AES graphic glitches (vertical line syndrome) NOW W/ PICS

[MKL]

There is some confusion in the above discussion (cfr. post #10 and #19) because some boards have two trimmer capacitors and some have only one.

This is the state of affairs on the various revisions:

NEO-AES
one trimmer cap (0-50pF) for the video encoder crystal

NEO-AES with daughterboard
one trimmer cap (0-50pF) for the video encoder crystal and one (value not specified) for the CPU crystal (on the daughterboard).

NEO-AES3-3
one trimmer cap (value not specified but most likely 0-50pF) for the video encoder crystal and one (value not specified) for the CPU crystal (near the memory card connector).

NEO-AES3-4
one trimmer cap (value not specified but most likely 0-50pF) for the video encoder crystal and one (value not specified) for the CPU crystal (near the memory card connector).

NEO-AES3-5
one trimmer cap (value not specified but most likely 0-50pF) for the video encoder crystal and a fixed value cap (6pF) for the CPU crystal (near the memory card connector).

NEO-AES3-6
one trimmer cap (value not specified but most likely 0-50pF) for the video encoder crystal and a fixed value cap (6pF) for the CPU crystal (near the memory card connector).

 

[LARGE KRO]

Damn, I've been lucky to find this thread.

I was stocking an AES for a long time playing mostly on MVS since a few years. I also made the critical error of reselling my old japanese system serial n°34... system a while back. Now I realise apart of the quality video output difference those old and early systems may also have a different calibration on the video.
My actual system is also a japanese model but has a serial arround 147... and last week I found the same graphical glitch you describe playing on Robo Army and Eight Man.

Taking a look at this thread I opened it and saw it' a NEO-AES3-5 motherboard with only one yellow trimmer. Adjusting it made the glitch disappear and contrast just a little clearer using the RGB video cable, but not that evident to notice.

I thank you very much for investigating on it guys, you rock

 

[FuriousT]

So,
What do you guys make of this?

I bought this Ghost Pilots (EUR) brand new. It shows the following on both a brand new US & PAL (converted to 60Hz) AES

I then got a different US Dog Tag version (the cart looks the same) and it doesn't show the glitches at all.

I don't see these glitches in any other games, so I guess it's a faulty cart?

 

[Xian Xi]

Clean the contacts. If it still exists you may need to open the cart.

 

[FuriousT]

Cheers.

The contacts have been well swabbed with surgical spirit, not really any dirt. If I open the cart, what am I looking for?

 

[FuriousT]

I've opened the cart, here are the boards.

Like I say, I bought the cart as new, and I can't see anything obvious.

*shrug* might just be a dead cart

 

[channelmaniac]

Clean those contacts with a big pink pencil eraser...

 

[FuriousT]

Cheers boys.

Those contacts are so clean now i can see my ugly mug in them...game still knackered though.

Ah well, to the loft with it.

 

[Xian Xi]

How would you fix this problem on an MVS?

 

[toofastforyahuh]

Wondered this myself. I see no trim pots on my (old decrepit bad BRAM) MVS. Just a 24 MHz crystal and an RTC crystal.
If the crystal really is the culprit, it would be quick to replace.

More interesting is what does this drift over time look like? Could you characterize it as frequency drift? Duty cycle drift? Vpp sagging? Or, worse still, jitter?
If someone has a bad system, it would be nice to see oscilloscope screen shots of the "bad" condition and the pot-adjusted "good" condition.
If I was looking at a high-speed clock, I would normally recommend 2 types of screen shots --- a regular instantaneous view to see the frequency/Vpp/duty cycle, and a scope persistence view to get an idea of jitter.
It would be nice to get these screen shots both at the crystal output itself and also after this trim pot circuit.
That way we could see both bad and good cases and see exactly what the trim pot is doing.

