SNES/Famicon PCB Revisions and RGB Video

[PixelDharma]

I just got a 1chip (love it!) and I’m on the cusp of installing the brightness-fixing resistors, so I’m keenly listening to this conversation.

Let me know if there’s actually any accurate correction for the brightness on my Framemeister. I don’t own an OSSC.

[James-F]

Yes "A/D/ Level" is the pre ADC gain adjustment.
No need to mod your vintage 1chip whatsoever.

[FBX]

Yes "A/D/ Level" is the pre ADC gain adjustment.
No need to mod your vintage 1chip whatsoever.

I've made videos explaining exactly this in demonstration. The A/D function of the Framemeister allows you to correct any over or under brightness of the console. For example on my Neo Geo AES3-3, the brightness is under spec, and moving the A/D level up using the color bars test pattern from the port of Artemio's Test Suite perfectly corrected the issue.

In the case of the OSSC, you can use the 'gain' settings to do the same thing, and even to the point of perfectly balancing the RGB levels.

[James-F]

In the case of the OSSC, you can use the 'gain' settings to do the same thing, and even to the point of perfectly balancing the RGB levels.

Not exactly, the OSSC currently in the official v0.80 firmware don't have Pre ADC Gain controls, the Fine Gain in the current firmware are Post ADC and do not fix already clipped signal.
The whole point of using Pre-ADC-Gain control is to fix the hotter 1chip signal WITHOUT modifying the console with 1.2k/750 resistors.

[TheShadowRunner]

Again, not convinced all 1CHIPs are affected..

[James-F]

Most are, that's where you should use the 240p test suite with the grey ramp, and adjust the pre adc gain.

[TheShadowRunner]

Most are, that's where you should use the 240p test suite with the grey ramp, and adjust the pre adc gain.

Oh it's acknowledged, alright. I thought all 1CHIPs were thought to be affected.

[SamIAm]

I've now got four Super Famicoms in my posession. Is this serial-number document still being updated? Because if it is, I'll share.

https://docs.google.com/spreadsheets/d/ ... utput=html

[James-F]

In FBX signature you'll find the biggest SNES Serial DB.
If your numbers are between others numbers from the DB with the same PCB model, no need to make additions.
If your number is an "edge" number or changes the PCB model, please do.

[FBX]

I've now got four Super Famicoms in my posession. Is this serial-number document still being updated? Because if it is, I'll share.

Let me have a look at your numbers, they might be valuable even if not an edge case.

[SamIAm]

Let me have a look at your numbers, they might be valuable even if not an edge case.

SM12053538 - SNS-CPU-1CHIP-02
S24864095 - SNS-CPU-APU-01
S24697281 - SNS-CPU-APU-01
S20325832 - SNS-CPU-GPM-02

There are also a bunch of 1CHIP systems on yahoo auctions that give the board revision and serial number. I can compile a list of those later, if you like. I recall one interesting case of a number starting with S247 that was a 1CHIP.

[FBX]

Let me have a look at your numbers, they might be valuable even if not an edge case.

SM12053538 - SNS-CPU-1CHIP-02
S24864095 - SNS-CPU-APU-01
S24697281 - SNS-CPU-APU-01
S20325832 - SNS-CPU-GPM-02

There are also a bunch of 1CHIP systems on yahoo auctions that give the board revision and serial number. I can compile a list of those later, if you like. I recall one interesting case of a number starting with S247 that was a 1CHIP.

Okay I entered them in. If you feel up to it, try to look for edge cases (i.e. the given board revision lands on the first or last serial of that batch in the spreadsheet) and report those you find to me in PM. Thanks!

[SamIAm]

Sure!

By the way, the link in your sig doesn't seem to work. I wonder if it's just me?

EDIT: Never mind! It's working now.

[SamIAm]

From Yahoo auctions, board verified with picture unless otherwise noted:

S24758322 SNS-CPU-1CHIP-?? (picture doesn't show board number completely, but is definitely a 1CHIP)
S2477**** SNS-CPU-1CHIP-01 (seller blocked last four numbers)
S2478**** SNS-CPU-1CHIP-01 (seller blocked last four numbers)
S24785142 SNS-CPU-1CHIP-01
S24785532 SNS-CPU-1CHIP-01
S248***** SNS-CPU-1CHIP-01 (seller blocked last five numbers)
S24803981 SNS-CPU-1CHIP-01
S24837059 SNS-CPU-1CHIP-01
S249***** SNS-CPU-1CHIP-01 (seller blocked last five numbers)
S24930849 SNS-CPU-1CHIP-01
S24956781 SNS-CPU-1CHIP-01
S25016619 SNS-CPU-1CHIP-?? (no board picture, no description)
S25058109 SNS-CPU-1CHIP-02
S25059856 SNS-CPU-1CHIP-02
S25077259 SNS-CPU-1CHIP-02
S25110627 SNS-CPU-1CHIP-02
S25120041 SNS-CPU-1CHIP-02
S25124870 SNS-CPU-1CHIP-02
S25150906 SNS-CPU-1CHIP-02
S25187826 SNS-CPU-1CHIP-03
S25202122 SNS-CPU-1CHIP-03 (no picture of board, but seller looks knowledgeable)
S25236322 SNS-CPU-1CHIP-03
S25237127 SNS-CPU-1CHIP-03


