Centipede repair log

pcjason

New member
Joined
Mar 29, 2010
Messages
4
Reaction score
3
Location
California
Hello,
Just thought I would post details on a few of the repairs that I just performed on a Centipede cocktail.

Background: This game was sitting in a garage without climate control in Lake Tahoe for over two years. It was placed there after it stopped powering on.

Issue #1: No power to the game.
Diagnosis: After ensuring that the power switch and interlock switch were functional with a meter, I checked the power cord for continuity from the plug prongs to the power brick. It was found that one wire had no continuity.
Repair: Replaced power cord and we now have power.


Issue #2: R30 was smoking on the AR II board.
Diagnosis: Should not be a surprise, but the edge connectors were not making very good contact, causing one of the sense lines used for feedback to the AR II to be floating.
Repair: For now, I have shorted the sense lines right at the AR II. New edge connectors are on order.


Issue #3: The monitor showed garbage and was waving all over the place. Almost looked like a loss of the sync signal.
Diagnosis: As with any first cut debugging, I started with checking the supply rails. The +5V rail measured 4.4VDC at the board. Before adjusting the AR II to raise this, I measured the AC ripple, as I did not want to fry anything if there was excessive ripple. The AC ripple measurement showed 800mVAC on my Fluke meter, which is *way* beyond a reasonable ripple value.
The power supply design is very simple, consisting of a step down transformer, a bridge rectifier, and a large filter cap to produce an unregulated 10.6VDC that is later regulated. Testing the unfiltered supply showed the excessive ripple there as well. There are only two things that could cause this. Either the bridge rectifier has an open diode, meaning a part of the AC cycle would be missing and the filter cap would have to do more work than it is capable of, or the filter cap is bad. Testing the bridge rectifier showed no issues, so the issue had to be the filter cap.
Repair: Replaced the large filter cap ("Big blue"). After replacing this, the +5V rail measured 5.15VDC at the board and the showed only 15mVAC ripple. This also caused the monitor waving to stop. At this time, I adjusted the supply down to 5.05VDC at the board.


Issue #4: The game shows non-static garbage on every power up. The garbage consists of random letters, numbers, and parts of sprites all of the screen. The background color changes randomly, garbage tones are heard, and the coin counter randomly increments.
Diagnosis: Firstly, I confirmed that all supplies were at the board and within spec. All supplies were fine. I had a working Centipede board on hand, so I threw that in just to verify that power was okay. The working board also worked in this machine, so power issues were ruled out.
Next, I swapped all socketed components between the two boards. After this exercise, the working board still worked, and the broken board was still broken.
Next, I did a thorough visual inspection of the board for bent leads, broken traces, or broken components. The visual inspection showed nothing abnormal.
Finally, I started probing the board. I have misplaced my logic probe, so I had to use my scope instead (poor me ;)). Probing the reset line on the CPU showed that the watchdog was firing continually, so the program was not running correctly. A quick probe on the clocks showed the correct frequency. Also, the address and data bus showed activity, so there was nothing grossly wrong there. I started to probe the program ROMs, and found that the /CS for the ROM at E1 was only toggling between 1V and 0V, so this ROM was always active! This is driven by the 74LS139 at J2. I removed the ROM in order to rule it out as the cause of the signal being pulled down. Even with the ROM removed, the signal still showed a 1V peak-to-peak swing, so the culprit had to be the 74LS139 at J2.
Repair: Replace J2. The board now boots and runs correctly!


Hope that helps somebody!

-Jason
 
Testing the bridge rectifier showed no issues, so the issue had to be the filter cap.

Jason, how is this done exactly? I've thought to check rectifiers for proper function myself, but lack the knowledge of how to test them.
 
Well done!! What is this AC ripple on a DC line thing? I've gotta admit, in my games if the 5V voltage was a scoche off, I think most people's thought would be to "turn the knob". I guess perhaps 4.4V is a bit out of the normal bias range for that adjustment.

Without an oscilliscope, how can one use a nice multimeter to make the same diagnosis and verify fix, or is this not possible?
 
You can check AC ripple on a cap by setting your meter to AC and testing across the ends of the cap. The variance in voltage is the "ripple".
 
Jason, how is this done exactly? I've thought to check rectifiers for proper function myself, but lack the knowledge of how to test them.