 

[Xian Xi]

Here's a good quote from MKL:

The pot is actually a variable capacitor and is part of the stock encoder circuit (it acts upon the crystal frequency). The neobitz is supposed to replace the stock encoder so that tweaking the trimmer cap shouldn't have any effect on the neobitz outputs. Unless the RGB inputs for the neobitz are the outputs of the stock encoder. If that's the case I suggest redoing the mod and getting rid of the stock encoder as well as the daughterboard, i.e. turning your system into a 1st gen NEO AES board model.

It would also be possible to have component from the neobitz (with trimmer cap having no effect on it) and composite/s-video from the stock encoder.

From what I remember, Tony said that vertical dots on an MVS usually happens on the older hardwares that have a PRO-BO IC. It could be an addressing issue with the internal fast ram.

On my board I checked the PRO-BO and it had a few lifted pins, reflowed the chip but it's still the same. So I don't know if the internal ram is bad or if it's something else.

 

[toofastforyahuh]

If B0 didn't have enough setup time margin, and over time if the clock is drifting too fast (or some noise component makes it feel the same) all kinds of fun can happen.

Today's chip designers have the luxury of expensive timing analysis tools so they know their margins (across the entire range of the process technology) before tapeout. ADK/SNK didn't have those tools in the late '80s.

If raising B0's voltage and/or lowering its temperature to 0C (run the MVS in a freezer) fixes the dots, it's a classic symptom of setup time failures. (Even today chip designers resort to this kind of testing in the lab when things seem wrong. Methodical experiments of "plot a 2D failure graph")

Meanwhile if the clock is drifting slower over time, there just aren't enough cycles to get the job done. Maybe those last few pixels are stuck in a FIFO somewhere?

If you have $$$ burning a hole in your pocket, the other fun thing would be to swap the crystal with a function generator and see what frequency ranges seem stable and which produce the dots. And then try the same with a B1 if you really wanted to be thorough. :-)

 

[carlos66]

hi there I was wondering is there anyone who could shed some light on a AES im having real problems with really strong vertical lines
any help would be great
trying to save aes japan
probly put this in the wrong place

 

[carlos66]

could anyone help me with a broken aes I have,sound works fine uni bios doesn't show any problems graphics are coloured squares all wording seems ok
any help would be great

 

[GadgetUK]

Those white lines I think are probably a dirty cart pin, or maybe a trace between cart connector and motherboard. Clean your cart socket up first, inspect motherboard for trace damage.

EDIT: If you've got a unibios try enabling the hardware testing so when it boots it will check out the VRAM etc. Just to rule out any kind of video RAM issue.

 

[carlos66]

thanks for the reply gadget.do u mean b.c.d on the controller or is there another option I have missed.did b.c.d and that loads no errors will clean contacts and get back to you.very nice work by the way with youtube stuff

 

[Xian Xi]

It's a fix layer issue. Is this the only game that does this? If all the games do this then you need to clean your cart slots. It's most likely a connection problem to the S1 rom in the cart. If you clean the slots and the game still does it then it's either the game itself or a bad trace from the cart slot to the AES.

 

[carlos66]

will check that out too... much work needs to be done thanks for the reply,will let you know the results
have tried the unibios hardware test(3.2 bios) that doesn't give any problems

 

[carlos66]

got another fix it job aes with these graphics problems will add photos
hope this is fixable

thanks

 

[GadgetUK]

My guess on that one is the VRAM.

 

[Atro]

It can also be a custom.
I had the same issue on a MV2. Swapped the neo257 and it got fixed.
Too bad the other 257 weren't good, so it only solved half the problem.

 

[carlos66]

Which chip is the v ram ....is that something I could get..

 

[Atro]

Which chip is the v ram ....is that something I could get..

You can order Vram on www.buyicnow.com, but it's hard to tell you where it is since we don't know what revision of AES you have.

 

[carlos66]

my aes is 3-5. have got 2 consoles (both 3-5)and both have similar issues.funny thing is the unibios doesn't pick up an error

 

[Xian Xi]

my aes is 3-5. have got 2 consoles (both 3-5)and both have similar issues.funny thing is the unibios doesn't pick up an error

That's because the Unibios only tests certain parameters on the system.