Is there a good resource for troubleshooting a board issue with an LCII?
Problem is strange, video has a pattern of vertical lines that are tied to bit-depth, almost like bad bit(s). But it's not the VRAM nor RAM. Apple Personal Diags tests both are fine.
If I could share a video it's intersting because it's also not a "stable" error, meaning there are parts of the screen where it's fine, but that's not static and can move around dependings on whta's displayed.
It's almost like there's a problem with just the video output maybe? The other thing to note, is even wiht the picture looks "good" there's movement, the underlying singal isn't stable.
Here are some shots to show some examples, notice on the hightest bit-depth the screen may look good, but the color bars are clarly bad. I don't have a schematic for the board and the CLUT chip seems to be a likely suspect, but I don't know this design at all.
The system was purchased as "no power" which is correct, the TDK PS only had stable 5V output, 12V and -5V are basically non-existant. Knowing the problem with the SMD electrolytics on the LCII I depopped all the electrolytics, thourghly cleaned and recapped (don't worry about my work, I know what I'm doing). Some may be interested to know that of the 17 caps only the 1uF, 100uF, one 10uf and one 47uF were good. Everything else on the board was dead including all 3 47's tied to the supply lines.
The VRAM stick wsa also in what looked to be surprisingly bad condition, but mainly because of what looked like old soldermask on pins.But... could have also been corrosion from wicking up electrolyte I don't know. The connectors were cleaned and restored best I could, luckily the bad ones are NC so this work was really just slapping lipstick on the piggy.
Sorry about the morie on the images, photos of the LCD didn't work well... but the problem can be seen when full-size.
Do all the pins of the video chip (Sierra semi, I believe) look in good shape? Does anything happen if you put pressure on them with a plastic spudger? Could also be electrolyte that wicked under chips or through vias.
One observation:
- When you have 2 colors (1 bit per pixel) every 16-th pixel is bad.
- When you have 4 colors (2 bits per pixel), every 8-th pixel is bad
- When you have 16 colors (4 bits per pixel), every 4-th pixel is bad.
This looks like a bad bit along a 16-bit bus. From the first picture it's possible to tell that it's one of the middle bits, either bit 7 or bit 8.
I'm not sure which "video chip" you're referring to. There's a chip with label "CLUT DAC" which is the only thing with a video related label. There the AMD 85C80, the 68030, the SWIM and other than that all I remember on the board is various 7400 logic stuff and the base RAM.
Yes totally possible some electrolyte wsa wicked up under the chips, but the board has been thoughly bathed and deep cleaned. The only "normal" thing which wasn't done was ultrasonic. But was in a wave bath, soaked and fluid injected under the chips to push out anything left. I felt it was a good deep clean... and I doubt it can be any cleaner but I'm also willing to give it another pass if that's the best suggestion!
I'm not sure about the via comment... anything stuck in vias may be there, but that's a single path so... confused by what you're suggesting with that.
Yeah I totally agree and which bit was my biggest question... But then RAM and VRAM testing with APD passed, both short and long RAM test.
I was thinking the same at the start and feel I've done everything to eliminate what I think would be bad. The dynamic nature to the overall distortion almost looks like a power thing almost like a cap can hold up, but the only caps left on the board are the tantilumns (I think that's all that's left) and probing those voltages look close to the "good" LCII I have. The fact that RAM and VRAM check out has me puzzled given the look of the VRAM stick I wouldn't have been surprised if that was the guilty party, but swapping VRAM with the two systems produces the same resuls, good one remains good with ugly VRAM, and bad system remains bad with good VRAM.
I am tempted to find a memory viewer and examine the VRAM data to see if I can see the problem, but given the dynamic nature I'd have to catch a little luck for this to work.
I did consider reflowing the CLUT DAC pins, but they look really good.
If it's not the VRAM stick, then it is its bus (socket, traces and soldering) or whatever chip it’s terminating in.
Thanks, but wouldn't APD's RAM test pick up on those?
Also don't see any MMU onboard, right?
Do you have a schematic for this? I'm kinda curious to look at VRAM in a memory viewer to seee if it's stable. I suspect it is, and scanning done after scanning the memory for output is where the roblem is... . do you know anything about the CLUT DAC? Assuming that does the conversion for the video output... right, just the color generator...
