Try using Pinwiki - I trust it a bit more than the Bally procedure book - there have been several revisions to that book, not sure what the revisions were for.

Locked on LED is the most difficulty issue to fix. Start here. https://www.pinwiki.com/wiki/index.php/Bally/Stern

Unless it's a typo, pin 2 of U9 should be around 4.8V and if you're getting 4.2V that would still be in range for a TTL circuit.

Post some nice clear pictures of the board.

4.2V is indicating there's no activity on that signal.

Pin 5 of the CPU is the "VMA" (Valid Memory Address) signal. When the CPU wants to access RAM, ROM or the PIAs, it puts their address on the bus and places the VMA signal high once the address bus has stabilised so the devices are sure they're being accessed. After the access period the VMA signal goes low.

In other words, pin 5 does not sit at one steady voltage. It's always pulsing. Placing a multimeter on the signal just gives you an average voltage reading.

The AMI brand of CPUs, PIAs and RAM from this period have a high failure rate. First thing you should try and do is swap the AMI CPU for a good brand part.
BTW are you getting about 5 volts on pin 40 (/RESET pin)?
What happens if you really quickly short pin 39 to pin 40 on the CPU causing a manual board reset?

By the way, were the socket replacements done before or after the board died?
The board has copper though the boards socket holes connecting the top solder pads to the bottom solder pads. These also route signal traces between the top and bottom of the board.
I'm not saying you've done it but the copper through holes will get ripped out if the old sockets were forcefully removed. It only takes one of those copper through holes to disconnect a top layer signal from the bottom layer signal causing a dead locked on LED board. Meticulous testing of every pin then needs to be done.

So it's not a reset circuit issue.

Have you determined if the clock circuits to the U9 CPU are working? They give the CPU "consciousness".
The CPU then needs address, data and control signal integrity to the U6 ROM and U11 PIA in order to kickstart and switch the LED off for the power-on flicker.

Square wave with a frequency of 533kHz on the clock signals is spot on.

DIP switches are irreverent for the power on self test.
Jumpers must be configured for the type of ROMs you're using otherwise the CPU can't read/execute valid code on startup.
U6 looks like it's a "E-720-52" under the label. If so this is a 9332 PROM chip and the board has it jumpered correctly for U6.

Since you have an oscilloscope, probe each pin on U6 during power up. All pins except 12 and 24 should show some pulsing activity (swinging between 0 and 5 volts).

Probably not damaged, but you now have another unknown.

At the U11 PIA check the following pins:
1 must have 0 volts (ground)
20 must have 5 volts (power)
21 through 33 all have some activity
34 should go from logic low to logic high on power up (reset line being released)
35 and 36 have some activity
39 should ideally go logic low to switch off the LED and to logic high to switch the LED on during the flash process

Note some of these signals will only briefly see activity before the board stalls/crashes. You may have to hardware reset the board to see activity on them to verify something is happening.

If U11 things seem ok, then try this:
Connect the middle leg of transistor Q1 to ground. Power up. This will force the board to stay in hardware reset mode. The CPU should output address $FFFE on the address bus, i.e. address line A0 should be logic low (0 volts) and all other address lines should be logic high (5 volts). Measure them at the U6 ROM using the schematics for U6 pin as reference.

No. That triggers an NMI to the mpu.

You duplicate what the reset circuit does by shorting mpu pins 39/40, then releasing (I use a screwdriver for this).

Other wise you turn the game off and on. There's no button on stock mpu board to reset it.

We all do it this way because it's convenient. But it's a little harsh on the Q5 transistor because it puts 5 volts across its collector-emitter pins that are conducting which could potentially damage it.
The elegant way is to ground the base (middle) leg of transistor Q1 which switches the transistors off. Yes I know we're all too lazy (sometimes myself included) to do it this way

So on power-up pin 39 started at 1 volt then went up to 5 volts? roughly how much time are we talking?

Without grounding the base of Q1, are you seeing the same active waveforms on A9 and A10? Are they possibly shorted together?

Have you got an EPROM programmer to check if your U6 is reading properly?

With only U9 installed (all other socketed chips removed), the CPU should be free running through every address. With your scope you should see mostly square wave activity on all the address lines. A14 being twice as fast as A15, A13 twice as fast as A14, etc, to A0 being twice as fast as A1. Are you vaguely seeing this?

