• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

What’s better - a checkerboard or a Sad Mac?

alexGS

Well-known member
Ok, so obviously the answer is neither, but I’m just about at my wit’s end :)

Macintosh Classic. The analog board had the usual leaked caps, leaky diodes, faulty optocoupler, etc. so the wobbling checkerboard was fixed by getting the voltages right - a happy bong, and all good.

But that was a logic board borrowed from a working machine. The other logic board, paired with the repaired analog board, produces a checkerboard pattern. This time it’s stable, not wobbling, because voltages are spot-on.

I recapped the logic board and cleaned it, tried swapping the ROM - no difference.

The (blue) lithium battery had not leaked - but the capacitors had, as you would expect.

Because the checkerboard appeared without the expansion RAM board fitted, it seemed the soldered RAM was faulty. I found a 1MB SIMM, checked the data sheet for the chips (seemed correct), got out the hot-air tool, and swapped the RAM chips onto the logic board.

That got me a screen with multiple, partial Sad Mac icons. Pressing on the RAM chips filled in the missing parts!

I took all the RAM off, bought some solder paste, and carefully soldered the chips on again.

This time, I got the correct Sad Mac display (photo).

It was at this point that I realised the chips I’d used from a 72-pin SIMM have three address lines, not four.

So I removed the chips from the Classic’s RAM expansion board (these being the only other ‘correct’ chips that I have). Soldered them on to get another multiple-Sad Mac partial display, and yes, pressing on the chips caused various lines to flash.

Reflowing all the RAM chips again and tidying up produced - a checkerboard :D

So I’m right back where I started, but the board looks lovely and clean now that I’ve cleaned it so many times, both with circuit board cleaning solvent and with hot soapy water.

The fourth photo shows the logic board as it stands now, with correct RAM (I think) - this produces the checkerboard display.

Was replacing the RAM a red herring? Or was I further ahead with the Sad Mac? Is it time to get out the oscilloscope?

Cheers for any words of inspiration
-Alex
 

Attachments

  • 3F776C2B-B1F1-4923-9EE5-7C633CEE0B2B.jpeg
    3F776C2B-B1F1-4923-9EE5-7C633CEE0B2B.jpeg
    175.8 KB · Views: 22
  • D7CCB74B-12CA-4FC3-B8F8-0AC00F606D3F.jpeg
    D7CCB74B-12CA-4FC3-B8F8-0AC00F606D3F.jpeg
    92 KB · Views: 28
  • 16A87936-5505-4DBA-B13E-B5F4BDF3EEA2.jpeg
    16A87936-5505-4DBA-B13E-B5F4BDF3EEA2.jpeg
    43.6 KB · Views: 34
  • 80BF6E21-2FE0-4BD5-A0B4-D216A4EB635F.jpeg
    80BF6E21-2FE0-4BD5-A0B4-D216A4EB635F.jpeg
    154.1 KB · Views: 33

Phipli

Well-known member
Sadmac because it means the processor and video are running ok, and probably most of the ROM.

Plus it gives you an error code :)
 

Phipli

Well-known member
Having actually read through your post, I think you were actually in a better situation when it was producing a clean sadmac. I would say keep messing with the RAM until you get back to a sadmac with RAM you believe is right. See if you can exactly match the original chips.

The checkerboard pattern means there is something extra wrong over the sadmac, and like I said, at least with the sadmac you can start investigating.

Edit : I mean, getting a sadmac meant your RAM was working pretty well. The video memory is spread through all of the RAM columns to do RAM refresh, so with video working, it means your RAM is at least mostly working if not completely.

I'd put the 72pin chips back as a starting point and investigate the sadmac code.
 

alexGS

Well-known member
Having actually read through your post, I think you were actually in a better situation when it was producing a clean sadmac. I would say keep messing with the RAM until you get back to a sadmac with RAM you believe is right. See if you can exactly match the original chips.

The checkerboard pattern means there is something extra wrong over the sadmac, and like I said, at least with the sadmac you can start investigating.

Edit : I mean, getting a sadmac meant your RAM was working pretty well. The video memory is spread through all of the RAM columns to do RAM refresh, so with video working, it means your RAM is at least mostly working if not completely.

Thank you Phipli for your carefully-considered reply :) that’s just what I was thinking, the clean display of the Sad Mac albeit with the error code (bus error?) makes me think it must have been working to some extent, where a checkerboard pattern feels like the RAM isn’t working. As you said.

Also, if it was a RAM fault in the first place, then the expansion board fitted to the same logic board perhaps suffered the same fate and also has faulty RAM (which I’ve now transferred to the logic board).

