Revived Quadra / Centris 610 weird pixels image glitch

[ChadVDR]

Previous thread details my path to resurrecting a stored Quadra with leaked caps but intact battery.

It now boots perfectly, reliably, and all of the inputs so far (ADB, External and Internal SCSI, Floppy) seem to work without fault. I've been playing all of the games that are on it and it's just fine in every regard including audio and speed/performance.

The final hurdle to a perfect repair is the weird video glitches. I'm working with a ViewSonic V70f CRT as well as an LG LED monitor. I have a 10-dip DB-15 to HD-15 adapter. So far it will only display anything if it is in "Composite Sync 1" or "Separate Sync" modes on either screen.

Here is the ViewSonic specs:

Fh:30~70kHz, Fv:50~120Hz
Up to 1280 x 1024
Sync: H/V separated (TTL)

Regardless of monitor used, I have strange video graphical glitches.

When booted in ROM waiting for a bootable image to load, it will have vertical bars. It's hard to see them but they have a slight dark/light gray pattern. They're all even-width and span the whole display. The difference in color isn't that great.

Once it loads far enough into the system, it will suddenly flash and the gray error will go away but a new error will appear. A peppering of inverted pixels. That is to say that across the screen in random but FIXED positions, a white pixel will actually appear black or vice-versa. They will flicker quickly like white noise but appear to stay in the same place. It kind of looks like static but its very very sparse. See image.



Any clue what may be going on? Is there possibly some grounding issue either on the board, on a connector, within the PSU? Is there possibly "dirty" or "noisy" voltage from the PSU that's creating the noise?

Things I've tried:
Tested resistance from the PSU PCB ground plane to the PSU case and it is at least 2Ω. Is that normal?
All compatible video modes (using the dip switches) on CRT and LED monitor.
All SCSI devices unplugged (internal and external)
Booting from a SCSI device (internal and external) with all or no other SCSI devices connected.
Booting from floppy.
Booting System 7 Disk Tools, 7.5.3, and OS 8.
Booting with NO SIMMS or one SIMM in various slots or the other SIMM in various slots.
Jiggling the connector and display adapter.
Different cables.
Poking things with a stick.

It's very useable, just would like to have that final finishing touch. Thanks all.

 

[joshc]

Do either of those monitors have a setting to automatically adjust settings to try and get the best picture possible?

It is worth cleaning the video port with contact cleaner just in case there is any corrosion or dirt causing an issue.

Which VGA adapter are you using? I've had mixed success with some of them. 10 DIP ones usually work well.

 

[lobust]

I'm thinking that almost has to be vram or ramdac right?

Does changing the bit depth to something lower make the problem go away?

 

[ChadVDR]

Do either of those monitors have a setting to automatically adjust settings to try and get the best picture possible?

It is worth cleaning the video port with contact cleaner just in case there is any corrosion or dirt causing an issue.

Which VGA adapter are you using? I've had mixed success with some of them. 10 DIP ones usually work well.

As for the setting, I don’t see anything that mentions auto-configuration if any kind. They really make no mention of refresh etc besides just an info pane that says what it currently being displayed.

I will clean the port and will actually check circuits to see that everything is going where they need to go ie ground pins have solid ground, attached pins are attached etc.

My adapter is a basic Amazon one. I can’t find any identifier, it’s just beige and has ten dips. If it helps to tell you how it pins out in a certain setting, I can do that maybe tonight if you let me know.

I'm thinking that almost has to be vram or ramdac right?

Does changing the bit depth to something lower make the problem go away?

That’s what awkward potato suggests. I’m hoping it’s not a dead RAMDAC. What gives you this idea? Is this exact symptom common?

I will let you know when I put it back together. Is that typically something that has been found to be true in the case of a VRAM/RAMDAC issue?

 

[lobust]

That’s what awkward potato suggests. I’m hoping it’s not a dead RAMDAC. What gives you this idea? Is this exact symptom common?

I will let you know when I put it back together. Is that typically something that has been found to be true in the case of a VRAM/RAMDAC issue?

I've never seen these symptoms on a vintage Mac so I can't say categorically, but on modern gpus that kind of artefacting is common and pretty much guaranteed to be a vram fault, usually caused by a sdram ic overheating, being overclocked too far, or simply dying.

The reason I suggested trying a different bit depth is that it will use less (or more for higher depth) vram to display the same image, and if the severity of the problem changes it would be a good indicator of a problem with the vram.

 

[AwkwardPotato]

Just to rule out anything easy: if there's any added VRAM in the slots, try testing the Mac without it.

Just to be clear, I hope nobody thinks the image I posted is my board. That is from the recap a Mac site. Is that what lead to your judgement about the VRAM circuit?