The moment you power on, pin 39 of the PIA defaults on reset to being 'open circuit'. i.e. it has no effect on the circuit it's connected to.
This leaves the two resistors at R107 and R28 to switch the Q2 transistor on resulting in the LED illuminating.
After approx 10ms when the board comes out of reset, the first thing the code in U6 does is to configure U11 pin 39 as an output and pulls it low which results in the LED switching off. This 10ms is the initial LED flicker time you normally see on power up.

The R107 and R28 resistor network should result in pin 39 reading about 3.7 volts the moment you power on. If you're getting no voltage have you checked that the Q2 transistor is good? Maybe it's shorted resulting in the LED being stuck on.

The U6 needs to be read/verified as a 2532 chip.

First off, let me applaud my fellow Marylander, Enochsmoken, for putting all this effort into trying to repair this board and being able to use an oscilloscope.
But if you are successful in repairing this dinosaur, that's just what you'll have: repaired 1976 technology.

Please keep in mind, a Weebly MPU is $167 shipped, 2021 technology. Just a thought.

Valid point. But it's just the age-old question: are you preserving, or upgrading?

People in the automotive and aviation collector/restorer communities (among others...anyone who is collecting historical artifacts) face the same kinds of issues. Do you put a "modern" aircraft engine (i.e. 50's technology instead of 20's technology) into your replica? Do you upgrade the stock components in your classic car, or even change it over to an EV powertrain?

Or do you instead seek to preserve the original technology as closely as possible for historical accuracy?

Every owner has different priorities. $167 is a reasonable cost compared to the time that would be spent repairing the old electronics even if one does the work themselves, never mind the cost to take it to a qualified professional. If all you want is a working pinball machine, that's the right route to take. But if you want to preserve your 30, 40, 50 year old machine, you'll seek to repair the original parts. Or even if one is specifically interested in the repair process, rather than the "unplug and swap" process.

I mean, for that matter...any old EM machine could have 90% of their circuitry replaced by a tiny circuit board that is way more reliable. But we don't see EM collectors swapping out the guts wholesale. Same thing. A big part of the appeal is to own the original electro-mechanical workings.

On the bright side, solid-state electronics can last decades without any problem. Other than capacitors wearing out, and damage to other components when something goes wrong (often owner-induced...battery leakage and short-circuits probably accounting for the large majority of cases), there's really very little difference between "2021 technology" and the original board anyway. Once you've re-capped the boards and fixed the blown components, you've got something not much different from a brand new board.

U6 is not an EPROM, it's a ROM that's manufactured with the code mask as part of the silicon die. The specific part (9332) is not supported by any programmer as a programmable chip.
It is however pin compatible to a 2532 EPROM, commonly made at the time by Texas Instruments and Hitachi.

If your programmer doesn't support 2532 EPROMs you'll need to make a 2532 -> 2732 adapter and then read it as a bog standard 2732.
2532 and 2732 EPROMs have the same memory capacity but they have slightly different pinouts.

Note, I don't recommend performing a "verify" function as in certain circumstances it can put high voltage on a pin which this ROM is not designed for.
Simply perform a "read" operation of the chip and data compare/checksum the result against what you download from IPDB.

Checksum of F000h means it couldn't read the chip or it's dead. If you click the buffer tab and all the data you see is FFh that will confirm it couldn't be read.
Checksum 13FCh is the correct checksum for that ROM.
I think it's time to replace it with an EPROM that's known to be good. A 2532 is a direct plugin. 2732 are easier/cheaper to get but you have to change configuration jumpers on the MPU board.

This is how I always jumper MPU boards for two 2732 EPROMs:

BY-35_After.jpg

Personally I would, yes.
The first call is to have a functioning U6 in order to get the flicker on power-up. The flicker indicates the CPU has seen U6 and begun the boot process. While the LED stays locked on, U2 isn't in the equation yet.

My last purchase was from AliExpress two years ago but you might not want to wait for the slow shipping.
Have a look on ebay to see what's available locally.

You also have the option to buy pre-programmed Flash Gordon EPROMs
These are 2732
ebay.com link: itm
These are 2532 (same pinout as the factory 9332 chip you have)
ebay.com link: itm

Presuming you got the pre-programmed Flash Gordon 2532 EPROMs, plug them in (the right way around ) and see how you go.

Out of curiosity, if you read the newly acquired EPROMs in your programmer do they report the correct checksums?

Please don't give up. This minty looking board deserves to live again.

Are you familiar with how the logic gates at U14, U17, U18 and U19 work? With your scope you can see if the outputs are doing what the inputs are telling them to.

With a bright light behind the board do you have enough space between the sockets and PCB on the top side to see what's going on between pins/traces incase of accidental shorts?

MPU_Sockets 004a.jpg