Undesired Jail bars on French NES and checkerboard pattern

Discuss technical or other issues relating to programming the Nintendo Entertainment System, Famicom, or compatible systems.

Moderator: Moderators

Post Reply
alfredocalza
Posts: 29
Joined: Sat Jun 22, 2019 1:06 pm

Undesired Jail bars on French NES and checkerboard pattern

Post by alfredocalza » Thu Nov 14, 2019 3:30 pm

Hi!
I have two French NES (RGB versions) which I play in my HD TV through an OSSC. One of them has a NES-CPU-11 board and the other one is a smooth top with a NES-CPU-07 board. Both of them compatible with the rgb scart cable. I purchased them both earlier this year through Ebay from a seller in France, and when I received them, the CPU-07 unit was broken, not even powering up.

I replaced all electrolytic caps on both consoles with high quality Panasonic and Rubicon caps (the CPU-11 unit had some blown-up caps in the RF silver box). I also replaced the ceramic capacitors with better quality X7R ceramic caps. Additionally, I replaced the voltage regulator on both consoles, added a 470uf 6.3v cap on the 5v line of the voltage regulator, replaced the AC to DC current converter on both units (was burnt in the CPU-11 unit), and replaced the crappy 72 pin connectors with more reliable ones.

After doing all those mods and fabricating a high quality RGB cable (trying to get the best quality out of the original hardware) I tested both. The CPU-07 played fine, although some very light jail bars were noticeable in only a few games and in specific blueish background colors. However, this was not the case with the CPU-11 unit, which as far as I know is a later unit (shouldn't it have been improved over previous versions?). Some very noticeable jail bars were present in almost every background color, except for very dark ones. Additionally, there was a very noticeable checkerboard pattern between the jail bars.

Would there be something I could do to fix this issue? I also forgot to mention that the previous owner disabled the lockout chip on the CPU-11 unit by cutting out one of the pins, instead of lifting it and soldering it to ground, which I think would have been the correct way of doing it.

I read another post from a user of this forum saying something like having to put an 83k resistor on pin a13 of PPU or something like that, but I could not quite understand the solution (or don't even know if it ever was fixed).

Here are some pictures of both my CPU-07 and CPU-11 units:

I would really appreciate any ideas on how to fix this!
Attachments
CPU-11
CPU-11
CPU-07
CPU-07
CPU-11
CPU-11
CPU-07
CPU-07

lidnariq
Posts: 8768
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Undesired Jail bars on French NES and checkerboard patte

Post by lidnariq » Thu Nov 14, 2019 4:15 pm

Current best practice seems to be to put a large low-ESR capacitor across the 2C02's power supply.

Last time:
https://forums.nesdev.com/viewtopic.php ... 97#p237597

Note that in the Famicom, pin 22 of the 2C02 is tied to +5V; this is not true in the front-loading NESes, so you won't be able to follow Ben Boldt's schematic exactly.

Simply breaking the pin on the CIC is fine. Other modifications (e.g. https://forums.nesdev.com/viewtopic.php ... 46#p209346 ) are more reversible, but it honestly doesn't matter.

alfredocalza
Posts: 29
Joined: Sat Jun 22, 2019 1:06 pm

Re: Undesired Jail bars on French NES and checkerboard patte

Post by alfredocalza » Fri Nov 15, 2019 7:57 am

Hey, thank you for your reply! I have started modifying and fixing consoles as a hobby only about a year ago. Moreover, a year ago I did not even know what a solder nor a soldering station looked like! (I did not study electronics or anything like that, but have read basic stuff online)

That being said, I am not very acquainted with the technical language nor the different components in a board (other than the very basic stuff). I read through the link you have provided me and could not find the solution to the jail bars! It looked to me like they were talking about something else (please excuse my ignorance).

So, according to what I understood from your reply, and following this diagram:
https://wiki.nesdev.com/w/index.php/PPU ... escription

All I need to do is to solder the positive leg of a high capacitance and low ESR electrolytic capacitor to pin #40 of the PPU and solder the negative leg to ground?

