Glitchy sprites on FDS

[P]

It's hard to read the chips on that pic though.

Socketed just means he soldered chip-sockets in place of the chips so he can just snap in and out the chips so he don't have to solder them every time.

[oare]

I've done a few more tests swapping things around. Nothing great to report, but here are the results anyways:

- SONY CXK5864PN-15L (5I 64) and HITACHI HM6264P-20 SRAM with either Motorola MN4264-15 D-RAM and Fujitsu MB81416-12 D-RAM chips cause glitches when used along a Famicom that has an HVC-CPU-GPM-02 motherboard, no matter the configuration of the chips. All were taken from RAM Adapters with a revision number HVC-FMR-03.
However, any configuration of those chips display no problem whatsoever with any of my HVC-CPU-07 Famicom boards, nor with my AN-505 Twin Famicom.
The only chip both boards had in common initially is the hex inverter (BU4069UB).

- SONY CXK5864PN-15L (6I 12) with one single SHARP LH2833-15 D-RAM chip (board revision = HVC-FMR-04) causes no problems on any Famicom board.

The only visible difference between HVC-FMR-03 and 04 boards is the D-RAM: 03 boards have 4 chips, vs 1 chip only for 04 boards.
The next logical step would be to try and slap-in an 04 D-RAM chip in a 03 board. Rewiring shouldn't be very difficult since everything else appears pretty much identical. But I'm wondering if it's worth the hassle; it wouldn't be very practical anyways.

I'm wondering about those NEC D-RAM chips on FredJ's photo, though.

[fredJ]

Well, in the other thread, the guy tested and said the glitching went away when he pressed on the chips with his finger...
I think that could be an important clue.

If I press onto the solder points of the 4 dram chips with my finger while running the graphics corruption goes away.

If you want better photos, i'll ask my colleague. He is a member here too.

Maybe we could consider what the glitching is? It seems to affect objects that move, sprites that are moving or movable. Such as characters, balls, parts of airplanes, and such. What memory is that?

[P]

Sprites are copied from the disk to the CHR-RAM like all graphics. Depending on the game they might be copied from the CHR-RAM to a shadow register before they are sent to the PPU (and thus sent to the TV), but I think those registers are inside the Famicom so the problem must be with the CHR-RAM. But it's only a small fraction of the CHR-RAM so that's why only some games are affected (all games doesn't use the whole CHR-RAM).

Well, in the other thread, the guy tested and said the glitching went away when he pressed on the chips with his finger...
I think that could be an important clue.

That sounds like there's a bad connection somewhere, but if he recently soldereded them it can't just be old cracked solder that needs reflowing. Maybe he just didn't push the chips down into the sockets properly.

[oare]

Reflowing the pins was actually the first thing I tried, but to no avail.
Since I've desoldered and resoldered a few of those chips, there's pretty much only fresh solder on those joints now, too, but one can never be sure. Maybe we should try bending the D-RAM pins onto the solder pads and see how that goes?

The thing is, though: why would bad connections whithin the RAM adapter solely affect later revisions of the Famicom, but not cause any issue with hvc-cpu-07 boards?

I'm beginning to think we should start looking for causes not only in the RAM adapter only, but also in the Famicom themselves if we want to find a practical solution?

[80sFREAK]

Sure. There is PPU, CHRRAM and latch. Nothing else 

hint: something is slow

[P]

The way the 03 board is configured is causing some kind of conflict with newer Famicom boards? I know nothing about electronics so I have no idea.

But didn't someone say that replacing the RAM chips in a 03 RAM Adapter works? If that's the case maybe there's something weird with the chips that makes them behave weirdly in very specific situations (like when a newer Famicom board is involved)? I'm guessing wildly here.

[80sFREAK]

http://www.geocities.jp/atx197/FC_HVC-CPU-05_03.pdf here is redrawn diagram. Thanks to Junker.

IIRC some earlier PCB's utilized 74HC373 (which is much faster, but there also was small capacitor on "Clk"), but later revisions 74LS373.
Also eveone knows about different revisions of PPU(i have D, E, G in my collection somewhere in the boxes).
What else? The capacitor on /RD line(it is behind slot, somewhere in the middle).

