flash anything other than the regular system card bios onto the Stupid Card. The would give maximum compatibility for unmodified games ... and the modified ones just load up whatever changes they want.

For most users, yes.
Personally this is a future investment so I can have a HuSound "loader" to receive SFX, MUS, and PCM packs over the controller port from a PC. My Lynx sound tools allow downloading like this and having the Compile -> Send -> Listen, repeat workflow with real hardware in the loop is pretty nice.

I guess that a question that really should be asked is ... does it even make sense to take up your valuable time and energy in actually going ahead and manufacturing these cards at this time?

I'd hate to see you get stuck with a lot of boards and parts.

I've been thinking the same thing. However, I'm going to make at least a few cards since I want to use them.

Once layout is done for next Friday's deadline, I'll figure out the final cost of the boards and parts and leave this topic open for a "formal" developer sign-up to get a final count. Again, even if this number is zero, I'll still fab a few cards and release all my pertinent work.

I've just found out how far the FPGA scene has come ... it looks like it's now possible (and afforable) to emulate complete old computers/consoles in hardware on an FPGA. It looks like there's at least one PCE-on-an-FPGA in development.

There are even people selling things like the MIST board, specifically for retro computing/gaming ...

https://code.google.com/p/mist-board/

There may be a Terasic DE1-SoC in my future sometime soon. The idea of being able to put together a complete computer without having to do any surface-mount soldering is pretty appealing. ...

http://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=English&No=836

It's really fantastic stuff, isn't it?
I invested in an old Altera DE2 years ago for my own prototyping work, and it's astonishing how much cheaper and more capable FPGAs have gotten. The DE0 Nano and friends are nuts in both price and gate count.

[TailChao]

Alright, everyone. Layout is completed and it's time for Developer sign-up.
Cards will cost $35/each + shipping and I have listed a component cost breakdown in the first post for those who'd like to manufacture their own cards in the future.

If you're interested in getting a card, please state so AFTER this post (all prior "give me one" posts are ignored). A total of 25 cards are available as first-come first-serve. You may sign up for more than one card.

This sign-up period will be open until April 10th. I will not collect anyone's payment until all cards have been built and tested. Please hold on to your money.

[MotherGunner]

Please put me down for one, thanks!

[elmer]

Alright, everyone. Layout is completed and it's time for Developer sign-up.

Yay! Thanks for doing this ... one for me, please. 

[CrackTiger]

I would like to get one please.

[ParanoiaDragon]

I'm tempted to get one.  I haven't read every last post on the the features.  Is there any advantage to getting this to play currently released games, or completely for upcoming homebrew/translations?

[SkyeWelse]

I kind of have the same question as ParanoiaDragon. I'm not a PCE developer, or a PCE Romhacker, but if I did decide to work on a project one day with a group here, would it benefit me to have this and would it generally have a benefit for the non-developer at this stage in the development process?

Also, would you plan to produce more batches in the future if it made more sense for non-developers to take a pass on it currently so that it can be used by developers as a priority?

-Thomas

[CrackTiger]

I'm tempted to get one.  I haven't read every last post on the the features.  Is there any advantage to getting this to play currently released games, or completely for upcoming homebrew/translations?

Since this uses the MC Genjin mapper, I'm assuming that you could play the sound roms that TailChao has made for it, if he makes them public, as well as anything else made in the future that uses the mapper to make development easier.

[elmer]

I'm tempted to get one.  I haven't read every last post on the the features.  Is there any advantage to getting this to play currently released games, or completely for upcoming homebrew/translations?

I kind of have the same question as ParanoiaDragon. I'm not a PCE developer, or a PCE Romhacker, but if I did decide to work on a project one day with a group here, would it benefit me to have this and would it generally have a benefit for the non-developer at this stage in the development process?

There is no advantage that I can think of to using this on currently-released games ... they won't even know that it exists.

As for the future ... who really knows?

I honestly can't see any translators making this card a requirement when there are going to be 25 or less of them in existence.

If translators or homebrew-developers decide that they need extra RAM on a PCE, then they can develop for either this card, or for krikzz's mass-produced (but more expensive) Turbo EverDrive 2 ... or both.

That will be entirely up to them.

Also, would you plan to produce more batches in the future if it made more sense for non-developers to take a pass on it currently so that it can be used by developers as a priority?

My understanding is that TailChao has already said that he's not interested in mass-producing these cards in the future ... but that all the plans and schematics necessary to do so will be freely available for anyone that does want to do so.

That's incredibly generous of him ... as is his offer to actually make some of these cards for the people that want them, at what is basically his cost price.

But think about that mass-production run in the future ... who is going to do it?