Would a 6.3V 470uf electrolytic capacitor be enough?

lidnariq
Posts: 8768
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Undesired Jail bars on French NES and checkerboard patte

Post by lidnariq » Fri Nov 15, 2019 11:16 am

Not just ground, but ground right next to the PPU. Jailbars seem to result from the NES's mediocre power supply traces.

470µf electrolytic almost certainly has too high an ESR. Ben Boldt in the linked thread found the sweet spot was a tantalum capacitor, in the 5-45µF range.

If you really want to use parts you have on hand, make a bundle of a bunch of electrolytic and ceramic capacitors. The datasheet for one commercial IC I read says something like 10nF ceramic + 100nF ceramic + 10µF electolytic

alfredocalza
Posts: 29
Joined: Sat Jun 22, 2019 1:06 pm

Re: Undesired Jail bars on French NES and checkerboard patte

Post by alfredocalza » Sat Nov 16, 2019 10:21 am

Hey, thanks a lot man! I will buy some tantalum caps at a local electronics shop and test different cap ranges to see which one works best. I will probably start with 10uf and go from there.

Just for curiosity, is there any reason why tantalum capacitors are the best option for this? Do they have lower ESR than that of ceramic capacitors?

alfredocalza
Posts: 29
Joined: Sat Jun 22, 2019 1:06 pm

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by alfredocalza » Wed Nov 27, 2019 1:04 pm

Just tried 15uf, 33uf and 47uf tantalum capacitors by connecting them to pins 40 (+5v) and 20 (gnd) of the PPU and tested the console three times with the same game (Super Mario Bros 3) and was not able to get the jail bars removed. What is more, I was not able to even appreciate any improvement by using any of the caps; just as if I hadn't done anything! Is there anything else I could try to get these jail bars removed? Remember that this is a French NES with the rgb adapter. Maybe the jail bars have something to do with the RGB adapter and not the PPU?

lidnariq
Posts: 8768
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by lidnariq » Wed Nov 27, 2019 2:54 pm

alfredocalza wrote:
Wed Nov 27, 2019 1:04 pm
Just tried 15uf, 33uf and 47uf tantalum capacitors [...] and was not able to get the jail bars removed. [...] as if I hadn't done anything! Is there anything else I could try to get these jail bars removed?
It's remotely conceivable that there could be some capacitive coupling between the video signal from the PPU and the PAL demodulator inside the silver box. I don't really have any ideas, unfortunately.
You could try following the rest of the previously-linked thread and seeing if any of the other suggestions help.
Remember that this is a French NES with the rgb adapter. Maybe the jail bars have something to do with the RGB adapter and not the PPU?
For jailbars, I doubt it. Jailbars are specifically coupling from various other PPU signals – especially PPU /RD and PPU A13 – and I don't think those are anywhere near the converter.

For the checkerboard... I dunno, maybe that's the 4.43MHz PAL carrier peeking through? It's very roughly the same width as a NES pixel – maybe a little wider – so that'd be credible. If so, maybe you need to replace some capacitors in the metal box.

Krzysiobal has a reverse-engineered schematic for the French PAL demodulator in this thread, if that's any help. I'd see if replacing the electrolytics there fixed anything.

alfredocalza
Posts: 29
Joined: Sat Jun 22, 2019 1:06 pm

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by alfredocalza » Thu Nov 28, 2019 1:30 pm

hey thanks for all your help! How can I find out if there is any capacitive coupling? and also, which component is the PAL demodulator? I also do not know which is the composite video signal line from the PPU (please excuse my ignorance!)

by the "4.43MHz PAL carrier peeking through" you mean that there could be a problem with the crystal oscillator inside the metal box? I could buy a new one and try replacing it; they are easy to come by here in Europe.

I also replaced all caps inside that metal box with brand new ones. Although one of the replacements was a small Chinese cap (which I had bought brand new) because I did not have a Panasonic nor Rubicon replacement cap for that specific value.

I also read another post from a user with the same issue and same PPU-11 console, in which you suggested connecting "PPU A13 (or maybe NOT(PPU A13) out of the inverter) to the video pin via a huge resistor" and that seemed to fix the problem but only for some of the cartridges, but I didn't really understand the NOT(PPU A13) part. Is this something you would recommend me trying, or is it not going to fix the issue?