At moment the fastest and easiest solution - replace CHRRAM chip in FDS adaptor with 120ns or faster. I would go at least 100ns.

[oare]

http://www.geocities.jp/atx197/FC_HVC-CPU-05_03.pdf here is redrawn diagram. Thanks to Junker.

IIRC some earlier PCB's utilized 74HC373 (which is much faster, but there also was small capacitor on "Clk"), but later revisions 74LS373.
Also eveone knows about different revisions of PPU(i have D, E, G in my collection somewhere in the boxes).
What else? The capacitor on /RD line(it is behind slot, somewhere in the middle).

At moment the fastest and easiest solution - replace CHRRAM chip in FDS adaptor with 120ns or faster. I would go at least 100ns.

That makes sense.

It seems that HVC-CPU(-01) to HVC-CPU-05 use 74LS373 chips, then 06 (maybe not all?), 07 and 08 boards use 74HC373 chips. And then HVC-CPU-GPM-01 and 02 use 74LS373 again.
Here are a few pics:
http://blog-imgs-38.fc2.com/o/f/f/offgao/IMG_0353_.jpg (HVC-CPU)
http://blog-imgs-38.fc2.com/o/f/f/offgao/HVC-CPU-06_OLD_FRONT.jpg (HVC-CPU-05)
http://blog-imgs-38.fc2.com/o/f/f/offgao/HVC-CPU-06_OLD_FRONT.jpg (HVC-CPU-06, interestingly the boards says "HC" but the chip is an "LS")
http://blog-imgs-38.fc2.com/o/f/f/offgao/img007.jpg (HVC-CPU-07)
http://blog-imgs-38.fc2.com/o/f/f/offgao/IMG_0288_.jpg (HVC-CPU-08)
http://blog-imgs-38.fc2.com/o/f/f/offgao/IMG_0483_.jpg (HVC-CPU-GPM-01)

I actually stumbled upon that exact site earlier today, there's an interesting schematic of the different revisions of HVC-FMR.
http://www.geocities.jp/atx197/FC_HVC-FMR-01-04_05.pdf
There  are a few more differences than what I thought when I took a quick glance to the boards themselves.
Notably, FMR-04 apparently has a connection to CLK from the D-RAM chip that is absent from the earlier revisions.

The SRAM in the FDS RAM adapter alone can't be the complete answer. I have an FMR-04 which has a CXK5864P-15 chip, and a FMR-03 which has a CXK5864PN-15L.
Both have the exact same response time (150ns), yet the FMR-04 works with HVC-CPU-GPM-02, and the FMR-03 doesn't.
So this LS/HC theory on the Famicom side definitely seems credible.

[80sFREAK]

The SRAM in the FDS RAM adapter alone can't be the complete answer. I have an FMR-04 which has a CXK5864P-15 chip, and a FMR-03 which has a CXK5864PN-15L.
Both have the exact same response time (150ns), yet the FMR-04 works with HVC-CPU-GPM-02, and the FMR-03 doesn't.
So this LS/HC theory on the Famicom side definitely seems credible.

And you forgot about date code  

[oare]

The dates, I gave in my previous posts (5I 64 for the FMR-03 CXK5864PN-15L chip  and 6I 12 for the FMR-04 CXK5864P-15 , if memory serves)
At this point, I'm really skeptical the 8K chip date and maker account for much, honestly.
I mean, at least 3 different makers (Sony, Hitachi and NEC) having a similar issue on chips produced over a 6 months span?
That seems unlikely.

Anyways...
Plot twist!

I took a look at my reference HVC-CPU-07 board. While the board says 74HC373, the chip itself is a 74LS373 (like the 06 board in the links given above).
Here's a pic.


I've looked at my two other 07 boards, they are both equipped with LS chips too, although the board itself says "HC".
So the practice appears to have at least been common.
Yet, all three work with the HVC-FMR-03 in all previously given configurations (Sony SRAM, Hitachi SRAM, Fujitsu D-RAM, Motorola D-RAM), whereas my ref. HVC-CPU-GPM-02 (obviously also equipped with an LS chip) doesn't.

