About capacitors on SNES boards

[magno]

As I have told several times in this forum, I'm trying to make my own SNES boards for developing purposes. I recently sent to production a lot of 10 boards and I managed to load my own code on the EPROMs in them.
But now I have soldered a PIC12F629 to those boards and the games run perfectly on my Super Famicom (as they did before adding it using an adaptor) but they don't in any of my 5 SNES PAL decks. Sometimes the game freezes randomly, sometimes it restarts, some others I lose any video output from the console desk. And those freezes occur when I've been playing 10 minuts, 50 minuts or a few seconds.

Since it sounded strange to me that the game runs fine on a Super Famicom, I recalled that Super Famicom has DC power inpunt, whereas PAL SNES has AC power input. I thought then that original boards have one capacitor around each chip on the board and then concluded that some AC power input noise was affecting +5V on my boards and made some data corrupt. This AC power noise was not present on Super Famicom and that was the reason the boards worked there. I used 10nF capacitors but... does anybody know which capacity have the ones found in the original boards? Do you think this decoupling capacitors are really necessary?

The fact is that those freezes have disappeared with the 10nF capacitors but the problem has been partially fixed: now sometimes the SNES reset itself again and I lose video output from the console, so I guess CIC's reset is being triggered, and that can only mean that SuperCIC is not giving the proper bit sequence and SNES lock CIC is resetting the CPU. However, this DOESN'T EVER happen on a Super Famicom, only in PAL SNES. What could be happening?
Maybe some noise coupled again on the PIC? Has anybody got some schematics with the proper PIC connection to the cartridge board?

Thanks!

[qwertymodo]

I believe the ones on the official boards are 22uF (or maybe 2.2, I don't remember). I would check the datasheets for the chips that you are using and use the size recommended by the datasheet. 100nF is pretty typical, but it can vary from chip to chip. A good rule of thumb is 1 capacitor per chip or 1 capacitor per Vcc pin for chips with more than 1 Vcc pin, and place it physically close to the Vcc pin. The 22uF cap is probably a decoupling cap for the entire board, which is probably a good idea as well. You might also want pull-up/pull-down resistors (10K) on chip enable lines to keep the chips disabled at startup until the system stabilizes. A pull-up resistor on the reset line might solve the reset issue.

[magno]

I believe the ones on the official boards are 22uF (or maybe 2.2, I don't remember). I would check the datasheets for the chips that you are using and use the size recommended by the datasheet.

The electrlytic capacitor in all cartridge boards is 22 uF. But I was talking about those pink capacitors that look like resistors:



They are labelled as C2, C3 and C4. I think they are 10nF, but not sure, and don't know if they are really necessary.

100nF is pretty typical, but it can vary from chip to chip. A good rule of thumb is 1 capacitor per chip or 1 capacitor per Vcc pin for chips with more than 1 Vcc pin, and place it physically close to the Vcc pin. The 22uF cap is probably a decoupling cap for the entire board, which is probably a good idea as well.

I knew that rule for high profile chips: processors, PICs, FPGAs, CPLD, DSP, LIUs or anything with clocks... but I though that was not the same when using EPROMs and SRAMs, purely asynchronous.

You might also want pull-up/pull-down resistors (10K) on chip enable lines to keep the chips disabled at startup until the system stabilizes. A pull-up resistor on the reset line might solve the reset issue.

I guess the reset is asserted by the SNES, not by the cartridge, since the lock CIC is on the deck and it asserts the /RESET when no key CIC is present on the cartridge. So I think the pull-up in the cartridge side won't solve the problem...

[qwertymodo]

From looking at a few PCB scans, those capacitors are 10nF and 1nF (black/brown/yellow/x/x is 10nF, black/brown/orange/x/x is 1 nF). Like I said, read the datasheets for your chips. It differs per chip, but most SRAM chips I've seen have at least suggested smoothing caps. As for the RESET line, thinking about it now, it definitely makes sense that the console would drive it rather than leaving it open, but for the /CE lines on the ROM and SRAM, I would definitely pull those up if you aren't already.

[magno]

From looking at a few PCB scans, those capacitors are 10nF and 1nF (black/brown/yellow/x/x is 10nF, black/brown/orange/x/x is 1 nF). Like I said, read the datasheets for your chips. It differs per chip, but most SRAM chips I've seen have at least suggested smoothing caps. As for the RESET line, thinking about it now, it definitely makes sense that the console would drive it rather than leaving it open, but for the /CE lines on the ROM and SRAM, I would definitely pull those up if you aren't already.

There is not more free space left on my board, so I won't be able to add those pull-ups... Anyway, I don't think they are necessary since they are not present on Nintendo's boards, and I don't see it related to the reset issue... The console keeps resetting and I don't know if it's due to Vcc noise, bad PIC12F629 programming, poor contact on the cartridge edge........... I don't knew what to do...
Thanks for your advice anyway, qwertymodo

[magno]

I solved the problem using 100nF decoupling capacitors.

Finally my board has not any pull-up, one 100nF ceramic capacitor for decoupling near each 27C322, another 100nF cap for SuperCIC, and one 22 uF electrolytic capacitor near Vcc to decouple noise in the whole board.

[benjaminsantiago]

Hey magno, do you have any pictures of the carts you've made anywhere? Just curious

[magno]

Hey magno, do you have any pictures of the carts you've made anywhere? Just curious

Yes

HiROM (64 Megs):



LoROM (32 Megs):

[getafixx]

is that 32mbit cart unfinished? You seem to be missing your '257s.

[benjaminsantiago]

These look great, I like the ad for buyicnow.com

What is going on with those discrete components under the batteries?

Thanks for the info on the caps as well magno.

[magno]

is that 32mbit cart unfinished? You seem to be missing your '257s.

Not, not missing them. When mounting a 16Mbit game, you can avoid mounting both multiplexers if you re-arrange the data in the EPROM so the data outputs from D0 to D7 only. The data is shifted and A20 becomes A0, thus the floating wire you can see in the board.

These look great, I like the ad for buyicnow.com

It's not an ad XDD They programmed the chips and they put the label in the EPROM to make clear what is burned in each one. They shipped the EPROMs the way you see.

What is going on with those discrete components under the batteries?

It's the SRAM-standby trigger circuit: it drives /CE and VCC_SRAM to keep the data in the SRAM when powering-off the SNES.

[getafixx]

Not, not missing them. When mounting a 16Mbit game, you can avoid mounting both multiplexers if you re-arrange the data in the EPROM so the data outputs from D0 to D7 only. The data is shifted and A20 becomes A0, thus the floating wire you can see in the board.

Ahh ok, I thought it was a 32mbit game on there.