SM11938794 SNS-CPU-1CHIP-02
SM11978981 SNS-CPU-1CHIP-01 (not verified in picture, but seller has sold multiple 1CHIPs)
SM11984464 SNS-CPU-1CHIP-01
SM12036231 SNS-CPU-1CHIP-?? (seller not specific, no picture)
SM12041567 SNS-CPU-1CHIP-02


I'm rather suspicious of a couple of auctions where the seller claims that the SFC is a 1CHIP, yet does not show pictures of the inside of the console. No information is given about the board revisions, either. The one starting with S246 would be the lowest 1CHIP SFC.

S24618767
S24781610


EDIT: I found this while searching the forums. It's an interesting overlap for the 1CHIP-02.
Serial #: S25200413
SNS-CPU-1CHIP-02
viewtopic.php?f=6&t=43689&start=120

[FBX]

Entered all but a couple from your list that didn't have enough information. It does resolve the SFC manufacturing process more clearly: APU and 1CHIP-01 were randomly intermingled. However, by the time the 1CHIP-02 line starts, the APU boards vanish. I get the feeling there was a time period where they stopped production until they cleared the plants of APU and -01 consoles, then started fresh with the -02 batch.

The couple of overlaps of the -02 into the -03 line just indicates an on-running transitional period into the -03 line, which of course all they did really was pinch a few pennies by omitting the components for csync.

[SamIAm]

Great! The more data the merrier, as far as I'm concerned.

I recently gambled and lost on one of those APU systems. It's such a shame that those are possibly the single worst of all SFC models in terms of quality. Until I get one more 1CHIP as a backup, it's S25 and up only unless they're practically giving it away.

I wonder what the story is with the SM line?

[paulb_nl]

Here are the SM serial numbers I collected when searching for a 1-CHIP Super Famicom on ebay:

SM11976966 1-CHIP-01
SM11977319 1-CHIP-01
SM11977939 1-CHIP-01
SM11982226 1-CHIP-01
SM11987450 1-CHIP-01
SM12036981 1-CHIP-02
SM12041209 1-CHIP-02
SM12070116 1-CHIP-03

I was thinking that the SM line was made in a different factory and it seems to be the case because SM says made in China on the bottom and the S line says made in Japan.

[SamIAm]

Made in China, eh? Now we've got to compare them!

Thanks for the numbers.

[SamIAm]

I just tried the following on one of my APU-01 boards:

-Removed R59
-Removed RF unit
-Replaced 7805 regulator
-Added 1uf ceramic caps at base of 7805 to decouple input/output (datasheet says you should use one at the input)
-Added 220uf OS-CON capacitor to decouple power rail.

I was using the FF6 title screen as a test. Unmodified, this APU board has diagonal-line noise, a pronounced vertical stripe, and weird artifacts showing up in the top area of the screen when the red FF6 logo appears.

None of the first four steps above made any appreciable difference. The diagonal noise was still there, although it may have become slightly less pronounced. The addition of the OS-CON reduced the vertical stripe only.

The most curious thing is that this APU board has two unpopulated solder pads behind the cart-connector for a capacitor to decouple the 5V power rail, and they are literally marked "OS-CON". I had picked up an OS-CON beforehand because I know how good they are at decoupling, so this was a surprise coincidence. Could it be that Nintendo prepared this pad for a serviceman to add an OS-CON in the event that someone sent their system in because of poor video quality?

Googling the board, I see that 1CHIP systems seem to also have unpopulated pads marked OS-CON. At a glance, I don't see one on the RGB-01 board or the GPM-02 board.

Anyway, I'm not sure what else to do with this APU-01. Any suggestions?

[FBX]

Anyway, I'm not sure what else to do with this APU-01. Any suggestions?

APU is THE WORST version of the SNES console to have because of the diagonal subcarrier noise interference. Voultar found the fix by lifting pin #3 on the PPU2 chip, and I did this myself to test on my own APU (I kept it even though I have two different 1CHIP consoles). Pics of my mod and results:





Be warned you of course lose color on composite video output, but that's a small price to pay for cleaner RGB.

[SamIAm]

Thanks for the pics, but doesn't removing R59 accomplish the same thing as lifting the pin 3?

[FBX]

Thanks for the pics, but doesn't removing R59 accomplish the same thing as lifting the pin 3?