Back to square one, I guess.
I'll definitely try that 100ns SRAM chip solution as soon as I can.
But IIRC there's apparently at least one user in the other thread who reports a failure with an even faster 80ns chip. So speed may not be the sole issue.

I'm really curious to know what exactly is causing those problems.
4 separate 16K D-RAM chips (FMR-03) vs 1 unique 32K D-RAM chip (FMR-04). A slightly different schematic. Different CPU revisions on the Famicom side. Different GPU revisions. Different motherboards in both the Famicoms and the RAM adapters, even. That's a lot of options...

Maybe it's just wear. But then again... Three different makers, over a span of 6 months at least? Plus a consistent working/non-working behavior depending on the Famicom board revision?
That's far fetched, at best.

[80sFREAK]

You can replace 74LS373 with 74HC373, but you have to add C21.

[fredJ]

My FMR-03s units both have a smooth plastic top enclosure, whereas my FMR-04s have a rough plastic top enclosure. I'm not sure how precise a visual clue that is, since there might be some overslap between the revisions. But if others can confirm this observation, it might be safer to try and avoid smooth plastic RAM adapters in general, especially if you own a Famicom that has the "FF" logo on the front panel.

Edit: a little update, I tried replacing the Sony S-RAM chip with a Hitachi chip (the "no-name" chip on FredJ's pic), and the glitches are still present.
I'll try replacing the D-RAM chips when I have some free time and see how that goes.
After all, those D-RAM pieces are the main difference between FMR-03 and FMR-04, so those might as well be the real culprit here.

I have now tried another 3 RAM adapters.
1 FMR-04, that worked fine.
1 FMR-03 with smooth plastic - worked fine
1 FMR-03 with rough plastic - glitched a little on one console. Not as much as my last glitching RAM adapter.

Funny thing is, I tried on two GPM-02 consoles. It only glitched on one of them


^^ rough plastic, some glitching on one GPM-02 console. (is playable)


^^ smooth plastic, no glitching


^^ BUT, my non-glitching famicom GPM-02 is of a late model. Serial no HC52xxxx. Though the chips appear identical to the one on http://famicomworld.com/workshop/tech/square-button-famicom/

It is probably some degenerating problem that affects some chips, in combination with some later consoles that might possibly also have degenerating chips. If they are degenerating, it isn't just because of bad chips but because of lots of use.

[oare]

I have now tried another 3 RAM adapters.
1 FMR-04, that worked fine.
1 FMR-03 with smooth plastic - worked fine
1 FMR-03 with rough plastic - glitched a little on one console. Not as much as my last glitching RAM adapter.
(...)

Excellent!
Thanks for the report.
The more evidence we gather, the easier it becomes to rule out hypotheses.

This definitely seems to point towards wear as a culprit, on at least one component.
This would be consistent with some of the posts in the other thread: one of the posters claims he possesses one FMR-03 with the 4 16KB D-RAM chips that "runs fine", while another one "causes glitches" (supposing he tried both on the same consoles).

I've tried to look for  clues on the Japanese side of the Internet. While some users just brushed off the issue as "a compatibility issues between some Famicom models and some RAM adapters", others were claiming their "RAM adapters went bad with time", seemingly implying they used the same RAM adapter on a unique console, and witnessed the degrading over the span of a few years. Which, again, would mean some manifestation of wear.
There is, however, very little information available, so nothing conclusive.

It might be components that go bad.
It might even be possible that the issue is entirely mechanical. Maybe its just the cartridge slot.
Or maybe it's components that need reflowing - but not always the same ones.
This would explain why 80'sFreak has had success with simply replacing the SRAM chip, while this solution proved useless for others.

I 'll try reflowing my 03 adapters' cartridge pins, it's pretty much the only area that has been left untouched.
It's a long shot, especially since I have already - unsuccessfully - tried pushing the pins while the device was on (during the loading sequence) in order to see if it was just a contacting issue.

In the meantime, if anybody can make more tests on a wider variety of machines, that would be really helpful.

[fredJ]

Maybe you could try replacing the S-RAM with a chip of 100ns or faster?
I think the Hitachi chip you tried is 120nS.

There is one in Dragon Quest V (SFC).

If there is degrading somewhere, maybe we need to increase the speed that data is passed through or calculated.