Famicom AV Mod Situation in 2019

Discuss hardware-related topics, such as development cartridges, CopyNES, PowerPak, EPROMs, or whatever.

Moderators: B00daW, Moderators

tepples
Posts: 21752
Joined: Sun Sep 19, 2004 11:12 pm
Location: NE Indiana, USA (NTSC)
Contact:

Re: Famicom AV Mod Situation in 2019

Post by tepples » Mon Apr 15, 2019 9:44 pm

Ben Boldt wrote:Has anyone run the BN-ROM version on real hardware before? I am using 240pee-0.16.
I've run it on a PowerPak but not on a single-game cart. I've built a release candidate of 0.18, but that may not fix it.
Attachments
240pee-bnrom-0.18rc1.zip
(27.94 KiB) Downloaded 465 times

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Mon Apr 15, 2019 10:50 pm

I put it on there and it is slightly different. I uploaded a video so you can see how it flickers:

https://youtu.be/-aoXixJZXTY

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Tue Apr 16, 2019 9:12 am

It works great in the Famicom AV mod test unit...
IMG_1685.JPG
  • Hi-Def NES (USA Front Loader) -> Blows up, loud electrical snap noise right after I pressed the power button, then dead
  • Unmodified USA Front Loader -> Sprite glitches
  • Original Famicom -> Perfect
I did take apart the RF box on the Hi-Def NES just last week but also did power it on several times after reassembling. So I am not sure if it is related or not.

I also tried 2 different 72 <-> 60 pin cartridge adapters, both did the same thing in the unmodified front loader.

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Tue Apr 16, 2019 6:05 pm

I find that the "color bars on gray" test makes the jailbars the most visible.

Here is a list of things I tried today, nothing of which made any observable difference:
  • Cheezy video cable with and without ferrite bead (video quality was worse with the cable but unaffected by ferrite bead)
  • Removed the 74LS373, installed socket, and changed to a 74HCT373
  • Completely removed the RF box and powered the famicom directly with 5V bench supply
  • Removed the PPU, installed socket, and did not connect video pin of socket to motherboard
  • Inserted the ferrite bead first thing in the circuit, between PPU and base of transistor
  • Shrunk the whole mod circuit -- used surface mount resistors and 560p very tightly on the bottom of the board
  • Wrapping the PPU in copper foil, foil not connected to anything
  • Connected the foil to all 5 GND pins (pin 20 and 14, 15, 16, 17)
  • Added ceramic bypass caps directly from +5V pins (pin 40 and 22) directly to copper foil
  • Sprinkled ceramic bypass caps all over the place on the bottom of the whole board
I also inserted a 3.3uH inductor at the very end of the circuit, i.e. pass the composite video through it. It made a slight difference but not an improvement.

At the end of all of that, I went back to Super Mario Bros. with all of these changes in place, I saw no difference from before. I am kind of out of things to try now, definitely open to trying any more ideas; I have only blown up 1 Nintendo this week. :lol:

So far, this is the best AV Mod circuit, and no crazy foil or cutting pins:

Code: Select all

                     (GND)
PPU.20 ----------------+---------------------------------+--------+
 |                     |                                 |        |  Ring
 |                  | / (c)  2SA937               560pF ---       +----O } Composite
 |         PPU.21 --|< PNP                      Ceramic ---              }   Video
 |                  | ^ (e)      220uF                   |        +----O }
--- 4.7-47uF           |     Electrolytic     110        |        |   Pin
--- Tantalum           +----------|(---------/\/\/-------+--------+
 |+                    |        +
 |   (+5V)     300     |
PPU.22 -------/\/\/----+

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Tue Apr 16, 2019 7:24 pm

One final copper foil test just to really put the nail in the coffin, since my PPU is in a socket and thus farther from the PCB, I put adhesive copper foil directly on the bottom of the PCB and under the whole mod circuit:
IMG_1689.JPG
No visible difference. We can call this myth busted.

User avatar
Ultron
Posts: 21
Joined: Tue Feb 26, 2019 6:36 pm

Re: Famicom AV Mod Situation in 2019

Post by Ultron » Fri Apr 19, 2019 11:42 am

I had some time today to do some testing.

I modded my top loader NES-101 to use the same circuit as the AV Famicom. Here is the stock video circuit for the NES-101:
NES-101 Video.png

I replaced R1 with a 310 ohm resistor, and removed R2 and R3. In R2's spot, I connected a 220uF cap and FC2 in series.
Cap and Bead.png
Then, I installed a 100 ohm resistor in FC2's spot.
100 ohm.png
Then (with the RF unit removed), I connected the middle pin of the RF mod connection to Pin 9 on my multi-out. Connected multi-out GND (Pin 5) to a ground plane, wired up Audio to pin 11 and 12.