Edit: I just found the manufacturer spec for the chip which includes the pin out for the chip itself, but if you can tell me how it connects to the VRAM then that is still more helpful.

In the Q700 I mentioned in the other thread: the 68040 writes to the VRAM with the help of the DAFB (framebuffer controller IC). The VRAM is actually a special dual-port DRAM -- there are two interfaces on the chip through which two different devices (in this case, the CPU-DAFB and the RAMDAC) can perform operations. The interface used by the CPU-DAFB is the same as any standard DRAM (8-bit-wide data bus and multiplexed address bus w/ row and column strobe lines). The second, "special," interface is used to serially spool out data, eight bits at a time to the RAMDAC, forming the raster on screen.

Each of the four VRAM chips' 8-bit serial bus runs in parallel to form a 32-bit-wide bus of pixel data. On the Q700 this 32-bit bus runs into one of two inputs on a multiplexer formed by 8x 74F399. The other 32-bit input of that multiplexer runs either to the DAFB, or the 68040, or both (can't remember ATM but will update when I find the RAMDAC pinout I mentioned in the other thread). At any rate, that CPU-DAFB input to the multiplexer is presumably used to set up the CLUT and the bit depth. The 32-bit output of the multiplexer runs through 4x 8-bit-wide RC filters, part number 115S0003, into the RAMDAC.

When the Mac is in 256-color (8-bit) mode, that 32-bit "pixel data" bus carries the data for every group of 4 pixels in any horizontal line on the screen, with 8 bits to each pixel. In 16-color (4-bit) mode, that 32-bit bus carries the data for every group of 8 pixels, with 4 bits to each pixel. And so on and so forth...

Back to the Q610: the functions of the aforementioned DAFB were rolled into the djMEMC (U26) and that multiplexer circuit was made unnecessary. If you have any rotten traces between the VRAM<->RC filters (RC3-6)<->RAMDAC you *will* get vertical lines as you describe (the fact that they disappear could be a function of the bit depth changing, or the Mac switching from indexed to direct color mode). I believe the 4x 74F245 (U20, 23, 24, 25) sit between the djMEMC and the filters on the Q610. Chances are, trace rot around those or the filters could also be the cause for the bad pixels on screen (incorrect values being written to CLUT), although personally I still doubt that.

In short, inspect the legs on RC6 and U25 carefully due to their proximity to leaking caps, as well as the traces near them. You can also check that the filter chips are good by checking continuity from each pin to the one opposite it (e.g. 2->19, 3->18, etc). Should be in the few tens of ohms maximum (1, 10, 11, and 20 may all be ground). Then on one side of the filter only (doing both sides would be redundant), check for continuity between each pin and ground. Each should be on the order of tens of kOhm.

(My Q700 problem I keep referring too was that in every fourth vertical line, each pixel whose color was one of a few certain indices in the CLUT would show up on the monitor with the wrong blue value, sometimes flickering. Only happened in 256-color mode. There are no electrolytic caps on the Q700 so no trace rot, filters checked fine, multiplexers & line drivers were all fine, moving VRAM chips around didn't change anything. Problem only went away after replacing the RAMDAC with one off an Apple 8*24 NuBus card)

 

[AwkwardPotato]

Found a photo I took of that problem, sorry about the moire. Note the blue dots on Doomguy's face and the yellow lines/dots surrounding all the icons.



Also, since I know you have a deadline, the easiest solution may just be to find the NuBus riser for the 610 if you don't have one already, and use any old NuBus video card. Just be aware that any given NuBus card will most likely perform worse than built-in Quadra video, save for the really expensive and rare cards.

 

[ChadVDR]

@AwkwardPotato I have to say this is an impressive write-up. You do this for free? Now I can wrap my head around what is going on here.

Since you mentioned it, I poked around RC6 and was shocked. The pins all have continuity to different places such as the DAC, at least one goes to the VRAM SIMM slot, another memory IC etc. However, I booted the machine and while it was running I poked pin 13 on RC6 and was able to effect the snow that appeared on the screen! I could press the pad where it meets the trace and it would get better or worse instantly. This proves Potato and @lobust theory about the VRAM being a culprit. It's very obvious that the legs have some corrosion to sort out. I know I shouldn't be poking around a live board but I'm not afraid to point the finger at myself if it all goes wrong.

You two have concluded my issue. It is clearly the circuits from VRAM IC's. The DAC itself is perfectly intact at least appearance and continuity-wise.

Here's where I'm leaving this for now although I will have an exhaustive write-up with diagrams etc. once I complete the permanent repairs.

FYI I'm using one of the Apple Multiple Scan 17" dip settings on the adapter now and it's excellent. Superb.

You both suggested messing with color depths. Well, here you go.

The default setting that caused concern initially:

256 color depth
Subtle vertical pattern with snow.


