The 5101= low power ram:
1) 2101 static ram = great for testing mpu board.

Post some clear high resolution pictures of the MPU board showing the repairs.
Did you try using a logic probe to check for signal activity on both of the 5101 chips?

At least you have genuine Philips 5101 chips.
Pin 9 of the 5101 in the U13 socket doesn't look right. Please check that the pin isn't bent at right angle under the chip such that it isn't in the socket.

Do you have a logic probe or oscilloscope to check for activity on the U8 / U13 pins?

If not you will have to carefully check that there is no short circuits under the new U8 socket pins and note that some signals to U8 come from all over the board. Confirming connectivity between U8 to U13 isn't enough, you need to check from the source (mostly the 6800 CPU) to U8 and U13.

check for shorts with a multi meter, do you have one? assuming yes if you've checked for continuity for traces to pins.

on the top of the board, say the 5101 at U8, start at pin 1, bottom left, and go from side to side, so 1 & 2, then 2 & 3, 3 & 4 etc.

except you will have continuity on the pins that have solder pads joined underneath.

if you do have continuity it is likely a solder pad on the top of the board is shorting to an adjacent trace.

even holding up to light, with the 5101 removed it can look okay, but a meter will confirm.

Also have to check for shorts between the pins and the thin traces that run between the pins.

I use an analog multi-meter for this.
Place a meter probe on pin 1 and swipe the other meter probe along all other pins on the chip. If the analog meter dial deflects quickly go back and recheck those pins. Then move onto pin 2 and swipe the other pins. Then pin 3 etc, etc.

Since you used strip sockets for U8, put the board up to a bright light looking for shorts between pins and traces:

MPU_Sockets 004a.jpg

Do you have an EPROM programmer?

Since you said you had corrosion - I would suspect a broken trace. Take note of the solder job on Quench's board in post #11. Note that he also soldered the pins on the top of the board. This will often fix traces that had damaged through-hole connections. Don't worry about the 90K between pins at this time, I would be more focused on traces which is where I believe your problems are.

With meter - you need to trace all pins that you affected. Start at U8, Pin 1 and follow thru to other pins that use Address Bus A3. Check using your meter at the IC, not on the bottom of the board.
Then move to pin 2, pin 3, etc

Also clean up your solder flux with alcohol and a small, soft brush. Often you can't see some solder bridges until you clean away the old flux.

I don't think your ROMs are faulty but I could write a quick program ROM to test U8 and U13 separately. But it's probably not worth it since the rework was around U8.

BTW did you check that pins 11 and 12 are connected together at U8?
Can you remove U8 and take some clear photos around the U8 socket so we can see the pin connection soldering on the top side of the board?

I can see the copper pads were not cleaned enough because the solder didn't attach properly to them. I still repair badly corroded MPU boards and can say it is very difficult to get the solder to attach where there was corrosion even after cleaning the copper to a bright color. I always solder tin the copper first to see how well it's sticking and re-clean where necessary before installing components/sockets.

Was the board booting ok before the corrosion repairs were done?

Can you confirm when you power on, the MPU LED:
Flicker,
Flash,
Flash
and that's it? Just want to make sure you're not counting the power on flicker as a flash.

The 6810 RAM and the 6821 I see are both AMI chips. Change them and any other AMI chips, and see what happens. I don't trust AMI chips at all any more.

Second that. The gray AMI chips are always the first ones to suspect.

Every chance, because the 5101 /CE1 signal is generated by signals not entirely related to the ROMs / U7 RAM.

The 5101 CE2 (Chip Enable 2) should remain high because it's connected to the /RESET signal. The /RESET signal is only active low on power up then it stays high. It should not be pulsing from time to time.

/CE1 should pulse low at the beginning of U8/U13 testing (just after the 2nd LED flash). When the software determines the 5101 has failed they will no longer be accessed/selected so /CE1 will remain high. In other words the /CE1 signal state is only interesting at the start of U8/U13 tests.