----------------------------------------------------------------------

For testing purposes, I used 240p suite on my Everdrive with a full screen color set to 00 (gray).

OK, so I took some o-scope and PVM screenshots of data. First, my unmodded AV Famicom:
AV Fami Screen.png
And the active part of the video, zoomed in to look for noise:
AV Fami Noise.png
As you can see, the image looks very clean, and there is no noise on the video signal.

User avatar
Ultron
Posts: 21
Joined: Tue Feb 26, 2019 6:36 pm

Re: Famicom AV Mod Situation in 2019

Post by Ultron » Fri Apr 19, 2019 11:54 am

Now, the NES-101, with the same setup as the AV Fami (First shot screen is blurry, but you can see the jailbars):
NES-101 Screen.png

And the active video part:
NES-101 Noise.png
NES-101 Noise 2.png


So, you can see, the jailbars have an approx. 667kHz period, or approx. 8 pixels. The jailbar itself is approx. 2.63MHz period, or approx. 2 pixels in width.

EDIT: I should note, the measurements were taken on the left side of FC2 (negative side of the 220uF cap).

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Fri Apr 19, 2019 4:50 pm

So why does modded NES-101 have jailbars and AV famicom perfect (as usual)? Does the AV famicom have a different PPU revision? I noticed when I had a wire attached to PPU pin 21, the other end not ever connected to anything, as I moved the wire around against the board, different jailbars would appear at different locations of this wire... But wrapping everything in foil proved not to keep the aliens out anyway. hmm Could this be a clock stability thing? Maybe each 8th clock pulse is a little wider or something? I never looked at that.

What part of the video signal are you looking at? Could you zoom out to a whole scanline or frame and circle the area you were looking at? I am a little confused and not very familiar with composite video. It looks like you have a great way to measure the jailbars, I would like to repeat that with my setup. Everything I did was qualitative by eye looking at an LCD monitor -- I would love to have a reliable way to actually measure it.

lidnariq
Posts: 8786
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Famicom AV Mod Situation in 2019

Post by lidnariq » Fri Apr 19, 2019 5:17 pm

Ben Boldt wrote:So why does modded NES-101 have jailbars and AV famicom perfect (as usual)? Does the AV famicom have a different PPU revision?
No, the 2C02 G-to-H transition happened in both markets about the same time, and IIRC happened before the NES-101.
I noticed when I had a wire attached to PPU pin 21, the other end not ever connected to anything, as I moved the wire around against the board, different jailbars would appear at different locations of this wire... But wrapping everything in foil proved not to keep the aliens out anyway.
I think PCB layout is the biggest factor here... which may be why there's so much superstition. You're using basically the same amplifier as the original, but it's in a different place.
Could this be a clock stability thing? Maybe each 8th clock pulse is a little wider or something?
That wouldn't cause this particular kind of artifact...
I would love to have a reliable way to actually measure it.
Use something with white or grey instead of color - the color carrier will make it hard to see the comparatively smaller magnitude jailbars.

(this was my 8255th post, a GPIO expander used as the keyboard interface in the original IBM PC)

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Fri Apr 19, 2019 9:26 pm

lidnariq wrote:I think PCB layout is the biggest factor here... which may be why there's so much superstition. You're using basically the same amplifier as the original, but it's in a different place.
I guess I am surprised that bad layout isn't made better with these copper foil attempts. I have grounded copper foil completely between my amplifier and the PCB. Interesting to think where this could be getting in and where from.
lidnariq wrote:That wouldn't cause this particular kind of artifact...
Okay
lidnariq wrote:Use something with white or grey instead of color - the color carrier will make it hard to see the comparatively smaller magnitude jailbars.
That was my plan all along, I had lost track of that! I have the 240p Test Suite cart, I just never ran a white screen and looked at it with the scope. I will do that next chance I get.

User avatar
Ultron
Posts: 21
Joined: Tue Feb 26, 2019 6:36 pm

Re: Famicom AV Mod Situation in 2019

Post by Ultron » Sat Apr 20, 2019 5:31 am

Ben Boldt wrote:So why does modded NES-101 have jailbars and AV famicom perfect (as usual)? Does the AV famicom have a different PPU revision?
Because the NES-101 has the worst PCB layout known to man. A first grader could have done a better job of laying out the traces. My NES-101 has a -G, the AV Fami is -H. Though I don't think that matters (I can swap the PPUs and I'm sure it won't make a difference).