If R59 is much later on in the trace, the noise in the trace might still be affecting RGB before it gets to that point. By lifting pin 3 on the PPU2, you eliminate the noise at the source.

Edit: Took my APU apart to have a look at the location of R59, and yeah, I'm betting you're still getting interference from that video output clock on pin 3. It's in very close proximity to the RGB lines well before it even gets to R59.

[James-F]

I also keep an original 2chip (GPM-02) for tests and comparisons, but use a SFC 1chip-02.
A 2chip might not be the best looking snes, but the DMA ,HDMA, Memory, CPU, bus timing in general are as intended by the original design, no integration, no shortcuts, no glitches or bugs.
Like it or not, a 2chip is still the most accurate snes around, and may show little difference to the 1chip on a consumer CRT with Composite/S-Video which 99.99% of non-fanatics in the 90s used to use.
RGB, emulation-like pixels, upscalers, BVMs, became a 'thing' in the past maybe 4-5 years by youtubers, no one noticed or cared for a little smear on a 2chip a few years back.
0chip,1chip,2chip,3chip,,, it's still the best console of all time though...

[SamIAm]

Thanks for the pics, but doesn't removing R59 accomplish the same thing as lifting the pin 3?

If R59 is much later on in the trace, the noise in the trace might still be affecting RGB before it gets to that point. By lifting pin 3 on the PPU2, you eliminate the noise at the source.

Edit: Took my APU apart to have a look at the location of R59, and yeah, I'm betting you're still getting interference from that video output clock on pin 3. It's in very close proximity to the RGB lines well before it even gets to R59.

Thanks again for looking into this.

Someone correct me if I'm wrong, but it seems like AC coupling shouldn't be possible if the line is cut regardless of where it's cut.

My understanding is that it's current flowing through a wire which causes a magnetic field to build up around it. The size of that field is dependent on the amount of current flowing, and if the amount of current fluctuates with rises and falls in voltage (as in a signal), the magnetic field will build and collapse accordingly. It's the collapsing of that field into a neighboring wire that gives the neighbor a momentarily boosted voltage, and that is AC coupling. Without current flowing, fluctuating, and causing magnetic field collapse, the effect doesn't happen. Voltage potential on a wire alone shouldn't mean much to any neighbors if it doesn't have anywhere to go.

Or should it?

I also keep an original 2chip (GPM-02) for tests and comparisons, but use a SFC 1chip-02.
A 2chip might not be the best looking snes, but the DMA ,HDMA, Memory, CPU, bus timing in general are as intended by the original design, no integration, no shortcuts, no glitches or bugs.
Like it or not, a 2chip is still the most accurate snes around, and may show little difference to the 1chip on a consumer CRT with Composite/S-Video which 99.99% of non-fanatics in the 90s used to use.
RGB, emulation-like pixels, upscalers, BVMs, became a 'thing' in the past maybe 4-5 years by youtubers, no one noticed or cared for a little smear on a 2chip a few years back.
0chip,1chip,2chip,3chip,,, it's still the best console of all time though...

If I can just get the vertical bar out of my GPM-02, I might use it as my main system. The level of blur otherwise doesn't bother me, and I do appreciate the accuracy.

Since I agree about the best-console part, I just want to have enough back-ups to keep playing however I like until I'm too old and doddery for video games in general.

[James-F]

Vertical line is non existent if I use the SD2SNES, but clearly visible with original carts.

I think it's more related to filtering the 5v bus at higher frequencies, so a few small ceramic capacitors 100nf, 1nF and 100pF do a much better job than a single 470uF.
At least that's what I think that's happening when using the sd2snes, it has filtering cap near every IC, a job of a good engineer clearly and how it should be done.

[SamIAm]

I was just thinking that the next thing I would like to try is adding more small-value ceramic capacitors around the APU-01 board to soak up high frequency noise.

The OS-CON I added should be better at catching noise above 500khz than typical aluminum electrolytics, but it's still not as good at >1mhz noise as ceramics. It's also probably not optimally positioned.

EDIT: Hey hey, adding a 100nf ceramic across the OS-CON and another in the power/ground holes left by the RF unit seems to have improved things. I haven't been measuring things properly, so it's hard to say what's actually helping, but at this point I'll be damned if the level of interference is anywhere near what it was before I had done anything. I've got boatloads of 100nf ceramic caps, so I'll see where else on the board I can add them as decouplers.

FBX, if you have the time and the interest, you should check the FF6 title screen with your APU-01. I just loaded up the LttP title screen on mine and it was crystal-clear, but the FF6 screen still has the problems I described. My GPM-02 and 1CHIP-02 are much, much more stable.

[FBX]

Thanks again for looking into this.

Someone correct me if I'm wrong, but it seems like AC coupling shouldn't be possible if the line is cut regardless of where it's cut.

