David Cook
Well-known member
I recently bought a Macintosh SE/30 with missing pixels / bad video. I thought it would help others to walk through how I found the fault.
Below, we see an image a startup image with the "insert disk icon" in the middle and the cursor on the upper-left. The cursor follows the moving mouse. This is good news as it means the most of the computer is working. If you are looking on eBay for an old SE/30, seeing even a partially valid image on the screen can tell you a lot about the state of the motherboard.
Vertically, the image appears correct. However, only 1 out of every eight horizontal pixels is displayed. Most of the video circuitry must be working. Data is getting from regular RAM to video RAM and out to the screen. Vertical and horizontal blanking is working. The video is stable and clear. We are just missing 7 out of 8 columns.
Important note: I recapped and cleaned the entire computer (analog board, motherboard, and power supply) prior to this. I also checked for broken traces around each capacitor as I replaced them. Don't try to diagnose any problems until you've performed these steps, because the solution might just be the recap, cleaning, and trace repair.
The chips shown in the image below are involved in pumping video to the screen.
UC6/UC7: Video ram chips. The computer places the image into video ram, and 8 bits come out at a time (4 from each chip) in parallel.
UE8: LS166 Parallel to serial chip. This orders the 8 bits being sent from video ram into a single serial line.
UG6: Custom PAL logic chip. This takes the serial bits and adds vertical and horizontal blanking periods (the dark areas outside the image on the screen)
C7: Capacitor. This was originally an old leaky electrolytic capacitor that I replaced with a solid tantalum. But, think about the capacitor juice and what area in the video path might be damaged. *hint* *hint*
Often, one or more traces from UC6/UC7 video ram will be corroded away by C7's leaky capacitor. In fact, the first thing you should do is visually inspect for damage. Because some of the traces run underneath the LC166/UE8, it is better to use a multimeter to test for continuity on all 8 lines between UE8 and UC6/UC7. If you have only one missing vertical line every 8 bits, chances are you have a single bad trace which you can fix by adding a bodge wire.
None of the traces on my board were damaged.
Using an oscilloscope, let's see what the video out (pin 13) of UG6 looks like on a good SE/30. Below, we see a steady high signal for the horizontal blanking (the dark area on the left and right of the screen). Then, we see the each pixel go high and low to produce the checkerboard pattern.
This next image looks like I zoomed in. Nope. Same time scale. This is the output of UG6 on the bad SE/30. One pixel goes low, followed by 7 pixels high. This matches the screen. So, the problem must be on UG6 or prior chips, because the "bad" signal is already present at this output location.
I compared UG6 pin 13 (output) to pin 9 (input) and saw the same bad pattern. Aha! UG6 is not at fault. It is receiving the bad video on its input. The fault must be earlier.
Next, compare UG6 pin 9 (input) to UE8 pin 13 (output). Same bad pattern. That means there isn't a bad trace between the chips. The fault must be in UE8 or earlier.
Next, look at the parallel input pins 2, 3, 4, 5, 10, 11, 12, (not 13), and 14 on UE8. These are the bits coming in from video memory. These commonly have corroded traces. But, in my case, I see each of these pins going up and down. That means each bit, over time, is properly being delivered from the video memory.
Okay. This narrows it down to UE8 or its other traces. That is, we know the signal is bad (only 1 vertical line switches on and off) coming out of UE8, but we know the bits coming in from video ram are "good" (all switching on and off).
Tone out all of the pins coming into UE8 to make sure the fault isn't with a non-video bit trace. Good?
Remove UE8. Take the opportunity to check that all the traces underneath look good. Replace UE8. Fixed!
Why? Well, capacitor C7 leaked into nearby UE8, resulting in damage to the die or its internal bond wires.
I hope this help you diagnose video problems in your SE/30, even if your UE8 is good.
