OK, sometimes I have the feeling my posts are too detailed and I'm boring people....
Here goes:
What you need to have for this are the schematics and a little understanding how these things work.
The schematics you can find at
http://arcarc.xmission.com/This one was of use the most for this situation:
http://arcarc.xmission.com/PDF_Arcade_Atari_Kee/Centipede/Centipede_DP-182-5th-01B.pdfSince my PCB was completely frozen I had replaced a lot RAMs before and replaced the CPU. Also some more 74LS chips near the CPU.
On the KLOV forum, Mark Speath pointed me into the direction that this was probably NOT a RAM problem, since the game doesn't actually need any RAM to start-up. It makes sense, first thing a computer (which an arcade PCB is of course) when it starts up is read the very first adress for it's first CPU instruction. My PCB didn't seem to get that far...
Here's a pic how it looked when started up:
I could push the reset button, but it remained completely static, like it was frozen.
When you check the schematics, you can see that everything "starts" with the clock generator, using a 12 Mhz crystal and a transistor Q1. I checked those with the Oszifox and it was oscillating happily. The clock signal is routed to some counters/dividers. One branch of these make the H-sync and V-sync signals. Since I had a stable picture, I figured these were OK.
The other branch of course goes towards the CPU. I checked the clock signals with the Oszifox on the test-points and on the CPU pins and they seemed OK. I had also checked the ROM select signals (there are test-points for each one) and these were constantly high. This meant the CPU was not addressing the ROM.
So, I figured something was holding the CPU. I also felt the overheating Q2 and Q3 so I measured the transistors and they seemed good. Then I measured (diode-check setting of DVM) across R7 and got a beep (=short). I re-measured it in Ohms setting: 0 Ohms. It should be 10k. I desoldered one leg of the resistor and it measured out fine, the short was somewhere else (usualy, the passive parts (like resistors, condensators etc. don't go bad, the active parts like diodes, transistors, IC's are most likely to go).
Checking the schematics I figured it could only be L2 since pins 6 and 7 are parallel to R7 and +5V. I cut off pin 6 and 7. The short across R7 was gone. Checked between pin 5 (+5v input of L2) and pin 6 and there was the short !
I fired up the PCB again with the two pins cut off. Now I could see something happening when pressing the reset button. (The garbage changed when pressed, and changed back when released....)
L2 controls the reset signal going to the CPU.So I checked the reset pin on the CPU (pin 40). When this is low, the CPU is in reset, so it should measure +5V when running normally. I checked it and it was low. So the CPU was constantly in Reset.
Then I figured "what the heck" and shorted a +5v testpoint to pin 40. First I did it very briefly. But it was enough to get the complete playfield
Then I shorted it longer and the attract ran....
Hello, everyone still awake ?
P.S. Anyone needs a almost brand-new (two times used !!!!) Centipede to Jamma adapter ?