Do you want to gamble that the community here will get together and commit to an order of 100 cards at say $60-each (if you get someone in China to manufacture them)?

IMHO, unless you're a developer that knows exactly why this card will benefit you ... and what you can use it for right now ... then I suspect that you'd just be taking up TailChao's valuable time and your money in buying a beautiful development card that you'll never use and possibly regret.

Then again ... perhaps in 5 years this will be a must-have "collectable" in it's own right! 

[synbiosfan]

IMHO, unless you're a developer that knows exactly why this card will benefit you ... and what you can use it for right now ... then I suspect that you'd just be taking up TailChao's valuable time and your money in buying a beautiful development card that you'll never use and possibly regret.

That is my thought. However, I don't want TailChao sitting on 5-10 cards for a year+ and regretting making 25. I'm holding off showing my support (buying one) for TailChao's efforts until I see his target audience has had their opportunity. Even then I'd just be a caretaker until a developer came along who missed out.

[Lochlan]

TailChao, please sign me up for purchasing one card.  Thank you!

[ParanoiaDragon]

Ok, I'll hold off for now.  I'm hoping these will be helpful & mass produced for the future, depending on who takes up the task.

[TailChao]

IMHO, unless you're a developer that knows exactly why this card will benefit you ... and what you can use it for right now ... then I suspect that you'd just be taking up TailChao's valuable time and your money in buying a beautiful development card that you'll never use and possibly regret.

This is fairly good advice.
I do not mind manufacturing 25 of these cards for the developer run, nor ending up with leftovers (things break, extras are good to have). But unless you have a specific development paradigm where this card would help, I'd hold off.

Since this uses the MC Genjin mapper, I'm assuming that you could play the sound roms that TailChao has made for it, if he makes them public, as well as anything else made in the future that uses the mapper to make development easier.

Actually, this card uses a new mapper (MCGenjin-CD) which is not 100% compatible with the original MCGenjin (which makes sense, since they're designed for different things). So my sound driver is not compatible (yet).

[wyndcrosser]

I'm interested. PM me when you want $$. Thanks

[OldRover]

I'm in... let me know what goes on.

[VenomMacbeth]

I wasn't interested in this initially, but then I read about loading hucard images from CD...and that sounds spectacular. :3 Methinks I will get one when they become available!

[elmer]

I wasn't interested in this initially, but then I read about loading hucard images from CD...and that sounds spectacular. :3

They're only going to do that if/when someone writes a small homebrew app to do that (which shouldn't be difficult).

AFAIK, it's not going to be a built-in function ... unless I've missed something, and TailChao or someone else has decided to modify the System Card to do that.

Methinks I will get one when they become available!

Err ... TailChao isn't a shop keeper. He's building a small made-to-order run of these as a favor to the community, and you've got until April 10th to place that order and commit to buying it when TailChao builds it.

Do you want to do that?

[TailChao]

AFAIK, it's not going to be a built-in function ... unless I've missed something, and TailChao or someone else has decided to modify the System Card to do that.

You're correct, the functionality is not built in aside from the mapper's support for linear mode.
An application to actually load the games would have to be written. I have no plans to do this as of writing.

[SamIAm]

I'm in Japan, so it's going to be slightly trickier to get one of these to me. I might work out a trade with someone. Anyway, please put me down for one. If Bonknuts hasn't ordered one, then put me down for two so I can get one to him.

[cjameslv]

Put me down, i'm in.

[akamichi]

I would like to join in on this as well.

Thanks.

- akamichi

[elmer]

Just bumping this thread up to remind everyone of TailChao's deadline tomorrow.

Last call! Time, gentlemen, please!

[TailChao]

Alright, update time.

I submitted the PCB gerbers for the fabrication of 30 cards on Wednesday, April 8th. This does not mean that I'll be populating and distributing all 30 of them, just that the PCBs are ordered in batches of 10 and I'd rather have extras than not.
This is my first time ordering PCBs from China rather than doing prototyping here in the USA, so I can't give 100% accurate time estimates. But the factory said the cards would be fabbed in about a week so the rest should be just shipping delays.
The components for 6 cards are already at my home, I'll order the rest once I get the first card working.

Prior to ordering, I reassigned a few pins on the MCGenjin-CD CPLD to ensure the cards will still be fine if my assumptions about write timing are incorrect. (WRn is promoted to a global clock). I've updated the card image on the first post for those who are interested.

During the fabrication gap, I'll be writing a test ROM which will exercise the various features of the card. Once I've gotten the cards working and all that is left is populating the boards for you guys, I'll release all the pertinent development materials: PCB Gerbers, MCGenjin-CD Verilog source + POF, and the card test ROM. I want you all to know exactly what you're getting.