My understanding is that it's current flowing through a wire which causes a magnetic field to build up around it. The size of that field is dependent on the amount of current flowing, and if the amount of current fluctuates with rises and falls in voltage (as in a signal), the magnetic field will build and collapse accordingly. It's the collapsing of that field into a neighboring wire that gives the neighbor a momentarily boosted voltage, and that is AC coupling. Without current flowing, fluctuating, and causing magnetic field collapse, the effect doesn't happen. Voltage potential on a wire alone shouldn't mean much to any neighbors if it doesn't have anywhere to go.

Or should it?

Well I can only give you anecdotal evidence in the example of early model 1 Sega Genesis consoles. Jail bars were hitting the blue trace from the subcarrier trace right next to it. Lifting pin 6 on the CXA was reportedly giving less than optimal clearance of the noise, while severing the trace on the VDP side completely eliminated the noise. I don't know if it's for a fact or not because I always cut the trace on the VDP side myself. It's just something I heard anyway about the CXA side.. If your theory is correct, then lifting pin 3 won't help, but as you can see from my before and after pics, it made a world of difference, as pointed out by Voultar.

When I get some free time, I'll take a look at the stripe issue on my APU, although I did replace the regulator on it with a 78S05, so I may not have any stripe to show.

[SamIAm]

Well I can only give you anecdotal evidence in the example of early model 1 Sega Genesis consoles. Jail bars were hitting the blue trace from the subcarrier trace right next to it. Lifting pin 6 on the CXA was reportedly giving less than optimal clearance of the noise, while severing the trace on the VDP side completely eliminated the noise. I don't know if it's for a fact or not because I always cut the trace on the VDP side myself. It's just something I heard anyway about the CXA side..

The stories of people getting better results by lifting the subcarrier trace at the VDP rather than the CXA of a Genesis are intriguing. That there is a difference really doesn't make much sense to me. I've been learning a lot about inductance recently, and it really seems like no current flow = no magnetic field = no effect on nearby circuits.

For the record, I had bad jailbars coming from my X'EYE, cut pin 6 of the CXA right at the chip, and solved the problem completely.

If your theory is correct, then lifting pin 3 won't help, but as you can see from my before and after pics, it made a world of difference, as pointed out by Voultar.

To be clear, I am saying that lifting pin 3 and removing R59 should have an equal effect, because both solutions stop current from flowing and prevent a typical AC coupling scenario from happening. In other words, I don't think that cutting pin 3 after having removed R59 will lead to any further reduction in the noise that I am seeing. I guess I could try it, though.

When I get some free time, I'll take a look at the stripe issue on my APU, although I did replace the regulator on it with a 78S05, so I may not have any stripe to show.

I also replaced my APU's regulator and have mostly eliminated the big vertical stripe. What I haven't eliminated is a minor amount of diagonal noise. Again, I'd suggest looking at the FF6 title screen instead of LttP, because with all of my modifications so far, I get a perfect LttP title screen yet a slightly fuzzy FF6 title screen. I also still get weird artifacts near the top of the screen when the main FF6 logo appears.

[Unseen]

Someone correct me if I'm wrong, but it seems like AC coupling shouldn't be possible if the line is cut regardless of where it's cut.

My understanding is that it's current flowing through a wire which causes a magnetic field to build up around it. The size of that field is dependent on the amount of current flowing, and if the amount of current fluctuates with rises and falls in voltage (as in a signal), the magnetic field will build and collapse accordingly. It's the collapsing of that field into a neighboring wire that gives the neighbor a momentarily boosted voltage, and that is AC coupling. Without current flowing, fluctuating, and causing magnetic field collapse, the effect doesn't happen. Voltage potential on a wire alone shouldn't mean much to any neighbors if it doesn't have anywhere to go.

Or should it?

AC signals don't just couple to other signals by induction, they can also couple by the mutual capacitance of the traces that are carrying them.

[paulb_nl]

I also replaced my APU's regulator and have mostly eliminated the big vertical stripe. What I haven't eliminated is a minor amount of diagonal noise. Again, I'd suggest looking at the FF6 title screen instead of LttP, because with all of my modifications so far, I get a perfect LttP title screen yet a slightly fuzzy FF6 title screen. I also still get weird artifacts near the top of the screen when the main FF6 logo appears.

I think the artifacts you are getting is CPU activity noise. Its more easily seen with the OSSC 256x240 optimized mode. I have captured a video with reverse LPF set to max so the noise can be clearly seen: https://youtu.be/syOW-w1q5UA

If you take a look at this video where they visualize the CPU instructions timing you can see the noise looks exactly like that. https://youtu.be/Q8ph2OVqZeM?t=8m18s

There is no noise at the DRAM refresh in the middle where the CPU is halted. Also what they call spinning where the CPU is waiting for something looks the same too with the diagonal lines. Once the game is waiting for controller input at the save file selection the diagonal bars go across the whole screen.