It is pretty easy to check bridge rectifiers for gross issues, since they are nothing more than four diodes. A bridge rectifier has four terminals, two of which are the AC input (usually marked by the "~" terminals, but I will refer to them as "~1" and "~2" to differentiate them), and two of which are the rectified output (usually marked by the "+" and "-" terminals). You need to test the four diodes while they are out of cicuit to make sure that they have the proper forward voltage drop (roughly .45V to .8V) when forward biased, and are open when reverse biased. This is trivial to do if your meter has a diode check function. You should do the following eight tests with your meter on the diode test function:
- Red meter lead on "~1" terminal, black meter lead on the "-" terminal --> Meter should show open
- Black meter lead on "~1" terminal, red meter lead on the "-" terminal --> Meter should show the forward voltage drop mentioned above

- Red meter lead on "~2" terminal, black meter lead on the "-" terminal --> Meter should show open
- Black meter lead on "~2" terminal, red meter lead on the "-" terminal --> Meter should show the forward voltage drop mentioned above

- Red meter lead on "~1" terminal, black meter lead on the "+" terminal --> Meter should show the forward voltage drop mentioned above
- Black meter lead on "~1" terminal, red meter lead on the "+" terminal --> Meter should show open

- Red meter lead on "~2" terminal, black meter lead on the "+" terminal --> Meter should show the forward voltage drop mentioned above
- Black meter lead on "~2" terminal, red meter lead on the "+" terminal --> Meter should show open

If any of the above tests show unexpected results, I would suggest you replace the bridge rectifier. There are also cases where the diodes begin to fail or become leaky when placed under a load. This is more difficult to test for, and I recommend just replacing the rectifier if you suspect it in any way. Usually they are cheap, so it should not hurt your wallet too much!

Also, regarding the out of circuit comment above, this is pretty easy on the Centipede power block. In order to do this, you will need to remove fuse F3, unscrew the terminals on "Big blue", and disconnect connector J5.



Well done!! What is this AC ripple on a DC line thing? I've gotta admit, in my games if the 5V voltage was a scoche off, I think most people's thought would be to "turn the knob". I guess perhaps 4.4V is a bit out of the normal bias range for that adjustment.

Without an oscilliscope, how can one use a nice multimeter to make the same diagnosis and verify fix, or is this not possible?

AC ripple refers to how much of variation there is on the DC supply. An ideal DC power source would have 0V ripple, meaning that the DC voltage does not change at all with time. In a typical arcade power supply, though, the DC voltage is made by rectifying an AC voltage. Since an AC voltage is sinusiodal, you get a waveform after the bridge rectifier that is essentially the absolute value of the sine wave (with some clipping due to the forward bias voltage of the diodes). If you look at this on a scope, it would look like many "humps" sitting next to each other. Digital logic does not cope with this, so a large filter capacitor is added to smooth out the valleys between each of the humps, making it appear as though the voltage after the rectifier and capacitor is a fixed, non-changing value. Over time, electrolytic capacitors dry out and lose their ability to store energy. This causes them to not be as good at smoothing out the voltage after the rectifier, leading to the DC voltage bouncing around slightly (introducing the AC component). Once this gets bad enough, the digital logic can no longer tolerate it and will begin to act erratically.

To answer your other question, many meters are perfectly adequate for testing for ripple. Your basic meter will allow you to test both AC and DC voltage. When you are told to check the voltage at the board, you typically measure the DC component of the voltage. This will give you more or less an average voltage, without telling you anything about how much the voltage is varying. You should then be able to switch your meter to measure the AC voltage, and it will give you an idea of how much the voltage is varying. It is difficult to say exactly how well a given meter will measure this, since it will depend on how it removes the DC component, and it will also depend on the frequency range of the noise, but for ballpark measurements I would get that most meters will suffice. FYI, when I was measuring the AC and DC components, both before and after replacing Big blue, I was using my Fluke 177 meter. The scope did not come out until I started troubleshooting the board.

-Jason
 
It is pretty easy to check bridge rectifiers for gross issues, since they are nothing more than four diodes. A bridge rectifier has four terminals, two of which are the AC input (usually marked by the "~" terminals, but I will refer to them as "~1" and "~2" to differentiate them), and two of which are the rectified output (usually marked by the "+" and "-" terminals). You need to test the four diodes while they are out of cicuit to make sure that they have the proper forward voltage drop (roughly .45V to .8V) when forward biased, and are open when reverse biased. This is trivial to do if your meter has a diode check function. You should do the following eight tests with your meter on the diode test function:
- Red meter lead on "~1" terminal, black meter lead on the "-" terminal --> Meter should show open
- Black meter lead on "~1" terminal, red meter lead on the "-" terminal --> Meter should show the forward voltage drop mentioned above

- Red meter lead on "~2" terminal, black meter lead on the "-" terminal --> Meter should show open
- Black meter lead on "~2" terminal, red meter lead on the "-" terminal --> Meter should show the forward voltage drop mentioned above

- Red meter lead on "~1" terminal, black meter lead on the "+" terminal --> Meter should show the forward voltage drop mentioned above
- Black meter lead on "~1" terminal, red meter lead on the "+" terminal --> Meter should show open

- Red meter lead on "~2" terminal, black meter lead on the "+" terminal --> Meter should show the forward voltage drop mentioned above
- Black meter lead on "~2" terminal, red meter lead on the "+" terminal --> Meter should show open

If any of the above tests show unexpected results, I would suggest you replace the bridge rectifier. There are also cases where the diodes begin to fail or become leaky when placed under a load. This is more difficult to test for, and I recommend just replacing the rectifier if you suspect it in any way. Usually they are cheap, so it should not hurt your wallet too much!