16 color depth
More apparent vertical pattern.


B&W color depth
Yikes! Looks like a Mac SE with bad RAM.


All Grays
FLAWLESS picture and performance!


The picture is so crisp and wonderful with the gray mode. I've been using the system and stress-testing it and if we were going to be using a monochrome monitor this would be a triumph.

I'd call this a fully functional machine. I'm going to zip it up and call it good until after the big day with the promise that I will revisit it once they get their fill of reliving their childhood and we make a solid backup of the original (yes, the ORIGINAL) hard drive. (I had to bump it on the table to get it unstuck).

You have all helped me cross this line and I owe it to you to provide as much information as I can, including pictures of the board under the chips and whatever circuitry I can trace.

I like Potato's recommendation of just throwing a card in to bypass the internal graphics system. Would that also bypass the on-board VRAM? If I can't effect a solid repair, it may come to this. Otherwise, I have high hopes for this once lost machine. The majority of the board was coated in cap crap and I had little hope in the beginning. In time it will receive a proper ultrasonic clean and micro-repair attention. Glad to make it this far.

Thanks and leave any questions for me to address as the project progresses in the new year!

 

[AwkwardPotato]

Glad to hear that it seems to be a corrosion problem rather than a RAMDAC one!! Yes, a NuBus card would also bypass the on-board VRAM.

 

[joshc]

Corrosion makes sense. The IOSB on one of my 610s was completely affected by cap goo causing non-boot. Hopefully you can get any remaining video issues resolved with further clean up/repairing solder joints.

 

[ChadVDR]

Only thing to add for now is how to backup or upgrade the current system install on another Mac.

The Q610 is running OS 8. I'd like to firstly back this up to the BlueSCSI. As far as I've been able to find, this involves simply creating a blank image on the SD card and then dragging and dropping from the top level of the HDD to the top level of the blank image. Does that sound right? Is there any chance that any apps or something would have an invisible file that wouldn't transfer?

Second step would be to transfer this install to another Mac and upgrade it. I'm currently running OS 9 on a Mac Mini. Any guesses as far as what it would take to transfer the system from the SD to this OS 9 fresh install?

Bit of a tall order but in the end I expect to maintain this original (circa 1997) system install on the Q610 by re-homing it to the SD card in the BlueSCSI. The HDD will be dead sooner than later. Putting a copy of the system onto the Mini in OS 9 is just for a sort of continuity across systems. It's not necessary since most things can be transferred easily but thought I'd ask.

 

[joshc]

You can drag the system to another drive and it should work as long as the system folder is blessed. However - SD cards don't last very long when read/written a lot, so they shouldn't be relied on. Once you've got the BlueSCSI working, personally I would make a copy of everything over a network connection to a different Mac. I've made backups of several of my old Macs onto my G4 in this way.

Second step would be to transfer this install to another Mac and upgrade it. I'm currently running OS 9 on a Mac Mini. Any guesses as far as what it would take to transfer the system from the SD to this OS 9 fresh install?

Not sure what you mean by this?

 

[ChadVDR]

Excellent. Then I will make a few archives of the HDD for safe keeping!

The second part is kinda just a dumb idea for no real reason. Imagine you have a Mac that shipped with OS 8. Now you upgrade that system to OS 9. All of your games and settings and files etc. from OS 8 are now in an install of OS 9. Is that correct?

Assuming so, I want to sort of upgrade my instance of OS 8 to OS 9 but on a different Mac that can run OS 9. So it would be as if I put the Q610 drive into my Mini then put in an OS 9 upgrade disc.

Like I said, no real reason for this but was curious. I suppose I could just drag everything that isn't in the system folder to the OS 9 instance I have on my Mini.

 

[ChadVDR]

Thanks for helping to revive another old clunker. Now how to revise titles as "resolved".

You guys did it!
View attachment 50495

I was able to give somebody a 1994 Christmas today. The Quadra was recapped, a few IC's repaired, the shell was bleached in hot peroxide and brand new feet attached. The only flaw is a bit of snowy pixels that will be fixed thanks to lobust and AwkwardPotato pointing out how the VRAM functions.

In Gray mode, the machine performs absolutely flawlessly. All three internal drives plus the BlueSCSI perform without a single glitch. Audio is great too. ADB devices work perfectly. The only remaining issue is the VRAM IC rot that will be re-soldered after cleaning. I'm actually surprised just how quietly this machine works. I can only barely hear the PSU fan over the drive and the drive sounds just as I have always remembered big old spinning platters.

I can't tell you how good it feels to see this setup and I still owe it to you to post my findings and circuit traces since no diagrams of this board seem to exist in the public.

Merry Christmas!

 

[imactheknife]

The background picture looks like moraine lake near lake louise in Alberta