In my collection, I can only find other RAM chips that are larger, with the same pinout - eg. 1Mx4 bit word (4MB SIMM) instead of 256Kx4. I wonder if they will work; whether the logic board will use only the first 1MB.

Thanks again for confirming what I was thinking :)
 
Last edited:

alexGS

Well-known member
Sadmac because it means the processor and video are running ok, and probably most of the ROM.

Plus it gives you an error code :)
The error code 0000000E 0000F000 seems to be a rare one, I had some trouble trying to find a direct translation.

The most plausible explanation to me seems to be that some of the RAM was working and the upper part was not. The original RAM had D0-D3 pins, while the RAM I used had DQ0-DQ1 and then two DQ3 pins. Maybe a typo in the Motorola data sheet. Perhaps the chips were damaged by the removal and installation methods. That lead-free solder paste takes a lot of heat to melt. This time around, I might try conventional solder and an ordinary iron.

Data sheet for the chips I used:
 
Last edited:

Phipli

Well-known member
The error code seems to be a rare one, I had some trouble trying to find a direct translation.

The most plausible explanation to me seems to be that the original RAM had D0-D3 pins, while the RAM I used had D0-D2 and then a second D2 pin, assuming that wasn’t a typo in the Motorola data sheet. Perhaps therefore most of the memory was working fine (making the screen display) and the upper part of memory was not.
That sounds like a typo to be honest.

Where are you based?

Edit : ah, New Zealand.
 

alexGS

Well-known member
That sounds like a typo to be honest.

Where are you based?

Edit : ah, New Zealand.
It does doesn’t it, I misread it actually, thought it had two DQ2s. Having two DQ3s and no DQ2, as it says, seems impossible.

I reckon these chips will work and I just damaged them. The 1MB 72-pin SIMMs were taken out of Macs during upgrades, long ago. I have at least half a dozen SIMMs; plenty to try again.

Yep, New Zealand, Papamoa Beach on the east coast
 
Last edited:

Phipli

Well-known member
Yep, New Zealand, Papamoa Beach on the east coast
Bit of a swim for me to pop over with a simm, I'm on the wrong side of the planet.

Sounds like you have some others spare. Do you have a hotplate? Try preheating them to 150°C and then using the hot air to just lift the solder above melting point briefly? It might be easier on the chips than just hot air - more controllable.
 

alexGS

Well-known member
Sadmac because it means the processor and video are running ok, and probably most of the ROM.

Plus it gives you an error code :)

I think, possibly, we were wrong.

Today I’ve fitted two different sets of RAM chips and figured out that this happens;

- When address pins are all correctly connected, a clear Sad Mac icon and codes appear. (If some address lines are not connected, there are vertical lines missing from the icon/codes.)

- When address and data pins are all correctly connected, a checkerboard appears. (If some data lines are not connected, there is the Sad Mac and the 0000000E - 0000000F codes, for example. I had 00000F0F at one point when two data lines were not connected).

It therefore seems that a checkerboard is “better” than a Sad Mac, in the sense that if the RAM is partially connected, a Sad Mac appears, and when the RAM is completely connected, a checkerboard appears.

I’ve been verifying the connections from pins on the BBU (the large custom chip) to the RAM chips DQ0-DQ3 (data bus), and A0-A8 between the RAM chips (address bus).

I wonder what to do next?
 
Last edited:

Phipli

Well-known member
I think, possibly, we were wrong.

Today I’ve fitted two sets of RAM chips and figured out that this happens;

- When address pins are all correctly connected, a clear Sad Mac icon and codes appear. (If some address lines are not connected, there are vertical lines missing from the icon/codes.)

- When address and data pins are all correctly connected, a checkerboard appears. (If some data lines are not connected, there is the Sad Mac and the 0000000E - 0000000F codes, for example. I had 00000F0F at one point when two data lines were not connected).

It therefore seems that a checkerboard is “better” than a Sad Mac, in the sense that if the RAM is partially connected, a Sad Mac appears, and when the RAM is completely connected, a checkerboard appears.

I’ve been verifying the connections from pins on the BBU (the large custom chip) to the RAM chips DQ0-DQ3 (data bus), and A0-A8 between the RAM chips (address bus).

I wonder what to do next?
Check for shorts on the address lines perhaps?
 

Phipli

Well-known member
If you work out which pins not being connected causes the checker board to go away, thats a major clue.