I will continue reading that other post that you referenced and keep trying other things to see if I can find a solution.

Thanks again!

lidnariq
Posts: 8768
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by lidnariq » Thu Nov 28, 2019 2:21 pm

alfredocalza wrote:
Thu Nov 28, 2019 1:30 pm
How can I find out if there is any capacitive coupling?
I've forgotten what the french NES mainboard looks like. Is it a standard NES-CPU-nn like the ordinary PAL and US mainboards?

If so, those boards are – accidentally, given the AV Famicom and NES-101 – usually ok for this, without any relevant signals lying physically adjacent.

Two adjacent conductive things form a capacitor; for example two parallel traces will loosely couple. (Say the trace is 40µm thick, and is 0.3mm away from another trace, and the traces parallel each other for 10cm. Then there's a parasitic capacitance of about ε₀ · 40 µm · 10cm ÷ 0.3mm = 0.1pF. Alternatively, two parallel traces on alternate sides of the NES's 2mm thick PCB would have ε₀ · 0.3mm · 10cm ÷ 2mm = also 0.1pF)

This is fine when you're dealing with digital signals, but with analog signals, especially low in magnitude (e.g. mV) or high impedance (e.g. kΩ), a small amount of capacitive coupling can swamp the output.

... All that said, the more I've looked into this, I don't think it's capacitive coupling in the NES. The video signal is amplified close enough to the PPU (the BJT is right there) that there's just not enough physical space for it to be likely.
and also, which component is the PAL demodulator?
The Sony/ASCII V7021 IC inside the metal box.
I also do not know which is the composite video signal line from the PPU
Pin 21. (nesdevwiki:PPU pin out)
(please excuse my ignorance!)
It's fine! One only learns by asking.
by the "4.43MHz PAL carrier peeking through" you mean that there could be a problem with the crystal oscillator inside the metal box? I could buy a new one and try replacing it; they are easy to come by here in Europe.
Maybe? If it's inexpensive enough to try, go for it. Alternatively, try adjusting the SMT capacitors on the opposite side of the PCB there. Maybe try adjusting the potentiometer in there - it goes to what the datasheet says is "ACC TC".
I also replaced all caps inside that metal box with brand new ones. Although one of the replacements was a small Chinese cap (which I had bought brand new) because I did not have a Panasonic nor Rubicon replacement cap for that specific value.
So much for that idea.
but I didn't really understand the NOT(PPU A13) part.
The PPU generates [PPU A13]; the adjacent 74HCU04 then inverts it ("NOT").
Is this something you would recommend me trying, or is it not going to fix the issue?
It's a dirty hack, and it's fragile and inconsistent. So it's definitely not a solution but it will change when things look bad.

Looking at your screenshots yet again, I see a clear pattern that goes _-_^_-__, which clearly somehow contains PPU A13 (period of 8 pixels) and also PPU /RD (period of 2 pixels) but ... I'm really at a loss as to what would cause this specific noise pattern ... other than power supply lousiness.

alfredocalza
Posts: 29
Joined: Sat Jun 22, 2019 1:06 pm

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by alfredocalza » Sat Nov 30, 2019 12:59 pm

lidnariq wrote:
Thu Nov 28, 2019 2:21 pm
alfredocalza wrote:
Thu Nov 28, 2019 1:30 pm
How can I find out if there is any capacitive coupling?
I've forgotten what the french NES mainboard looks like. Is it a standard NES-CPU-nn like the ordinary PAL and US mainboards?

If so, those boards are – accidentally, given the AV Famicom and NES-101 – usually ok for this, without any relevant signals lying physically adjacent.

Two adjacent conductive things form a capacitor; for example two parallel traces will loosely couple. (Say the trace is 40µm thick, and is 0.3mm away from another trace, and the traces parallel each other for 10cm. Then there's a parasitic capacitance of about ε₀ · 40 µm · 10cm ÷ 0.3mm = 0.1pF. Alternatively, two parallel traces on alternate sides of the NES's 2mm thick PCB would have ε₀ · 0.3mm · 10cm ÷ 2mm = also 0.1pF)

This is fine when you're dealing with digital signals, but with analog signals, especially low in magnitude (e.g. mV) or high impedance (e.g. kΩ), a small amount of capacitive coupling can swamp the output.