Also, regarding the out of circuit comment above, this is pretty easy on the Centipede power block. In order to do this, you will need to remove fuse F3, unscrew the terminals on "Big blue", and disconnect connector J5.





AC ripple refers to how much of variation there is on the DC supply. An ideal DC power source would have 0V ripple, meaning that the DC voltage does not change at all with time. In a typical arcade power supply, though, the DC voltage is made by rectifying an AC voltage. Since an AC voltage is sinusiodal, you get a waveform after the bridge rectifier that is essentially the absolute value of the sine wave (with some clipping due to the forward bias voltage of the diodes). If you look at this on a scope, it would look like many "humps" sitting next to each other. Digital logic does not cope with this, so a large filter capacitor is added to smooth out the valleys between each of the humps, making it appear as though the voltage after the rectifier and capacitor is a fixed, non-changing value. Over time, electrolytic capacitors dry out and lose their ability to store energy. This causes them to not be as good at smoothing out the voltage after the rectifier, leading to the DC voltage bouncing around slightly (introducing the AC component). Once this gets bad enough, the digital logic can no longer tolerate it and will begin to act erratically.

To answer your other question, many meters are perfectly adequate for testing for ripple. Your basic meter will allow you to test both AC and DC voltage. When you are told to check the voltage at the board, you typically measure the DC component of the voltage. This will give you more or less an average voltage, without telling you anything about how much the voltage is varying. You should then be able to switch your meter to measure the AC voltage, and it will give you an idea of how much the voltage is varying. It is difficult to say exactly how well a given meter will measure this, since it will depend on how it removes the DC component, and it will also depend on the frequency range of the noise, but for ballpark measurements I would get that most meters will suffice. FYI, when I was measuring the AC and DC components, both before and after replacing Big blue, I was using my Fluke 177 meter. The scope did not come out until I started troubleshooting the board.

-Jason

Repped. This is one of the most educational posts I've read in a while. Thanks!

Evan
 
Figured I'd piggyback on this thread since it comes up in searches.

Centipede PCB repair -

Problem: Game was working fine, but lost horizontal sync. Adjusting sync revealed screen was duplicated into 4 seperate screens!

Solution: I've heard this will happen on pac man if a counter goes bad, specifically the 74161 @3R. That counter creates the 2/4/6/8etc. H lines that are used to drive all kinds of stuff.

Looking at the centipede schematics, there are 74163 counters in the sync section... the Horizontal sync runs off of 32h and 64h lines... that are created by the 74163 @2N. used a logic probe, 64H was stuck high. Replaced the chip, fixed the game.

So, if you get a screen split into 4 small versions of the game, 2N may be your problem.
 
So I don't make another "Centipede help" thread, I'll post here because the problem I have with a board is similar to issue #4 in the OP.

This board is a POS, it clearly sat in water at some point in its life. Roughly 10 of the chips have a significant amount of rust on them. All those, and sockets for them, are on their way, but I'm trying to diagnose the issue.

I swapped all roms and CPU with a known good set. CPU is getting a good clock, and the reset line is toggling, so I'm stuck in a reset loop. Will not boot, just shows garbage on screen. Trying to boot into test gives me the following screen, so it's trying to boot into test, but somehow can't. Looking for a little guidance
 

Attachments

  • 5A91BC9B-B535-4B2A-BA99-B959CFE334EB.jpg
    5A91BC9B-B535-4B2A-BA99-B959CFE334EB.jpg
    981.5 KB · Views: 20
So I don't make another "Centipede help" thread, I'll post here because the problem I have with a board is similar to issue #4 in the OP.

This board is a POS, it clearly sat in water at some point in its life. Roughly 10 of the chips have a significant amount of rust on them. All those, and sockets for them, are on their way, but I'm trying to diagnose the issue.