Not sure how it is able to display correctly with partially connected RAM. No graphical glitches? That tells us something about where the issue is. Somewhere the VRAM isn't.
 

alexGS

Well-known member
Check for shorts on the address lines perhaps?

If there are shorts on the address lines, the Sad Mac does not display completely (ie. has glitches). If the data lines are all connected but not the address lines, the checkerboard has glitches. I found that out when I fitted 4MB-worth of RAM instead of 1MB and interestingly had a checkerboard with squares instead of rectangles. The glitches were fixed when I resoldered the pins for the address lines; I then had a perfect checkerboard (still not much use but twice as many squares as usual; as though it detected 2MB RAM on the logic board).

If you work out which pins not being connected causes the checker board to go away, thats a major clue.

Not sure how it is able to display correctly with partially connected RAM. No graphical glitches? That tells us something about where the issue is. Somewhere the VRAM isn't.

If any of the RAM data lines are disconnected, the checkerboard goes away (and is replaced by the Sad Mac). The second code, e.g. 0000000F, indicates which data lines are missing. The display is without glitches.

My guess is that when the RAM is fully connected, it passes to the next stage of the power-on self test and displays the checkerboard - it makes a noticeable pop from the speaker that wasn’t apparent when the Sad Mac was displayed.

I’ve been checking the lines between the ROM and the CPU. Haven’t found a fault so far.

I wonder what the checkerboard really means…
 

Phipli

Well-known member
If there are shorts on the address lines, the Sad Mac does not display completely (ie. has glitches). If the data lines are all connected but not the address lines, the checkerboard has glitches. I found that out when I fitted 4MB-worth of RAM instead of 1MB and interestingly had a checkerboard with squares instead of rectangles. The glitches were fixed when I resoldered the pins for the address lines; I then had a perfect checkerboard (still not much use but twice as many squares as usual; as though it detected 2MB RAM on the logic board).



If any of the RAM data lines are disconnected, the checkerboard goes away (and is replaced by the Sad Mac). The second code, e.g. 0000000F, indicates which data lines are missing. The display is without glitches.

My guess is that when the RAM is fully connected, it passes to the next stage of the power-on self test and displays the checkerboard - it makes a noticeable pop from the speaker that wasn’t apparent when the Sad Mac was displayed.

I’ve been checking the lines between the ROM and the CPU. Haven’t found a fault so far.

I wonder what the checkerboard really means…
It means the Video circuit isn't being told to draw the halfttone. Which is weird because it is clearly capable of drawing the sad mac.
 

techknight

Well-known member
The error code 0000000E 0000F000 seems to be a rare one, I had some trouble trying to find a direct translation.

The most plausible explanation to me seems to be that some of the RAM was working and the upper part was not. The original RAM had D0-D3 pins, while the RAM I used had DQ0-DQ1 and then two DQ3 pins. Maybe a typo in the Motorola data sheet. Perhaps the chips were damaged by the removal and installation methods. That lead-free solder paste takes a lot of heat to melt. This time around, I might try conventional solder and an ordinary iron.

Data sheet for the chips I used:
On the Macintosh Portable, I got an E code due to the R/W line being broken from the CPU and GLU
 

alexGS

Well-known member
On the Macintosh Portable, I got an E code due to the R/W line being broken from the CPU and GLU
Thanks.
The interesting thing is that I had the codes when there were soldering faults, and once I find and fix all the soldering faults (including the output enable lines, etc.), then I have a checkerboard instead. That’s happened with two different sets of RAM chips now.

I don’t know what my next move would be with that logic board. What do you reckon would cause a checkerboard?

-Alex
 

enigmaesque

New member
Alex, by any chance, have you tried simply allowing the machine to run for a bit? I was having similar issues (sad mac, wobbly checkboard) with the logic and analog board, and noticed that if left running for a bit, my Macintosh Classic would eventually auto-boot after about 10 minutes. After this, rebooting would only take about 1 minute. I wonder if it has to do with the voltage, or if the process of waiting a bit allows the hardware to extract the voltage needed for the OS to run properly. I mean, I know it's not a permanent fix, but I searched online for months, replacing all sorts of capacitors, batteries, even the optocoupler, only to find out that the machine only needed a bit of time to boot up. I still haven't figured out why it does this, but after cleaning it, replacing parts, and doing all sorts of tweaking, I figured if it works, it works haha, I don't mind waiting the initial 10 minutes. Just thought you should try, the first time it took me an hour or so, then after that it did it in 10 minutes. Try plugging in your machine in different outlets too, sometimes being closer to the breaker makes a difference.
 
Top