Since I'm not likely to be in front of a computer tonight to "officially close" the developer sign-up, here's our current list:
- MotherGunner
- elmer
- Black Tiger
- Lochlan
- wyndcrosser
- The Old Rover
- VenomMacbeth
- SamIAm
- Bonknuts / Tomaitheous
- cjameslv
- akamichi

[elmer]

Thanks for the update! 

I look forward to hearing how everything works out with the new PCB manufacturer.

[Debvgger_]

I think it's too late for asking for one? 

[cjameslv]

I think it's too late for asking for one? 

who's alt account is this?

[Debvgger_]

My own account, cjameslv. Lost the access from the version without the final '_' so I had to register again 

[TailChao]

I think it's too late for asking for one? 

I won't be able to make a decision to manufacture extras or not until the PCBs arrive from the fab and I've made sure the mapper logic is correct.
I'll PM you if there is a card available.

[Debvgger_]

Thank you! :-) I will keep my fingers crossed, feet included.

[TailChao]

Picked these up at the post office yesterday.

Looks like everything was fabbed properly. The only immediate mistake is actually my fault (I made the pad rows on the RAM footprint a little too close together, but this isn't anything that will break the cards, just complicate soldering).

Luckily the wait gave me enough time to get my test program finished and fix my Turbo CD (the seek gear finally shredded itself, quite lucky that we're able to order new ones). I'll try and get the first card up and running, passing the test program, and playing Gradius II by May 8th. Then it's time to manufacture everything for you guys.

[elmer]

Picked these up at the post office yesterday.

Thanks for the update, they look really cool! I can't wait to get one in my hands. 

[ddoubledee]

OMG, WANT! :O

[Debvgger_]

How is it going TailChao? :-)

[TailChao]

How is it going TailChao? :-)

Very well! Around 11pm last night I finally got to see the two screens attached below.
Quite a relief, as the first card I assembled did not work properly (there is likely a bad solder joint somewhere, the TSOP-IIs SRAMs with curled under pins are a little tricky).

This means that it's resource release time, and now I'll be moving forward to manufacturing. Since this will take awhile, and I cannot give a good estimate as to exactly how much, I'll be continuously updating the first post's new "manufacturing tracker" so you can get an idea how things are going.

Before that, though- I'm going to perform a few tests with different speed grade CPLDs and region mods. One concern I have is that the region mod most people use (the one with the analog switches) adds a very large propagation delay, which could cause the card to not work properly. I'll keep you all posted.

[elmer]

Very well! Around 11pm last night I finally got to see the two screens attached below.
Quite a relief, as the first card I assembled did not work properly (there is likely a bad solder joint somewhere, the TSOP-IIs SRAMs with curled under pins are a little tricky).

Excellent news! 

Did you always intend to curl the pins under the chip, or was that a cunning solution to ...

The only immediate mistake is actually my fault (I made the pad rows on the RAM footprint a little too close together, but this isn't anything that will break the cards, just complicate soldering).

[TailChao]

Did you always intend to curl the pins under the chip, or was that a cunning solution to ...

The only immediate mistake is actually my fault (I made the pad rows on the RAM footprint a little too close together, but this isn't anything that will break the cards, just complicate soldering).

It was the latter. But I always intended to have the NC pins removed during assembly since it greatly simplifies layout.

As an aside, I've finally *removed* the region mod in my TG16 since this card now takes care of the only reason I installed it in the first place. However, during that process I did a few quick checks as to how the CD Stupid Card and a real System Card 3.0 react to different propagation delays (either due to wire length or the MC14551s). They both start to fail around the same time (especially with the MC14551s driven with a 5V supply), so I'm not too worried.  But I'll still order the 6ns CPLDs and see if things improve (we're using 15ns right now).

The TG16 has verrry long dataline traces on its mainboard and can be a real jerk in regards to signal integrity. The unmodded white PCE I have seems to run nearly anything. I imagine you Duo owners never have to think about any of this as everything is condensed on one board.

[Nio]

Just signed just because of this. I am in. *happy*

[TailChao]

Alright, I just found something seriously messed up:

The reset timing on the PCE and TG16 is different, in that the TG16 will occasionally have the CPU released from reset long before the signal rises on the HuCard slot.

This was causing the mapper to occasionally miss its first write (which sets the operating region) on the TG16. Putting a delay there will solve the problem, and I'll be flashing the patched System Card BIOS with the predelay added for the developer cards.

I guess this explains why some owners see different behavior among different consoles.
If any of you want to make your own mappers, this definitely needs to be kept in mind.

[Debvgger_]

Just signed just because of this. I am in. *happy*

Well done ;-)

Tailchao: Very good news indeed, and an interesting find.