I swapped all roms and CPU with a known good set. CPU is getting a good clock, and the reset line is toggling, so I'm stuck in a reset loop. Will not boot, just shows garbage on screen. Trying to boot into test gives me the following screen, so it's trying to boot into test, but somehow can't. Looking for a little guidance

If your board is still resetting due to the watchdog, disable it by grounding the watchdogdisable pin (WDDIS on the PCB). This should let the self test run and maybe signal a bad ram chip.
 
If your board is still resetting due to the watchdog, disable it by grounding the watchdogdisable pin (WDDIS on the PCB). This should let the self test run and maybe signal a bad ram chip.



This did not work. With the watchdog disabled, booting into test gives me the same screen, booting with the test switch off gives me literally nothing on the screen

To add, I'm doing this all with the POKEY removed, as it's most certainly bad(the 4 of the legs are broken off into the socket), and I've read Brian saying that it should be removed when troubleshooting
 
Last edited:
I'm leaning towards the 2114s being bad, because, in this thread, Brian mentions that the game would not boot without them. If it were the 2101s, I would be seeing other issues. I still can't get it to give me any beeps or anything in test mode

https://forums.arcade-museum.com/showthread.php?t=395618&highlight=centipede+pcb&page=2

I've read over 100 threads today :D

You said your symptom was like op's #4 but his screen was random garbage that was constantly changing and yours appears to be static.

did you check all address and data lines like the op did? Did you check clock with logic probe or frequency counter? I would start there.
 
You said your symptom was like op's #4 but his screen was random garbage that was constantly changing and yours appears to be static.

did you check all address and data lines like the op did? Did you check clock with logic probe or frequency counter? I would start there.

"non-static garbage that is changing" sounded to me like it was reseting, which is what mine is doing. My board is only static when the test switch is flipped.

I did check the address and data lines like the op did. They are all toggling. I checked the clock using a logic probe, as I don't have anything else to use. I can't verify that it is indeed 12MHz, but I can verify that it is toggling and it is uniform. I have an LP-560

I know how to use, but unfortunately don't own an oscilloscope
 
Thats a neat one. Probably multiple issues. If the board was as messed up as you say.

Based on the equipment you mention you have I'd go the path you going. Sound like your working with just a logic probe and multimeter.

This is what I'd do in your situation. (Note my suggestions would be completely different if you have a fluke since there is a more direct and straightforward manner to test in that case)
Normally I'd say don't go replacing sockets, but I'd they are as bad as you say I would. However I'd first only replace the cpu and pokey socket and well as the one code socket that contains the eprom with the game self test code (I'd have to check with this is its one if the end ones but I forget which)

I'd also replace the lm324 just because having that and a known good pokey can help you if you have to have both known good items.

So basically what I'd do in this case
Replace those sockets
Install new lm324
Install known good cpu
Install know good pokey
Install verified good test code rom

Now try booting into test mode and listening fir the beeps
If you hear any use the manual to figure out which ram or rams are bad. (I normally do say first pull out pokey but since you already did it and no change now put in a known good if you have one.)

If you hear no beeps or things beep in manners that don't make sense, just replace both 2114s and hope that it's not something eith the supporting circuitry. If you install known good 2114 and it can boot past the 2114 checks the rest of the self test code should help you get past the rest.

Brian
 
Last edited:
Ugh no looking after that pic again I'm thinking "those tiles are way to uniform to be the game crapping before the 2114 test.. only problem is I cannot think of a set of issues that would lead to the pattern your seeing. (Half screen mushroom with alternating upside down) and half screen white

Only think I can think of is that the game is clearing all ram regions, banks 1 and 3 are messed up and something is messed up with the 2114#.

Ahh without more equipment I guess I'd still go with my original post
 
So basically what I'd do in this case
Replace those sockets
Install new lm324
Install known good cpu
Install know good pokey
Install verified good test code rom

Parts for all of this are on the way. I've got multiple working Centi boards here that I can swap socketed chips between. I'll report back when I get that stuff replaced. Thanks
 
Digikey messed up my order so I have to wait for a Pokey socket again, but I may have made progress today

I started piggybacking chips, literally anything that I had on hand, it was piggybacked. I've heard piggybacking RAM doesn't really work, but I did it anyway.

The chip at L6, when piggybacked, changed the screen. Now, I get A's and O's vertically, in every other row, instead of the mushrooms. I'm sure this is because the chip is immediately next to the Video RAMs, and I don't know if it's actually bad or not. May just replace it to be sure

The chip at H8 got rid of the horizontal lines. You can't see it in the previous pic but they were there. With the piggyback, they are gone.

I also replaced and socketed the LM324. Tested the old one in a working board and it was bad

Pic attached, except the static shown isn't there anymore
 

Attachments

  • IMG_5163.jpg
    IMG_5163.jpg
    914.5 KB · Views: 16
Back
Top Bottom