The decoupling scheme is horrible. There is no decoupling cap anywhere near the power pin of the PPU. I added a 0.1uF ceramic, but this did not make a difference. It may need some more decoupling.

hmm Could this be a clock stability thing? Maybe each 8th clock pulse is a little wider or something? I never looked at that.
A lot happens on the 8th pixel of every line. Tiles are 8x8 pixels, so the PPU is doing a lot of work every 8 pixels. Because of this, I believe the noise is coming from a current surge, not necessarily from a noisy pin or trace. I also need to check other pins nearby. I can see some noise on the PPU power pin (Pin 40) of this frequency. Though not as large as on the video signal, the noise does pop up on the 5V line; the noise is MUCH less in amplitude on the AV Fami.

What part of the video signal are you looking at? Could you zoom out to a whole scanline or frame and circle the area you were looking at? I am a little confused and not very familiar with composite video. It looks like you have a great way to measure the jailbars, I would like to repeat that with my setup.
I am looking at the active video part. Like lindariq said, use a white or gray screen in 240p suite. The active video part is located after the color burst.
Composite Video.png


Even with the crap layout of the video path on the PCB, I don't think it's picking up interference from it. When the Everdrive starts up, it flashes a blue and green screen, before showing the menu. The blue and green screens DO NOT have jailbars on them. The menu does, though.

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Tue Apr 23, 2019 3:48 pm

I have a very good view of the jailbars now with white screen. I believe that the jailbars are apparent each time there isn't a ripple with my setup.
tek00014.png
Quite fascinating, when I apply a 1k pull-down on the ALE signal, the ripple reduces:
tek00015.png
I can't visually tell a difference on the white screen though. If I apply a 1k pull-up, it has no effect. If I put a 0.1uF ceramic cap in series with the pull-down, it creates the same reduction in ripple. Comparing 74LS373 to 74HCT373 with this setup, I see no measurable difference. While running, if I completely yank out the 74x373, it also has no effect on the ripple. (Amazing to me that it keeps running that way...) Since ALE is only connected to the '373, these clues seem to point to noise entering through the ALE pin, or being generated by the ALE pin. I tried the 1k and cap, pull-up and -down on all other pins and found no similar effect elsewhere.

Notice the correlation between the PPU AD pins and the ripple. The ripple does not occur at the points where the PPU AD pins don't change. Since the only other thing these are connected to is the PPU's SRAM, it might be interesting to swap SRAM chips and see what happens.

lidnariq
Posts: 8786
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Famicom AV Mod Situation in 2019

Post by lidnariq » Tue Apr 23, 2019 4:19 pm

Ben Boldt wrote:Since the only other thing these are connected to is the PPU's SRAM, it might be interesting to swap SRAM chips and see what happens.
May as well just try pulling it altogether. Won't be any more wrong than removing the '373.

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Tue Apr 23, 2019 5:12 pm

lidnariq wrote:
Ben Boldt wrote:Since the only other thing these are connected to is the PPU's SRAM, it might be interesting to swap SRAM chips and see what happens.
May as well just try pulling it altogether. Won't be any more wrong than removing the '373.
Ooo neat idea, I like it. I will dig around for a narrow DIP-24 socket, I am sure I have one somewhere.

Edit:
I must have used my last damn dip-24 socket for the RAM in that 240p test suite cart... There aren't any left where I usually store them. I will order some and keep looking. 50% chance I will find 100 of them right after ordering more. ;)

Edit 2:
I ordered various sizes last night, and sure enough, I found tons of some of the ones I ordered. But no narrow dip-24s... I have plenty of wide dip-24s and narrow DIP-28s, but not too excited to go that route. I will keep looking.

Edit 3:
Turns out that RAM chip I put into the 240p cart is actually a dip-28, so it looks like I never had any narrow dip24s to begin with.

User avatar
Ben Boldt
Posts: 419
Joined: Tue Mar 22, 2016 8:27 pm
Location: Minnesota, USA

Re: Famicom AV Mod Situation in 2019

Post by Ben Boldt » Fri May 10, 2019 2:10 pm

I put this off for so long that I actually got the correct sockets in the mail. I pulled the RAM chip with it running, and while it does glitch everything up, it seems to still have the jailbars amidst the garbage on the screen. I also ordered a CY7C128A-25PC SRAM chip (25 nsec). I will be able to take measurements comparing the original RAM to this new one. And as a matter of fact, both this famicom and my broken Hi-Def NES have socketed PPUs, so I can swap the PPUs too and see if anything measures differently.

Edit:
I just threw the other PPU in there for the heck of it and interestingly, it has the sprite glitch now with this PPU. So both of my front loader PPUs have this glitch.

Post Reply