I got a FirePower with a System 6 MPU that wouldn't boot. I've reviewed threads from Vid and Clay, and taken all the recommended steps on the MPU, including:
○ Replaced all chip sockets on MPU & driver, and confirmed no shorts between pins (all were SCANBE)
○ Replaced 40-pin interconnect, male and female, and confirmed no shorts between pins
○ Replaced IC19 CMOS RAM (5101) with an NVRAM
○ Replaced IC13 and IC16 MC6810 RAMs (and swapped various replacement 6810 chips into these sockets while debugging)
○ Replaced IC3/IC4/IC8 on the MPU (8T97 hex buffers, replaced with 74LS367)
○ Replaced header pins and connectors for 5v power supply

Now the MPU is on the bench with a computer power supply. Voltage at the CPU itself is slightly low, around 4.6-4.7v. Leon's test ROM flashes the MPU LEDs, and correctly toggles the PIA outputs on MPU IC18 and driver board PIAs IC5, IC10, and IC11. All good. But when I run Leon's RAM test, it shows that IC13 (6810 RAM) is bad. Specifically, the MPU LEDs stay dark for about 5 seconds when I push the diag button, unless I push the diag button at the moment the MPU LEDs are lit in which case they stay lit for about 5 seconds. No flashing of single LEDs. I swapped out IC13 (as well as its neighbor IC16) for new-old-stock 6810's. I have 5 of them but get the same result from Leon's memory test no matter which ones I install.

With Leon's ROM, I can see the blanking signal toggling HI/LO at pin #37 on the 40-pin interconnect.

I have been getting this same result from Leon's ROM for awhile now. I replaced address buffers IC3, IC4, and IC8 with 74LS367s because I thought the inputs and outputs sounded slightly different from each other based on the tones from my logic probe (I don't have a scope unfortunately). But the input/output pairs on IC3/IC4/IC8 were always toggling with Leon's ROM. Anyway I replaced all 3 buffers with 74LS367's and saw no change. The outputs essentially match the inputs (i.e. high speed switching) although the tones from the logic probe sound a little different. Perhaps there is some improper low impedance on the address bus, which is inappropriately loading the output of the buffers and distorting the output signals? (But if that were true, how the heck could Leon's program load from the ROMs, which are also addressed by IC3/IC4/IC8?)

Data buffers IC9 and IC10 appear to be working perfectly - I measure identical signals on their input/output pairs with my logic probe. Hi speed switching and they sound identical.

I've confirmed the pins of the 6810 RAM chips (and the 5101) are either toggling during Leon's toggling, or they at least blip momentarily during Leon's memory test (in the case of the chip select signals). And I've at least confirmed life from IC7, IC8, IC11, IC12, and IC25.

I started getting suspicious of Leon's RAM test, so I put in a set of Scorpion ROMs and connected the driver board, still on the bench. At powerup, the MPU LEDs turn on momentarily and then off - they are not stuck. I ran the built-in Williams memory test and it lights the lower MPU LED. That means there is a problem with 6810 RAM IC13 or IC16. So, the Williams memory agrees with Leon's memory test, even though I have tried multiple 6810 RAMs in the sockets for IC13 and IC16.

With the game ROMs installed on the bench and the driver board connected, I checked the blanking signal at pin 37 of the 40-pin interconnect and found it LOW. Not good. I traced this back to pin 4 of the MPU board PIA (IC18). At power-up, pin 4 of IC18 toggles HI/LO about 4 times, and then goes LO. So, I have a blanking problem that may be a symptom of my RAM problem. Again, with Leon's ROM, the blanking signal does propagate through the 40-pin connector to the driver board - it's only when I install game ROMs that it fails, and it fails upstream at the PIA. So I don’t really think the problem is the blanking timer circuitry itself.

At this point I am not sure what to test/replace next. I guess I could shotgun all the chips that touch the 6810 RAMs, like IC7, IC11, IC12, and IC27. But I've already confirmed those chips are alive, so I doubt they're causing the problem.

Thanks Rikoshay. I do have two 6808's that give the same result. Would be interesting to try it with a 6802, jumpered to use internal memory. I guess if there is a connectivity issue between the CPU and the IC13, the 6802 would neutralize it. But, the other two memories need to use the address/data bus as well, so it seems like any connectivity issue would simply reoccur for those chips.

I don't have the original 5101 anymore - it was damaged when I removed it, and I think it was suffering damage from battery leakage anyway.

Thanks Budwin. Fair point, and frankly my parts cost with shipping so far is teetering toward the price of your board. I still have some patience left for this, but may come knocking at your door eventually.

I made a big head-slapping realization. Even though I changed all the sockets and swapped some ICs on the MPU, I had never changed the J3/J4 jumper. When I got the machine, it had all 6 original Firepower ROMs and was jumpered at J4. For my debugging, I'm switching between Leon's test ROM and some Scorpion ROMs, which don't include chips for the upper row of ROMs (IC21, IC22, and IC26). My understanding is that if you're not using the upper row, you should jumper J3 and not J4. So I made that change. Now, the Scorpion ROMs appear to be booting up on the bench (both lights turn on briefly and then off) and the blanking signal is *HI* at the 40-pin connector. AND, when I run the Williams memory test, it appears to pass (two flashes, followed by lights out). I have not reinstalled the boards in the machine yet so can't confirm they're working all the way.

HOWEVER, Leon's ROM is still giving the same result on the memory test. Specifically, the MPU LEDs stay dark for about 5 seconds when I push the diag button, unless I push the diag button at the moment the MPU LEDs are lit in which case they stay lit for about 5 seconds. No flashing of single LEDs.

Recall that the machine was not showing anything on the score displays when I got it, and was showing only GI lights at power-on. Now, I'm seeing a blanking signal and the boards are passing internal Williams tests. So I am leaning toward ignoring Leon's RAM test (which people seem to distrust anyway) and reinstalling the boards in the machine. But first, I need to change some header pins on the driver board, and then I guess it's time to make the modifications to the MPU board so that it can use the Firepower combo rom (which I have). I am a little torn because I want to see whether things are working before I modify the board and switch ROMs. But I don't really want to repeatedly take the boards in and out of the machine. I suppose I could just hook up the connectors for MPU power and the score displays, and test it without hooking up all the lights and solenoid connectors...