uze6666 wrote:I just read PSX docs and I think I got over exited a bit. Its much more complex both in term of protocol and cycle count. I'm really divided on that joystick issue!
Ok, so let me comment on that for the sake of debate. I have a few concerns about relying on the NES/SNES pads...
a) does anyone have a source for the NES joypad connectors in volume? I've *seen* the NES connectors on Andre's Hydra board (xgamestation), but I don't have a source for them. I can get a friend to ask the guy that owns Parallax if he knows their supplier. I had to buy ~2K of the PSX connectors as a minimum order from Taiwan a couple years ago so I'm worried that NES connectors might have the same volume commitments. The couple links I've seen online have been pulls from dead NES's and they only had a few parts. Even a 10 pc run of boards would wipe 'em out, much less 100...
b) does anyone have a source for new NES control pads with the original style connector? Most of the ones I've seen recently have been Famicom clones using DB9's. I found a couple in the $8/ea range and again, I notice that Andre found 'em for the Hydra, but I kinda doubt he'd want to help us out since he'd undoubtedly see Uze's design as competition.
c) dual-shock PSX joysticks are <$2/ea new from China, PSX mice are ~$3/ea... I've bought digital PSX joypads for $0.76/ea, new, in small quantities. (!) The only NES pads I know of in production at the moment are the Generation NEX wireless for ~$45. There's a SNES aftermarket pad for about $8/ea (cost), but the mating connector is a problem.
Basically if this turns into a "100+" volume market I don't much fancy relying on used NES pads and pulled connectors to try to build hardware. Just been burned by component shortages way too many times.
1) We need a "standard" board with all bell and whistles. A thumb could be nice afterwards.
Fair enough. (Although technically the 'standard' board could just be a carrier board with a little thumb/stamp type module on it.)
2) Lets stick to the simpler NES/SNES serial protocol.
(See above.)
3) A board should have the two NES PH connectors and also related empty through holes for direct wire soldering. It should have the requested D-sub connectors with two 4021. That will support in one shot genesis and Amiga/C64 joysticks and perhaps PSX with the help of your
adapter. (could it be on the board directly?). A jumper will be needed due to 5V not being on the same pin (I think).
Losing some of the elegance and packing on some costs there, IMHO. It's doable, but the PCB is growing and the costs increasing.
If you gotta have DB9's I'd suggest getting ones with plastic shells (otherwise the metal hoods will shave down the plastic joystick plugs and they'll cease to fit 'snugly' after a while). Unfortunately I don't have any sources for
cheap plastic hooded DB9's... More like ~$2.50/ea in 100's.
I admit that I'm not a fan of the serial shift registers really. I'll have to go look at the page discussing them and see if I can't get my enthusiasm up.
I already have an AVR assembly implementation of the PSX stuff. Maybe it could just run in game-code land and not in kernel time or something. (as long communications happen at about 60Hz things are OK-- most controllers have no 'time' requirement at all, but the wireless ones do.)
4) We need SD interface and ethernet with the ENC28J60/Magjack, composite, svideo and MIDI IN. There should be a jumper to disconnect MIDI from the UART to make it free for other uses.
Load MIDI on all boards then? (I wasn't including that in my earlier costs-- I figured that would be a "small minority" feature.) Are you saying ethernet is standard then too? (I fear we're rapidly approaching $100 retail price boards with all this)
5) A header to get access to most pins and a separate standard spi header socket
Affirmative. (Are you meaning the 6 pin AVR-ISP socket?)
6) I know you prefer SMT, but I strongly prefer DIP mounted on a socket for the MPU. For two good reasons: to replace it if it blows and to be able to upgrade to the new 1284P.
It's possible to go DIP MCU, but man... That's pretty much negating everything I was shooting for. (small size for lower PCB cost, parts commonality with other products I have in volume production to get the price breaks on components, little if any through-hole to keep automated assembly costs low-- throughole is a hand-op and as soon as a tech touches it we're adding dollars per board in labor). Do people really smoke MCU's that often? I had more people mangle DIP parts putting them in and taking them out than I've had SMT devices fail.
Here you go. I know this will raise the cost a bit, but I think we'll have a winner. And your price will be mine (and ours). If you want, I'm willing to buy you a batch to resell. Hope you're still interested?
Yeah, I don't want to push back too hard on your plan (it is *your* gizmo!), but that's really diverging from what I was hoping to do... Don't get me wrong, it's doable, but the up front costs will be higher (need to buy more 'new' parts in quantity), the PCB costs are higher (larger), assembly costs higher (hand-adds, probably need to run it over the wave instead of just the SMT oven), BOM costs higher (multiple connectors, more parts).
Maybe at this point we should agree to disagree? I can still do a design for you to your specs and get all the manufacturing package and stuff ready and then you can just pay for the parts and assembly on them (or just make 'DIY kits' if you prefer). I'd still go ahead with my smaller, SMT version and/or a stamp+carrier board instead.
-Clay