PAL 1CHIP + SuperCIC - no color via composite in PAL mode

[Freedy]

Sorry for the late answer!
I captured a new sample, for some reason I could not choose wav files, this is why I compressed it into a zip file.
In the meantime some random 27MHz crystal arrived from Aliexpress I ordered a few weeks back, discoloration on top of the screen is present with it, too.


Later I will capture some of the composite video signal with the 1chip console, as well.

[lidnariq]

In octave, I:

Code: Select all

SOUND = audioread ("SFC_PALmode.wav");
LENGTH = size(SOUND)(1)
semilogy([1:LENGTH] * 48000/LENGTH,abs(fft(SOUND )));

and then zoomed in to find the relevant maximum is at 15601.8 Hz.

The frequency should be

4433618.75 x 6 ÷ 5 ÷ 341 = 15602.18Hz

This is a difference of 24ppm, which is small enough that we can't be certain whether it's your SNES or your computer; in any case it's less than 300ppm so it's not the cause of the weird artifacts.

[Freedy]

Thank you for the quick answer!
Did the signal recording with the 1chip system, too.
~15600,5Hz and unless I messed up something(1-(15600.5/15602.18))*10^6, there is a ~107ppm difference, thats still significantly lower than 300ppm.

I do not really have any good idea how to proceed.

For the 1chip:
-I'll put the n64 PAL crystal back, and I'll capture some sample with that, as well
-I think next time I'll visit my parents, I will take the SNES with me with the current crystal and check it on other TVs, as well. Maybe the CRT I have with me has some failing caps which are causing significantly lower tolerance range, but I am just guessing, I do not know how likely that type of failure mode is

For the SFC
I am thinking about replacing the Altera CPLD with a Microchip one from a reputable seller on the multiregion board. I do not think its very likely, but maybe I got some weird part from aliexpress thats causing this weird discoloration(for the color carrier, the CPLD does the division based on the CDCE913's output)

EDIT:
Checked my CRT, there is only 1 ceramic cap in the signal path between composite video input and video processing IC.
Checked C18 electrolytic cap in the composite video signal path on the 1chip SNES, it tested ok

[Freedy]

I've put back the PAL n64 crystal, I measured ~15604.5 Hz.
Does not really make sense to me why this is happening.(only having color in PAL mode with n64 crystal, not with crystal from other 1chip snes or crystals I ordered)
Unfortunately I do not have any other PAL SNES to compare the colors with, I am seeing some color artifacts but I can not judge if that is just simply how PAL composite looks like, or something is still not ok with the color signal.

[lidnariq]

Let's assume that this 140ppm fast is "correct" and that your measurement device is slow by this amount, so what appears to be 20ppm off isn't. Furthermore let's assume that your TV is more sensitive than I thought it should be and rejects colorburst well before it gets as much detuning as 300ppm.

In any case, all loose crystals (the kind with only two wires) have a specific expected ballast capacitance, usually around 10-25pF. You should be able to adjust (decrease) those to speed up the OEM crystal until it's running at the right frequency.

[Freedy]

In the meantime I tried RF output as well with the 1chip, and there is no color that way either.

I ordered a 2 more PAL crystals from 2 different manufacturers from a mouser partner, but currently my best guess is that the S-CPU or the S-RGB went out of spec. It would be nice to look directly at the color carrier frequency coming out of the S-CPU, or in the case of the 3chip SFC, the color carrier output of the multiregion board.

I've been looking at oscilloscopes, but basically I would buy it only for these two consoles and I can not really justify the price. I saw that cheap frequency meters exist, but I do not know how good they are for this purpose. It would be also nice if I could look at the composite video output spectrum somewhere around the 4.43361875MHz frequency.

I used a soundcraft notepad 12fx for recording the sample from the composite video output, I do not know how accurate it is.
I guess I have no other choice left other than to buy an oscilloscope maybe I'll buy one next year. Until then I'll try out some other TVs and crystals.

[lidnariq]

Most of the time, oscilloscopes don't maintain the kind of frequency stability needed to really tune this in. They just use a quartz crystal, with the commensurate 30ppm precision that you get if you followed the manufacturer's specification for what ballast capacitors are present.

The specific software you used won't matter; this is just a question of the reference clock in your computer (before it ultimately goes to the soundcard). (It's probably a 24MHz crystal somewhere. That seems to be a standard choice.)

To measure things to better precision than 30ppm, you'd need something like a Oven Controlled Xtal Oscillator, or a Global Positioning System Disciplined Oscillator.

If I remember correctly, you said that 60Hz video with NTSC3.6 color worked correctly. If you record that with the same apparatus, what horizontal sync frequency does it show? In NTSC mode the SNES hsync rate should be 15745.8Hz

[Freedy]

I received the Atmel CPLDs from mouser and I already replaced the Aliexpress Altera CPLD with one of them. Unfortunately (although not so surprisingly) the issue still persist.
I will do the additional measurements in NTSC mode

[Freedy]

Redid the measurements with the 3chip + CPLD based multiregion board in PAL and NTSC mode, as well. I guess they seem fine.

I am thinking about replacing the SN74LVC2G17DBVR IC on the multiregion board, it is in the signal path between the CPLD's output for the color carrier frequency and the output of the multiregion board.

An other idea is, to replace the BA6592F IC with an S-ENC (Are these two ICs the same? )

[Freedy]

I did the measurements with the 1CHIP, as well. i do not know which PAL crystal is in it, but there is no color with it in PAL mode.