Super Famicom VRAM issue

[kamiboy]

Okay, I've had this defective SFC for a long time now and it is high time I get to the bottom of what is wrong with it.

Symptoms are graphical glitches in certain games, seems to be mostly related to the background planes. The test above led me to think that one of the two VRAM modules on the motherboard were the culprit. I first tried a reflow, then did continuity tests, and lastly I outright replaced both chips.

None of the above led to a solution. At this point I am ruling out the VRAM chips themselves, the issue must be something else and as per the screen above I am thinking the problem surely must lie with the DMA functionality. I am way above my head here though.

I could continue doing continuity checks on the CPU and GPU at random but with so many tiny legs the chances of mistakes is huge. I need to figure out which traces on the motherboard I should check that relate to the DMA functionality, specially DMA transfers to the VRAM.

Anyone care to shed some light on this matter?

[korpse413]

I'm afraid I can't exactly shed any light, just speculate. What model board is your SNES? Not sure if there is a trend with the original model getting fried easily, as the one I describe my issues with is SHVC from 1990.

These are issues I am having with my SNES which I 'think' are related to VRam. I discovered that test suite rom from your post above and might give it a try with my buddies SD2Snes, but as we can see from my photos I'm afraid I won't be able to make out anything due to the garble. People seem to think my issue is with the Picture Processing Unit (PPU), where others the CPU. May I ask your method for replacing chips/reflowing? I've heard reflowing with a soldering iron is a no-no, so maybe I'll toss it in a controlled toaster oven for a few seconds at around 200 degrees. Do you own a hot air gun (not sure if I need to invest money in one of these, not sure if I'll even use it again)?

Sorry if I haven't exactly helped, but maybe if my symptoms are truly a VRam issue we could dismiss yours as VRam and really PPU? (or vise versa?)

[kamiboy]

My board revision is from '94, a SNS-CPU RGB-01.

As for reflow, i just use a soldering iron, same thing for replacing the RAM modules. The legs on the VRAM chips are far enough apart as to make it feasible. The PPU and CPU though, not so much.

[kamiboy]

Okay, today I decided on new and simple approach. I decided to measure voltages on each pin of the U4 and U5 S-RAM chips.

This has uncovered a discrepancy. On U4 pins 13 and 17 receive 5 mV whereas on U5 these pins are supplied with 5V.

These two pins are IO pins and every other IO pin receives 5V so I think that is the problem. Both these pins are connected to pins on PPU1, which then would have to be the source of the problem since the traces are not broken.

Could I trouble any of you to help me confirm this?

Just fire up your SNES without a cart and measure voltages on pins 13 and 17 on the chip labeled U4 on the motherboard and report back.

[mvsfan]

i tested both u4 and u5 on my Gpm board and they both read 5v on pin 13 and 17.

[kamiboy]

Well, there you go then, much obliged of course.

It seems there is something wrong with PPU1 then as it bears the two pins connected to 13 & 17 on U4.

My knowledge of electronics cannot carry me any further than that. I suppose there must be some sort of malfuction inside of the PPU1 chip. Though I cannot rule out that the problem might might be sourced elsewhere.

I see on the circuit diagram that four 5V lines are supposed to be fed into PPU1, so a last desperate hope might be that the problem lies there. Alas, if not there remains naught but to replace PPU1 or just give up.

I lean towards the latter.

[mvsfan]

if you use a can of flux and lightly brush some on to the pads on the board as well as dip the very end of the legs on the chip in it, you can use a soldering iron no problem. It keeps the solder from sticking the pins together. You dont need hardly any solder with surface mount chips a little bit goes a long way.

I tack 1 pin down on each corner of the chip to make sure it stays properly aligned, add a bit of solder to the tip of the iron and then i just drag it across the pins.

to reflow, just add flux (solder usually isnt needed) and then drag your iron across the pins.

and I also use a fine pencil tip on the soldering iron.


to get surface mount Chips off of a board I use hot air with success. I heat up all the pins evenly then use a pick to lift it off.

[kamiboy]

Ah but before I can even bother to think of such details I would need to source a replacement PPU1, which, unlike S-RAM modules, is a custom chip, so the replacement would need to come off of another SNES.

Not sure it is worth the trouble, really. I do all my playing on a RGB modded Jr anyway.

[CkRtech]

Pin 13 and 17 aren't Vcc lines. They are data lines. They aren't meant to have a constant reading - the values are going to change during the read/write cycles.

You may want to continuity check all of your traces to each pin of both of the ram chips (and of course all work should begin with a clean cartridge and a clean cart slot).

[mvsfan]

well, thats true too. you can still find several snes parts new but not the cpu/ppu. it might not be worth repairing if they are bad.

[kamiboy]

Pin 13 and 17 aren't Vcc lines. They are data lines. They aren't meant to have a constant reading - the values are going to change during the read/write cycles.

You may want to continuity check all of your traces to each pin of both of the ram chips (and of course all work should begin with a clean cartridge and a clean cart slot).

I know they are data lines, but I think they are supposed to be between 5 and 2.2 V, one being high, the other low.

I do not think they are ever supposed to be in the mV range. For the record, I did measure them several times, all the other IO lines were solid 5V across the line.

[Unseen]

I know they are data lines, but I think they are supposed to be between 5 and 2.2 V, one being high, the other low.

Nope. Depending on the type of input of the chip, a 5V-powered system would read 2.2V either as high (classic TTL) or barely-below-the-threshold-but-probably-low (CMOS logic).

I do not think they are ever supposed to be in the mV range.

Sure they are. If they really were 2.2V over a longer time span (milliseconds or more) tht would point to a possible problem in the circuit.

For the record, I did measure them several times, all the other IO lines were solid 5V across the line.

For lines that are active (changing their value), measuring their voltage with a multimeter is not very useful.

[kamiboy]

Well, I suppose I need a logic probe of some sort to be sure then. Alas I am done investing money in this repair project.

I do have access to two other SNES systems though. If I fire those up without a cart inserted and measure 5V on those IO pins I am going to go ahead and conclude that I have located the problem, damn be whether measuring the voltage of those pins is technically sound or not.

[SukkoPera]

Well, I suppose I need a logic probe of some sort to be sure then. Alas I am done investing money in this repair project.

Well, if you feel... adventurous, google for DSO138 .

[mvsfan]

btw, I dont fix every snes. when i run into problems that are more effort than is worth my time to locate and fix it, wich is kinda what it sounds like your having with this one,

It ends up being parts unless its a 1-chip.

The 1-chips that arent fixable end up going on a shelf, complete in the hope that i will eventually come across the parts i need.

although i was upset that i just bought one recently that had been underwater at one point. it was full of sand and corrosion. it doesnt happen often but those are useless.

[kamiboy]

Well, I do not know what you intend to do with the SNES' that you do not deem worthy of ever fixing, but if you ever end up with a spare PPU1 for cheap do let me know.

[mvsfan]

Im not sure if they still work. the few snes i have that arent 1-chips ive been picking parts off of.

before i put them on the shelf they were completely dead and it wasnt the fuse or the 7805. its just a chance that parts work or dont.