CyberVision64 Problem (corrupted screens)

[StompinSteve]

Hello,

I lost my RTG virginity today. It has not been a pleasant experience... After waiting for the mail a very long time (thank you French postal-workers who lost the package at Marseille Airport for several weeks and did not give a shit), I finally received my Phase5 graphics-card today. It's the CyberVision64 (and not the CyberVision64/3D).
The card has 4MB VideoRAM onboard.
My Monitor is a DELL U2410F
The computer is my 4000T

I use a short DB32 to VGA buffered cable to go from the 4000T native output to the "input" on the CyberVision64. Its output goes to the Monitor.
The standard amiga modes (early boot screen etc. etc.) thus all work fine thanks to the passthrough.

I installed P96 v3.3.2 (current version, bought it from Jens while card was in the mail). Installed P96, chose CyberGraphics 64 (was pre-selected) and a 31kHz vertical scanrate Monitor. All good. Rebooted, started P96Mode and tried out some resolutions. All fine. Saved the default config and started the normal ScreenMode Prefs. Selected CyberVision 1120x832x256 colors and it worked fine.

I then went about finetuning the monitor controls (pixel clock to remove some vertical banding) and was mostly happy. Picture Quality was almost perfect.

I then ran P96Mode again, selected 1120x832x256 colors and clicked on "Edit". There i played around a little. I wish I hadn't...

At one point, I did "something" in "edit mode" (cannot remember what, I wish I did) and the screen jumped to the left and went all borked and corrupt (see screenshots). Maybe something like "increase/decrease HTOTAL/VTOTAL, Pixel Clock, Toogle polarity". I really can't recall what made the screen bork and get all screwed up.
No problem I thought, I pressed ESC and I was back to Workbench. But I must have saved this setting with the corrupt display config because after a reboot, the RTG'd workbench was totally corrupt. Same kind of screwed up graphics as can be seen in the "test mode" screenshots (the last one is the "edit" mode from 800x600 I believe).

Long story short: ALL the screen-modes / resolutions I try, end up corrupt in test-mode. ALL OF THEM. Nothing is normal anymore.

I was able to select a regular PAL screenmode so that I have a workable workbench now. But in the P96mode application, every single resolution and colourdepth I select (the non-supported ones cannot be selected) shows a totally corrupted display. Nothing works anymore. The card has become totally unusable.

I then uninstalled P96 incl. the environment settings. I had installer create a logfile during install and now used it to make sure the cleanup was complete and nothing was left behind. Then I rebooted.

After the reboot, installed P96 again with the exact same settings and rebooted. Result: all the resolutions etc. etc. are still corrupted. Nothing changed.

I also put the monitor back to it's factory settings just in case.
I cannot, for the life of me, get a normal picture when clicking "test" or "edit" in P96mode

I have no idea what is going on. Could the card itself be permanently set to some setting it cannot handle?
Shutting down the 4000T, wait 30 seconds and starting here up does not improve the situation. The passthrough is unaffected. Selected standard native Amiga-Modes works fine.

[pipper]

This looks very much like one of the RAM chips gave up it’s ghost. Probably just a coincidence that it happened while you were tweaking things. I had the same on my CV3D and swapped the faulty chip with a new one. That solved the issue.

[StompinSteve]

Originally Posted by pipper

This looks very much like one of the RAM chips gave up it’s ghost. Probably just a coincidence that it happened while you were tweaking things. I had the same on my CV3D and swapped the faulty chip with a new one. That solved the issue.

Now that you mention it. I started getting artifacts, funny lines and messed up pixels after 15 minutes or so, before the image got totally corrupted (maybe 15 minutes later). That is actually the reason I started playing with the settings. Maybe the pixel clock or some other setting was not correct was my thinking.

Just my luck. 800 Euro's (card + shipping + import-taxes) and the card lasted half an hour before it broke. Crap and Merde!
Is there a way to find out which VRAM Chip is defective?
There are 8 VRAM chips, of which 4 are soldered onto the board (the first 2 MB) and 4 are in sockets (the other 2 MB).

[Docent]

Originally Posted by StompinSteve

Is there a way to find out which VRAM Chip is defective?
There are 8 VRAM chips, of which 4 are soldered onto the board (the first 2 MB) and 4 are in sockets (the other 2 MB).

Remove these 4 chips that are socketed and check if the board still has problems.

[StompinSteve]

I did. I carefully took out the 4 socketed chips, effectively making it the 2MB Version of the card.
As expected, the problem is still there because the resolutions and 8-bit colordepth (or 16-bit when going really low) are supported by the 2MB card also. In other words, the card has not used the upper 2 MB address-regions yet.
This means one of the 4 soldered chips (the first 2MB) is defective.