... All that said, the more I've looked into this, I don't think it's capacitive coupling in the NES. The video signal is amplified close enough to the PPU (the BJT is right there) that there's just not enough physical space for it to be likely.
and also, which component is the PAL demodulator?
The Sony/ASCII V7021 IC inside the metal box.
I also do not know which is the composite video signal line from the PPU
Pin 21. (nesdevwiki:PPU pin out)
(please excuse my ignorance!)
It's fine! One only learns by asking.
by the "4.43MHz PAL carrier peeking through" you mean that there could be a problem with the crystal oscillator inside the metal box? I could buy a new one and try replacing it; they are easy to come by here in Europe.
Maybe? If it's inexpensive enough to try, go for it. Alternatively, try adjusting the SMT capacitors on the opposite side of the PCB there. Maybe try adjusting the potentiometer in there - it goes to what the datasheet says is "ACC TC".
I also replaced all caps inside that metal box with brand new ones. Although one of the replacements was a small Chinese cap (which I had bought brand new) because I did not have a Panasonic nor Rubicon replacement cap for that specific value.
So much for that idea.
but I didn't really understand the NOT(PPU A13) part.
The PPU generates [PPU A13]; the adjacent 74HCU04 then inverts it ("NOT").
Is this something you would recommend me trying, or is it not going to fix the issue?
It's a dirty hack, and it's fragile and inconsistent. So it's definitely not a solution but it will change when things look bad.

Looking at your screenshots yet again, I see a clear pattern that goes _-_^_-__, which clearly somehow contains PPU A13 (period of 8 pixels) and also PPU /RD (period of 2 pixels) but ... I'm really at a loss as to what would cause this specific noise pattern ... other than power supply lousiness.
Hi, thank you again for your explanations! I read other posts and tried a few things involving ceramic and electrolitic capacitors close to the 5v pin of the ppu, but saw no improvement whatsoever. I think that capacitive coupling issue that you mentioned might be worth checking; I read at another forum that this was an issue with cpu10 and cpu11 consoles. Mine is a cpu11 one. What if i lift pin 21 of the PPU, which if I am not mistaken it is composite video out, and connect that to a transistor and then straight into the video out, bypassing any noise that it might pick up from other lines on the board? I can even use a shielded cable that I have. For this, I would need to know what type (specs) of transistor do I need in order to do do the composite video amp? Also, after I amplify that signal, where exactly should I connect the cable?

lidnariq
Posts: 8768
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by lidnariq » Sat Nov 30, 2019 1:48 pm

alfredocalza wrote:
Sat Nov 30, 2019 12:59 pm
What if i lift pin 21 of the PPU, which if I am not mistaken it is composite video out, and connect that to a transistor and then straight into the video out, bypassing any noise that it might pick up from other lines on the board?
I mean, that's more-or-less what it's already doing. I haven't been able to find a good picture of the NES-CPU-11 mainboard to compare to the NES-CPU-07 mainboards I have around here, but there aren't that many signals around...
I can even use a shielded cable that I have. For this, I would need to know what type (specs) of transistor do I need in order to do do the composite video amp?
Do use coax. It's a PNP transistor; almost anything will do. (Since you're bypassing the existing amplifier, you could scavenge the parts used for the amplifier that's on the mainboard, which is a 2SA937 and a 510 ohm resistor).

Krzysiobal generated a reverse-engineered schematic from his (imported) french NES-CPU-11.
Also, after I amplify that signal, where exactly should I connect the cable?
The same place where video currently goes into the metal box, unless you want to convert the whole thing to composite output.