When liquid leaks from components, it can be so thin as to wick into areas that appear sealed. Vias are an example. Because a via is a metal ring plated into fiberglass epoxy, there is a small gap on the outside of the via where electrolyte can wick into. If there are connections to interior copper layers at that via, the electrolyte can wick into them as well. I don't know any way to effectively clean this when it happens, but it is a reliability concern because over time and with an electric field applied, the electrolyte can facilitate copper migration or corrosion. What can then happen is that a trace on an inner layer breaks (corrodes through) and the only way to repair that is with a jumper/fly wire.
The LC II has a MC68030 onboard which is a CPU with a built-in MMU. The only Macintosh computer with a separate MMU is the original Mac II.
Here is the schematic:
LC II schem.pdf
The 'CLUT DAC' (more commonly known as a RAMDAC) is a off-the-shelf component that should have its own published datasheet.
Nice! The first thing you would want to ring out with a multimeter is the 16-bit bus between the VRAM slot and the RAMDAC. Also make sure none of those caps are shorted to ground. Check the values of the resistors too.
VRAM Bus.png
Thanks, was hoping that was a little extreme but understand anything can happen and given the state of the cap... I shouldn't be surprised by anything!!
Thanks for the schematic, I'll start probing the board later and see where that takes me!
@CVT thanks for the suggestion too, the schematic will be my map, and totally agree with your suggestions. The fun of incircuit testting begins!
Following up, all resistors in the input to the CLUT DAC are as expected, the caps... those are a little more interesting. They're tacked to the board so I didn't want to try and tombstone them for readings so I grabbed and ESR meter and those all show like they aren't reading. I checked a few others of the Bar2A markings on the board and those were showing an ESR. I'm guessing the ones in that input are fine but can't verify. I did ohm them out with a dmm and nothing stood out different than the others... I removed the VRAM socket and eveyrthing underneith looks ok except near one of the plated pin holes has some scaring which doesn't make sense. It alos looked like that's a NC pin so may be unimportant.
The socket does not look great, (ie not plated) so that's a concern. I suspect there's nothing new I can grab and slide in, but who knows.... I'll take a look if I can even figure out what it's called I suspect the VRAM may use an Apple pripority socket. All contacts basically look the same as this half, it's not great but also shoudn't stop it from working. Ignore the fiberglass brush bristles, I was trying to clean it before deciding I should just pull it and look for something to replace with...
2020_1106_111538_003.JPG
Guess I should also note, there is no "boing" when starting. I didn't think to chekc the headphone jack before pulling out the VRAM socket so will have to wait on that or maybe I could try without the VRAM which I know is required to run, but not about the boing.
Been looking around for that socket and found this on vcf which reminded me of the boing.
VCF link with the boing
ESR meters will not detect ceramic capacitors below 0.22 µF, due to their method of testing. Since the capacitors are just bypass to ground, the only failure mode that would explain a stuck bit would be if one of the ceramic caps failed short.
Ugggh, looks like the guess about propritory was spot-on... 68 pin eh?
Since 72 seems to be the standard, would it be possible to disassemble a 72-pin and migrate all minus 4 connectors to mac socket? Seems like a PITA to do, but also seems doable if the sizing of the current TE sockets will match. Need to figure what left polarization mean...
3 donor options found on mouser
Thanks I think the ones on board are 100p so yeah way below, and I did check there are no shorts.
The "stuck" isn't really stuck, as it changes between off and on. Sometimes there's a black line other times white line! That's part of the reason I was wondering about the signal "support" from other caps in that part of the circuit. I'm going to walk around the output past the CLUTDAC which BTW I coudn't find a datasheet on, at least not the one mentioned in the schematic. Exepct that's an Apple custom part so need to find a compatible model. The one on this board is even branded AT&T and I don't remember them as a big supplier.
New replacement 68-pin SIMM sockets are available here: https://www.peconnectors.com/sockets-pga-cpu-and-memory/hws4036/
Thanks!! I took a look and am concerned the description states gold plated contacts. The VRAM has tin fingers. Tin and gold are dissimilar metals that don't play well together.
That said, the picture doesn't look like gold contacts so that may be wrong or is just super thin plating... may call and ask.
Confirmed only gold contacts only on the peconnecotrs.com, so looks like I have to depop and repop one by hand.
While there is still continuity and low resistance with this guy, I'm sure once we start putting speed into the mix this could be problematic.
SIMM Contacts.png