If you are saying CE1 never pulses low after the 2nd LED flash then you need to look at the logic gates that generate the signal to /CE1. These start at pins 12, 13 and 11 of U19 then to pins 9 and 8 of U18 and finally pins 9, 10 and 8 respectively of U17. You should see some activity on all of these pins. Maybe you have an open circuit to one of these pins caused by the battery corrosion or U8 socket install.

The ROM test adds the value of each ROM location as a sum. At the end of the test, the summation should be zero '00'. Technically speaking there is a 1 in 256 chance a bad ROM could result in the ROM test passing.

In your diagram I see three low going pulses on U17-8 (/CS1).
The first pulse reads the existing value at the lowest U8/U13 RAM location (for backup).
The second pulse writes the test value '01' to the lowest U8/U13 RAM location.
The third pulse reads the test value that was written to lowest U8/U13 RAM location.

The test value read does not match what was written so the program basically halts meaning there was a fault detected with U8 and/or U13.

Now if you have enough analyser probes, I would attach one to each of the four data lines on U8 and the four data lines on U13 and also probe /CS1 as the reference for when U8/U13 are written/read.
You will then be able to see what is written to these RAM chips and what is being read out to see the non-matching data.

Indeed, the diagram down below further illustrates it:

I presume the R/W signal you probed was directly at one of the 5101 and not somewhere else? The activity looks correct.

Measure the supply voltage at both of the 5101 chips with the board powered ON - i.e. make sure both the 5101 have power.

Set your multi-meter to DC voltage, place the black meter probe on pin 8 and the red meter probe on pin 22 of the 5101's.
You should measure about 5VDC

MPU200_5101_test_waveformsB.png

Hmm, 0.5V changes seems unusual. Do you currently have any batteries connected?

You mentioned you already changed both 5101 so it would be strange that the replacements are also faulty.
Do you have any other MPU board you can test them in?

Do you still have the original 5101 and can you redo the data analysis with them?

Ok, so the first test value of $01 (00000001) passes, the next test value of $02 (00000010) fails.

Can you swap the U8 and U13 chips over at the sockets and redo the tests? Lets see if the results are better.

Where did you get the 5101 chips from? If you flip the chips over do the markings in the plastic look the same? There are lots of fakes and remarked 5101 chips out there. Some are 5101 chips but slower speed chips that may only work reliably in an MPU-100 or Bally board. I’ve also heard that some are 2101 chips or just dead.

The ones pictured on the MPU board in the fifth post look original, I've bought heaps from China that look the same with no issues. I also have a box full of Chinese remarks that look very different with a high failure rate.

First picture of real ones with white markings, second picture of units from my fake box with enscribed/brown markings.

PCD5101P_Real.jpg
PCD5101P_Fakes.jpg

Any that are laser etched are going to be remarked chips or fakes. May or may not work. I seem to recall getting a batch that did have printed white markings but they were off a bit and ended up being earlier remarked chips. Often that will wear off much easier than the original printing.

From the pictures posted they look like they should be real chips but you never know. There are so many fakes floating around that every chip has to be tested. Can no longer assume a supposedly new chips is good. If they were ordered off ebay from China or from Aliexpress then the likelyhood of fakes goes up a bit.

I just ran into a batch of 20 fake 6802 chips from ebay (all dead) that I had to open a case on.

The memory chips the original poster has may be ok but I had to raise the issue since it is coming up more often and if those were fakes then possibly chasing a board issue that is actually ok.

They are probably ok then. But had to mention it was a possibility.

The corrosion on the MPU board here is actually rather mild - you and I have repaired much worse examples. We've got to be missing something simple..

So we can now assume you have confirmed one good 5101 if you can get $00 - $0F data to pass when U8 is $0x?
Can you test all other 5101 chips in U13 and try to get the initial value in U8 at $00 in order to confirm which other chips are good / bad? It seems from the early test results you did have a faulty chip in U13 (where it passed on the value $01 but failed $02).