You should be able to remove the ferrite bead and re-inject the video there.

alfredocalza
Posts: 29
Joined: Sat Jun 22, 2019 1:06 pm

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by alfredocalza » Sun Dec 01, 2019 5:00 pm

I followed the path of the composite line out of the PPU to see which components are being used and how the signal makes its way to the video out metal box, and this is what I found:

1) The signal goes out of pin 21 of the ppu into a pnp transistor marked as "ECB" on the main board. What caught my attention here is that the middle pin of the transistor is connected to ground instead of it being connected to the 5v line. I have read very little about transistors, but it seems counter intuitive to amplify a signal with a transistor connected to ground instead of +5v. Is this correct?
2) The "amplified" composite signal comes out of the transistor and is connected through a resistor to the +5v line. The resistor is marked as 510 ohms on the mother board, but when I tested it, it was a 140 ohm resistor (attached photo). Could this be causing the jail bar issue or the checkerboard pattern? Should add a 510 ohm resistor instead? Why is the composite line being connected to the +5v line through a resistor anyway? This doesn't make sense to me.
3) The amplified composite line is then connected to the first pin of the silver metal box through a ferrite bead and also a 330pf capacitor (blue ceramic cap in the picture).
4) The first pin of the silver metal box is then connected to some of the pins belonging to two metallic little cubes inside the silver metal box through a 550 ohm resistor. What are these two metallic little cubes and what do they do? Could the problem be hidden here?
5) I also noted that the composite line at this level is also connected to ground through a 477 ohm resistor. This is weird, because that same line is also connected to +5v through a resistor.
6) After this, I lost track of the line because I do not know what the two little metallic cubes are.

Does this seem to be correct? Is there something out of whack in this setup?

Thanks!
Attachments
IMG_20191201_202040.jpg
IMG_20191201_201956.jpg
IMG_20191201_201918.jpg
IMG_20191201_180537.jpg

lidnariq
Posts: 8768
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Undesired Jail bars on French NES and checkerboard pattern

Post by lidnariq » Sun Dec 01, 2019 6:03 pm

alfredocalza wrote:
Sun Dec 01, 2019 5:00 pm
1) The signal goes out of pin 21 of the ppu into a pnp transistor marked as "ECB" on the main board.
"ECB" there is marking the orientation of the transistor, the emitter, collector, and base.
What caught my attention here is that the middle pin of the transistor is connected to ground instead of it being connected to the 5v line. I have read very little about transistors, but it seems counter intuitive to amplify a signal with a transistor connected to ground instead of +5v. Is this correct?
The transistor there is a PNP one. It's serving as a voltage buffer (current amplifier), emitting a voltage that is about 0.7V above what comes out of the PPU.
2) The "amplified" composite signal comes out of the transistor and is connected through a resistor to the +5v line. The resistor is marked as 510 ohms on the mother board, but when I tested it, it was a 140 ohm resistor (attached photo).
The bands on the resistor show that it's supposed to be a 150 ohm resistor (brown green brown gold). 140 ohms is just a little lower than you "should" get from a 5% tolerance resistor, but you're measuring it in-circuit, so I would assume it's fine.
Could this be causing the jail bar issue or the checkerboard pattern? Should add a 510 ohm resistor instead? Why is the composite line being connected to the +5v line through a resistor anyway? This doesn't make sense to me.
Here's some articles on wikipedia:
english wikipedia:Common collector
especially note
wikimedia commons:PNP emitter follower
4) The first pin of the silver metal box is then connected to some of the pins belonging to two metallic little cubes inside the silver metal box through a 550 ohm resistor. What are these two metallic little cubes and what do they do? Could the problem be hidden here?
Those are ... RF transformers? I think? I don't think they're relevant to your symptoms.
5) I also noted that the composite line at this level is also connected to ground through a 477 ohm resistor. This is weird, because that same line is also connected to +5v through a resistor.
The two resistors together serve to move the average voltage around. The V7021 has specific requirements on how the video enters it: namely, it must
1- be roughly 2.7V DC
2- be 0.36Vpp

If those aren't complied with, video may be off, but I wouldn't anticipate jailbars as a plausible symptom. It'll be hard to check whether the latter is right without an oscilloscope, but you could measure the voltage on V7021 pin 27 ("VIDEO IN") and see if it's roughly 2.7V.
6) After this, I lost track of the line because I do not know what the two little metallic cubes are.
You can use krzysiobal's schematic here: https://forums.nesdev.com/viewtopic.php?f=9&t=18861
Does this seem to be correct? Is there something out of whack in this setup?
Nothing seems off to me.

Post Reply