- David
Aside: I also spent time fixed the lack of sound. Usually, traces near the Sony chips are corroded. Instead, the speaker was bad. So, check the resistance of the speaker wires to make sure it isn't open or shorted before you waste time inspecting the sound circuitry. : )
Below, we see an image a startup image with the "insert disk icon" in the middle and the cursor on the upper-left. The cursor follows the moving mouse. This is good news as it means the most of the computer is working. If you are looking on eBay for an old SE/30, seeing even a partially valid image on the screen can tell you a lot about the state of the motherboard.
Vertically, the image appears correct. However, only 1 out of every eight horizontal pixels is displayed. Most of the video circuitry must be working. Data is getting from regular RAM to video RAM and out to the screen. Vertical and horizontal blanking is working. The video is stable and clear. We are just missing 7 out of 8 columns.
Important note: I recapped and cleaned the entire computer (analog board, motherboard, and power supply) prior to this. I also checked for broken traces around each capacitor as I replaced them. Don't try to diagnose any problems until you've performed these steps, because the solution might just be the recap, cleaning, and trace repair.
The chips shown in the image below are involved in pumping video to the screen.
UC6/UC7: Video ram chips. The computer places the image into video ram, and 8 bits come out at a time (4 from each chip) in parallel.
UE8: LS166 Parallel to serial chip. This orders the 8 bits being sent from video ram into a single serial line.
UG6: Custom PAL logic chip. This takes the serial bits and adds vertical and horizontal blanking periods (the dark areas outside the image on the screen)
C7: Capacitor. This was originally an old leaky electrolytic capacitor that I replaced with a solid tantalum. But, think about the capacitor juice and what area in the video path might be damaged. *hint* *hint*
Often, one or more traces from UC6/UC7 video ram will be corroded away by C7's leaky capacitor. In fact, the first thing you should do is visually inspect for damage. Because some of the traces run underneath the LC166/UE8, it is better to use a multimeter to test for continuity on all 8 lines between UE8 and UC6/UC7. If you have only one missing vertical line every 8 bits, chances are you have a single bad trace which you can fix by adding a bodge wire.
None of the traces on my board were damaged.
Using an oscilloscope, let's see what the video out (pin 13) of UG6 looks like on a good SE/30. Below, we see a steady high signal for the horizontal blanking (the dark area on the left and right of the screen). Then, we see the each pixel go high and low to produce the checkerboard pattern.
This next image looks like I zoomed in. Nope. Same time scale. This is the output of UG6 on the bad SE/30. One pixel goes low, followed by 7 pixels high. This matches the screen. So, the problem must be on UG6 or prior chips, because the "bad" signal is already present at this output location.
I compared UG6 pin 13 (output) to pin 9 (input) and saw the same bad pattern. Aha! UG6 is not at fault. It is receiving the bad video on its input. The fault must be earlier.
Next, compare UG6 pin 9 (input) to UE8 pin 13 (output). Same bad pattern. That means there isn't a bad trace between the chips. The fault must be in UE8 or earlier.
Next, look at the parallel input pins 2, 3, 4, 5, 10, 11, 12, (not 13), and 14 on UE8. These are the bits coming in from video memory. These commonly have corroded traces. But, in my case, I see each of these pins going up and down. That means each bit, over time, is properly being delivered from the video memory.
Okay. This narrows it down to UE8 or its other traces. That is, we know the signal is bad (only 1 vertical line switches on and off) coming out of UE8, but we know the bits coming in from video ram are "good" (all switching on and off).
Tone out all of the pins coming into UE8 to make sure the fault isn't with a non-video bit trace. Good?
Remove UE8. Take the opportunity to check that all the traces underneath look good. Replace UE8. Fixed!
Why? Well, capacitor C7 leaked into nearby UE8, resulting in damage to the die or its internal bond wires.
I hope this help you diagnose video problems in your SE/30, even if your UE8 is good.
- David
Aside: I also spent time fixed the lack of sound. Usually, traces near the Sony chips are corroded. Instead, the speaker was bad. So, check the resistance of the speaker wires to make sure it isn't open or shorted before you waste time inspecting the sound circuitry. : )