[StompinSteve]

Look at this chip. There are 8 of these “Texas Instruments 3427XAK F245” chips, 7 are next to each other near the connector and one is a bit further away on the board. One of them has 3 pins broken off (see attachment).

So there are 8 of these F245 chips and there are 8 RAM Chips. Could these F245 have a "controlling function" of the RAM chips. One F245 controls one RAM Chip? 1:1
I find that these are "Octal Bus Transceivers" when I google for "F245 Chip".

"These octal bus transceivers are designed for asynchronous communication between data buses. The devices transmit data from the A bus to the B bus or from the B bus to the A bus, depending upon the logic level at the direction-control (DIR) input. The output enable (OE) input can be used to disable the device so the buses are effectively isolated."

Even a total noob like me understands that this is extremely likely the source of the problem.
The packaging of such chips I found, are also DIP-20 like on my board.
On my broken chip, pins 11, 12 and 13 which are B8, B7 and B6 respectively, are missing.

Regardless, I must send this board to a repair-specialist :-(

The RAM Chips are easy to find. But when I google for "3427XAK F245" or just "3427XAK" i get nothing so far.

[Jope]

This chip is like this in CV64s.

The chip is a 74F245 part in a SOIC package. The other numbers are not relevant for identifying it. But as said, the missing pins are missing in all CV64s.

[StompinSteve]

Dunno what a CV64 is other than the Aircraft carrier USS Constellation CV-64), but i'm sure you mean something else ;-)

Joking aside, when looking at the 8 chips with a microscope (= smartphone camera zoom...), I see that every one of the 8 F245 chips is wired up differently when we talk about their pins 11, 12 and 13.
The 3 missing pins (11, 12 and 13) are connected to the same track (i have continuity on their 3 respective pads). On the other 7 chips, this varies strongly. Some are connected directly to a "golder finger" on the Zorro connector. Others are not.

Because they are different, I guess my theory about each of them doing something with a RAM chip is probably wrong.

So flush that theory, back to the "a RAM chip is broken" theory....

Edit: I have a Revision 2.0.1 Board and on that revision, these 3 pins are not connected at the factory. It can be clearly seen on this high-res image of this version of the board on BBOAH https://bigbookofamigahardware.com/b...63&fileId=4538

Zoom in and look at the 7 identical F245 chips directly above the Zorro connector. Then look closely at the right most F245 chip. You will see it has those 3 pins "empty" as on mine.
I found a picture of a revision 1.whatever and despite it being hard to see, on that revision, those 3 pins are not soldered on. So I guess on the Revision 2, they snipped of those legs on that one IC for some reason.

[grelbfarlk]

I would delete your devs:Picasso96Settings file or wherever it is and start a new one.

Get UMC from Aminet. Run it from a shell like umc 640 480 60 and put those numbers in the corresponding mode for Picasso96Mode and tweak them a little bit for sizing and positioning.

Or upload your Picasso96Settings file and someone with a CV64 could test it themselves.

It's "easy" to add a mode or tweak a mode so that an RTG board freaks out as soon as it's initialized.

[pipper]

If you live in the US, you could send it to me and I can swap the soldered RAM chips.

[StompinSteve]

Thanks for the offer pipper. I live across the great pond. I have already sent the card to EAB Member and Amiga Repair Guru "Salocinx" here in Switzerland

Shameless plug: Salocinx's website is https://amiga68k.com

[StompinSteve]

Originally Posted by grelbfarlk

I would delete your devs:Picasso96Settings file or wherever it is and start a new one.

Did that by uninstalling and cleaning up, then re-installing.

Originally Posted by grelbfarlk

It's "easy" to add a mode or tweak a mode so that an RTG board freaks out as soon as it's initialized.

Yep. But it should return to normal when everything is cleaned up and starting from scratch again. The card does not store anything as far as I was able to find out.

[Reynolds]

S3 Virge VGA cards has similar socketed ram ICs which are kinda cheap, I had to exchange on my CV64 too and that solved the issue.

[StompinSteve]

Update: It's bad VRAM. My repair-guru swapped de-soldered the first 4 chips, took the second set of 4 from their sockets and soldered them down on the location of the first 4 -> Problem gone.

So definitely one, maybe even two VRAM chips is/are defective and we are looking for replacement chips now. We will replace all 8 chips as one never knows if there are other chips with defects. Better to swap them all out.

[StompinSteve]

Update: after waiting a long time for the new RAM chips to arrive (the exact same ones as that where on the card), all 8 RAM Chips have been replaced. The card runs stable as a rock and the picture quality is great.

I would like to thank fellow EAB user Salocinx for doing the repair-work. The quality of the soldering is amazingly good. A really professional job.