Try a bally ROM or stern m100 ROM if you just want to see if you get past the third flash with just u8

Recommend to get PCD5101P with an original imprint. Date codes would be like mid 80s to mid 90s. Anything with a 2000s+ date code is questionable remark. I bought and resold I think 1000 salvaged pulls of PCD5101P about ten years ago and had like less ten or so dead and I never had one go bad on me once in service that I have heard about. I think someone mentioned PCD5101P brand has clamp diodes that most of the others do not. For whatever reason this 5101 memory has extremely high fail mode. When I did PCB repair dead 5101s was super common. I am not sure if it is ESD that kills them or what. But every brand, WMS, Stern, Bally, and Gottlieb boards all had bad 5101 too often.

Also check the stupid 12v to 5v circuit they use to power the RAM. Very often to find that voltage clocking in at like 5.5vdc and higher.

5101's are extremely sensitive to static discharge. Philips was the last manufacturer of 5101s and I think Andrew is correct about clamping diodes on the interface pins.
One thing the chinese have not figured out on blacktopping and remarking: tooling holes on the top of ICs. If the tool holes are clear and shiny then they are most probably real. If the tooling holes are rough then they *might* be fakes but there are some real parts that have rough tooling holes. Your image in post #5 - the PCD5101s look legit.
Quenches images from post 35 have some interesting parts. First set (the ones he said are legit) have nice, shiney tooling holes on top. Yes, i would agree they are legit.
The remainder look awful. Especially the third set where you can see the direction the sander went when removing old numbers. Some are marked with 5101 variations on the bottom. They were probably 5101s to begin with but were handled so roughly during removal that they probably didn't survive. The 5101E parts look to be oldies from RCA.
An example of how china gets a lot of their 'new' parts:

Only time I change them is when they've broken off. 35V Tantalums are usually used here and they're on the 5V supply rail so there is plenty of margin for the breakdown voltage. Don't think I've ever had one fail on a MPU board.
But it's up to you.

I don't think they recycle chips anymore like this. There are some dump sites in Shenzhen that look like this though.

I have seen videos of a multi million dollar high tech factory in china that all it does is salvage chips from pallets of E-waste for them to be sold back into the market. They can stamp on new legs to chips, Pull them cleanly, probably laser etch, and everything. I wouldn't be surprised if you bought brand X transistor from LCSC and took it too this company they would change it to whatever your favorite western brand you want. They have a process that can hide all the scuff marks on used but not remarks pretty well, if you rub a chip with acetone it will expose all the scratches from the boards being stacked on pallets 10ft tall.

I am sure very little testing goes on, but they will erase EPROMs they salvage. They will even write NVRAM chips back to all 00 so it looks like a new chip.

These are all genuine. Around 1990 dated and later should have a printed vertical bar on the left side and the Phillips shield logo.

I don't have a ton of examples left, but each one that has "kfm" also has "Korea" stamped into one of the bottom side tooling holes. "mfm" ones do not. Tooling marks are all similar to the same. Except the mfm ones do not have a country stamped on the bottom. M for Malaysia? The other tooling hole has a one or two digit number.

20220916_172904 (resized).jpg
20220916_173652 (resized).jpg

Those used to be quite easy to find but supplies have long dried up. Philips made fast and reliable parts.
The right two from 1987 were probably made before Philips finished changing the branding from Signetics to Philips - it only took them 12 years to accomplish this.

Oh - and the chinese don't laser etch remarked parts. They actually use a brownish colored ink that only looks like its laser etched. I remember when Texas Instruments started laser etching parts. 'This will bring an end to all counterfeiting!' Well, no it didn't. Hold up a brown ink part to a laser etched part - to the naked eye, you cannot tell a difference.

I bought some PCD5101s from China where the seller had pics of 100% genuine chips on the listing & no reports of blacktopping/fakes or remarks parts in their feedback, also they were more expensive.

I received remarked parts, so it's a gamble regardless buying this way. The first thing I did was look underneath the chip to look for the shiny borders running along the sides to identify the chip - all the later chips with the Philips logo I have seen have these very narrow shiny borders with a rounded edge on the top & bottom as seen in the pic below. At least I knew if they then worked they were the correct PCD5101s.

Remarked part on the left & genuine on the right.
IMG_1974 (resized).JPGIMG_1970 (resized).JPG

Manufacture date: 31st week of 1982

Probably short for Signetics. Philips bought Signetics which might also explain the early date code.

Post pictures of the chips when you get them.